{
    "categories": [
        {
            "name": "市場觀察",
            "name_en": "Market Watch",
            "desc": "每週追蹤市場結構、成交與資金變化。",
            "desc_en": "Weekly tracking of market structure, turnover and flows."
        },
        {
            "name": "籌碼分析",
            "name_en": "Positioning Analysis",
            "desc": "解析籌碼流向與持倉結構。",
            "desc_en": "Analysis of positioning flows and holder structure."
        },
        {
            "name": "風險管理",
            "name_en": "Risk Management",
            "desc": "建立有紀律的風險控制框架。",
            "desc_en": "Building a disciplined risk control framework."
        },
        {
            "name": "投資教育",
            "name_en": "Investor Education",
            "desc": "投資知識與觀念養成。",
            "desc_en": "Investment knowledge and mindset building."
        },
        {
            "name": "市場評論",
            "name_en": "Market Commentary",
            "desc": "對市場事件的觀點與解讀。",
            "desc_en": "Views and interpretation of market events."
        },
        {
            "name": "公司進展",
            "name_en": "Company Progress",
            "desc": "涵遠資本的重要發展與里程碑。",
            "desc_en": "Key developments and milestones of Farsight Capital."
        },
        {
            "name": "合作動態",
            "name_en": "Partnership Updates",
            "desc": "合作夥伴與服務體系進展。",
            "desc_en": "Partners and service network progress."
        },
        {
            "name": "線下沙龍",
            "name_en": "Offline Salons",
            "desc": "投資人沙龍與分享會回顧。",
            "desc_en": "Investor salons and sharing sessions."
        },
        {
            "name": "行業論壇",
            "name_en": "Industry Forums",
            "desc": "行業論壇參與與觀點摘要。",
            "desc_en": "Forum participation and takeaway views."
        }
    ],
    "articles": [
        {
            "id": "r1",
            "section": "research",
            "cat": "市場觀察",
            "homeSlot": 1,
            "tag": {
                "zh": "每週觀察",
                "en": "Weekly Watch"
            },
            "title": {
                "zh": "本週市場觀察11.02-11.06：AI交易熱度降溫，市場進入風險重新定價階段",
                "en": "Weekly Market Outlook | Nov 2–Nov 6, 2026"
            },
            "desc": {
                "zh": "台股一週跌 5,510 點、回撤 20.31%：跌的是折現率，不是 AI 需求；籌碼未清，宜看營收。",
                "en": "AI Trading Momentum Cools as Markets Enter a Phase of Risk Repricing"
            },
            "body": {
                "zh": "本週核心問題：基本面沒有轉壞，台股為何在五個交易日內跌掉 5,510 點？\r\n- 折現率先行：政策時間表、信用成本與利率同步上移，受壓的是估值，不是 AI 需求。\r\n- 結構放大：電子權值高度集中，費半重挫經 ADR 傳導，11 月 6 日出現台股史上最大盤中跌點 3,737.67 點。\r\n- 籌碼未清：外資六日賣超 6,144.71 億元，融資餘額距高點僅減 11.8%，槓桿仍在場內。\r\n- 驗證優先：10 月營收、2027 年資本支出與 AI 企業 CDS 利差，比預測指數點位更能回答下一個問題。\r\n\r\n [img:https:\/\/images.openai.com\/static-rsc-4\/01RZU12FIoJkSDfe8Tr64ZxCFWYgRP9GZKXORBxWxz4XkCQyD8FtoqhgBuBkVGzcUVn9Y6AwzFvZ0K0lGyj5Y09v2DQ4UsTAR4drBFLTmzbSWvhyoyv2W61Jpig6JFwzsSlbZriBx0gHAkgbXn59N4n7mQu-T6ckuNAF4vb2Hyn9SJWoNbsu5yzcAviyFo_a?purpose=fullsize]\r\n\r\n一、五個交易日跌掉 5,510 點：一次評價中樞的下移\r\n加權指數由 10 月 30 日的 45,211.83 點跌至 11 月 6 日的 39,701.64 點，五個交易日下跌 5,510.19 點、跌幅 12.19%，為 2022 年 6 月以來最長的連續七個交易日收黑[1]；自 10 月 6 日歷史收盤高點 49,822.55 點起算，回撤 20.31%，已符合技術性熊市定義。\r\n\r\n跌幅是階梯式而非單日跳空：11 月 3 日、5 日、6 日分別下跌 3.59%、3.20%、4.50%，三個交易日貢獻本週逾九成跌點；11 月 6 日盤中最大跌點 3,737.67 點創台股紀錄，跌停家數一度逾五百檔，終場收斂至 −4.50%[5]。11 月 3 日與 5 日成交值均接近 1 兆元，量能未隨價格萎縮——這是主動降低曝險，不是流動性抽離。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FCiADlwi2uA&scene=im&aid=582478&lang=zh]\r\n\r\n10 月 6 日之後仍是緩坡，10 月 29 日起轉為陡降，11 月 6 日直接壓到回撤 20% 門檻之下。價格路徑說明這不是漲多後的休息，而是評價中樞的下移；下一個問題是，中樞為何下移。\r\n\r\n二、觸發點不在需求，而在折現率\r\n市場習慣用「需求」解釋 AI 行情，但本週三條線索指向同一件事：被重新定價的是 AI 資本支出的時間表與資金成本，而非需求量。\r\n\r\n政策線。資料中心審批權主要落在州政府；本次期中選舉翻轉的州長席位中，德州、喬治亞、俄亥俄為裝機重鎮，其中德州州長當選人主張暫停核准。受限的是時間表，不是訂單。\r\n\r\n信用線。AI 相關企業今年發債近 5,000 億美元，甲骨文 5 年期 CDS 約 263bp、CoreWeave 約 982bp，甲骨文 2060 年到期債券折價至 0.63[3]。算力不再只是資本支出科目，而是需要持續再融資的資產；再融資成本上升，折現率跟著上升。\r\n\r\n利率線。10 年期美債殖利率升至 5.2%–5.3%，實質利率約 2.95%，為 2008 年以來最高[4]。折現率上修與風險溢酬擴大同時發生，正是長天期成長股最敏感的兩個變數。\r\n\r\n三條線並非獨立事件：政策決定「能不能建」，信用決定「借不借得到錢」，利率決定「借錢划不划算」。三者同時惡化，受影響的是整條 AI 供應鏈的評價基礎；台股是全球 AI 硬體密度最高的市場，反應因而劇烈。\r\n\r\n三、跌幅的結構來源：權值集中與 ADR 傳導\r\n台股不是獨立的風險源，而是全球 AI 交易的高 beta 代理。11 月 5 日美股費半重挫約 12.2%、台積電 ADR 下跌 10.8%；11 月 6 日台股加權指數下跌 4.50%，其中台積電 −4.69%、日月光投控 −7.86%、台達電 −6.63%、鴻海 −4.46%、聯發科 −2.02%[6]。\r\n\r\n [img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FEw5wMzkpVe&scene=im&aid=582478&lang=zh]\r\n\r\nADR 跌幅約為現貨的兩倍：現貨有 10% 跌停機制、盤中低接買盤把台積電從跌停拉回、台股收盤時 ADR 隔夜賣壓尚未完全反映。價差收斂未完成，若美股弱勢延續，下週現貨仍有補跌壓力；成立條件是溢價未回到長期區間。指數由少數權值股定價，11 月 6 日跌停家數外溢至中小型標的，說明賣壓已擴散到流動性較差的地方。\r\n\r\n四、籌碼未清：外資撤退、投信承接、槓桿仍高\r\n價格修正的深度最終由籌碼決定。本週同時存在三個方向相反的力量，這是急跌後缺乏有效反彈的原因。\r\n\r\n外資撤退。10 月 29 日至 11 月 5 日連六日賣超 6,144.71 億元，11 月 3 日單日 1,073.06 億元[1]；10 月單月賣超 317.8 億美元、淨匯出 280 億美元，雙雙創單月新高[8]。新台幣同期貶至 33.036 元，為 2025 年 4 月以來低點。\r\n\r\n投信承接。連 18 個交易日買超 3,270.40 億元，內資透過 ETF 持有電子權值的比重上升，「外資賣、內資接」成為常態；代價是籌碼分散、反彈推升效率下降。\r\n\r\n槓桿仍高。融資餘額 11 月 5 日 5,764.09 億元，距 10 月 12 日高點 6,533.30 億元僅減 11.8%，年初 3,434.10 億元，年內仍增 67.9%；單日減 205.37 億元為史上第九大[2]。融資減幅明顯小於指數跌幅，槓桿尚未出清。\r\n\r\n清洗完成的條件不是指數跌深，而是融資餘額連續加速下降、外資賣超收斂、期貨淨空單回補三者同時出現；任一項缺席，反彈都容易變成逃命波。\r\n\r\n五、下週觀察：市場能否從恐慌轉向理性定價\r\n涵遠的做法是不預測點位，只確認條件。以下五個訊號，比指數的絕對位置更能說明風險是否真正下降。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FAFun6YCizK&scene=im&aid=582478&lang=zh]\r\n\r\n跌深不等於便宜。在折現率與政策預期穩定之前，把「跌深」當成買進理由，等於用舊的評價框架定價新的風險結構；研究要做的是把「被折現率殺掉的估值」與「被需求下修殺掉的盈利」分開——前者會隨利率回穩而修復，後者不會。\r\n\r\n六、結論：盈利沒下修，是估值先被壓縮\r\n只看實體經濟，讀數並不支持 20% 的回撤：前 9 月累計經濟成長 14.22%（第 3 季單季 14.36%），9 月出口 872.2 億美元創單月新高[7]。但 10 月 CPI 年增 2.33%、連續 6 個月突破 2%，壓抑寬鬆空間[9]；11 月 5 日本益比 22.46 倍、股價淨值比 3.41 倍[2]，修正一週後評價仍在偏高區間。折現率上修、風險溢酬擴大，22 倍的本益比需要更高的成長支撐，而成長的時間表正是本週被質疑的部分。\r\n\r\n「基本面沒壞」與「股價該跌」並不矛盾：盈利預估沒下修，折現率與風險溢酬先上修。要追問的是，AI 資本支出會不會從「遞延」變成「下修」。目前看到的是遞延與觀望，而非訂單取消；若州級審批凍結延續至 2027 年，資本支出預估就會被下修，屆時才輪到盈利被下修。這是下一階段最重要的分岔點。\r\n\r\n七、涵遠觀點：市場波動之中，更需要保持判斷的獨立性\r\n\r\n市場快速下跌時，投資人容易陷入兩種極端：一種是因為恐慌而急於離場，另一種則是因為價格回落而急於抄底。\r\n\r\n但從機構投資的角度來看，價格下跌本身並不足以構成投資理由，價格反彈也不等於風險已經解除。\r\n\r\n我們更重視的是資金與籌碼的變化、企業基本面的持續性，以及市場情緒與估值之間的落差。\r\n\r\n本輪台股修正，正好提供了一次重新檢視投資組合與市場定價的機會。\r\n\r\n---\r\n數據來源\r\n[1] 臺灣證券交易所：《上市股票每日收盤行情與三大法人買賣超》，2026\/11\/06 取得。\r\n[2] Goodinfo 臺灣股市資訊網：《臺股指數與信用交易統計》，2026\/11\/06 取得。\r\n[3] 鉅亨網：《AI 相關企業發債與 CDS 走勢》，2026\/11\/06 取得。\r\n[4] 富邦期貨：《臺股盤前分析》，2026\/11\/06 取得。\r\n[5] 經濟日報：《臺股 11\/6 收盤與盤中跌點報導》，2026\/11\/06。\r\n[6] 三立新聞網：《權值股單日表現與跌停家數》，2026\/11\/06。\r\n[7] 行政院主計總處：《GDP: Advance Estimate of 2026Q3》，2026\/10\/30。\r\n[8] 中央銀行：《新臺幣匯率與外資進出統計》，2026\/11\/06 取得。\r\n[9] 中國新聞社：《台灣 10 月 CPI 年增 2.33%，連續六個月突破警戒線》，2026\/11\/05。\r\n\r\n---\r\n本文由涵遠資本研究團隊整理。\r\n資料來源為公開資訊，僅供研究與教育參考，不構成任何投資建議、要約或招攬。投資決策應自行評估風險與自身財務狀況。\r\n涵遠資本——深涵洞見，行穩致遠",
                "en": "Taiwan equities shed 5,510 points in one week, with a peak-to-trough correction of 20.31%: what is being repriced is the discount rate, not AI demand. Positioning remains unsettled; revenue trends deserve closer attention.**\r\n\r\nThe Central Question This Week\r\n\r\nIf corporate fundamentals have not materially deteriorated, why did Taiwan's stock market lose 5,510 points in just five trading sessions?\r\n\r\n- Discount rates moved first: Policy timelines, credit costs, and interest rates all shifted higher, putting pressure on valuations rather than AI demand itself.\r\n- Market structure amplified the decline: Heavy concentration in large-cap technology stocks, combined with spillover from the semiconductor sell-off through ADRs, contributed to a record intraday decline of 3,737.67 points on November 6.\r\n- Positioning remains unsettled: Foreign investors recorded NT$614.471 billion in net selling over six consecutive sessions, while margin balances remained just 11.8% below their peak, indicating that leverage is still elevated.\r\n- Fundamental verification takes priority: October revenue figures, 2027 capital expenditure plans, and CDS spreads of AI-related companies offer more meaningful signals than attempts to forecast index levels.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/01RZU12FIoJkSDfe8Tr64ZxCFWYgRP9GZKXORBxWxz4XkCQyD8FtoqhgBuBkVGzcUVn9Y6AwzFvZ0K0lGyj5Y09v2DQ4UsTAR4drBFLTmzbSWvhyoyv2W61Jpig6JFwzsSlbZriBx0gHAkgbXn59N4n7mQu-T6ckuNAF4vb2Hyn9SJWoNbsu5yzcAviyFo_a?purpose=fullsize]\r\n\r\n1. A 5,510-Point Decline in Five Sessions: A Shift in the Market's Valuation Anchor\r\n\r\nThe Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) fell from 45,211.83 on October 30 to 39,701.64 on November 6, losing 5,510.19 points, or 12.19%, over five trading sessions. This marked its longest consecutive losing streak of seven sessions since June 2022 [1]. Measured against its historical closing high of 49,822.55 on October 6, the index had retreated 20.31%, meeting the conventional technical definition of a bear market.\r\n\r\nThe decline unfolded in stages rather than through a single overnight gap. On November 3, 5, and 6, the index fell 3.59%, 3.20%, and 4.50%, respectively. These three sessions accounted for more than 90% of the week's total point losses. On November 6, the TAIEX suffered an intraday decline of 3,737.67 points, setting a new record for the largest intraday point drop in Taiwan's market history. More than 500 stocks briefly hit their daily price limits before the index recovered somewhat to close down 4.50% [5].\r\n\r\nTrading value on November 3 and 5 approached NT$1 trillion on both occasions. Trading activity did not contract alongside prices. This suggests active exposure reduction rather than a simple withdrawal of market liquidity.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FCiADlwi2uA&scene=im&aid=582478&lang=zh]\r\n\r\nThe market maintained a relatively gradual downward trajectory after October 6, before entering a much steeper decline from October 29 onward. By November 6, the correction had pushed the index below the 20% threshold.\r\n\r\nThis was more than a routine pause following an extended rally. It represented a downward adjustment in the market's valuation anchor. The next question is what caused that adjustment.\r\n\r\n2. The Trigger Was Not Demand, but the Discount Rate\r\n\r\nMarkets have traditionally used demand expectations to explain the AI-driven rally. However, three developments this week point toward a different conclusion: what is being repriced is the timing of AI capital expenditure and the cost of financing it, rather than the underlying volume of demand.\r\n\r\nPolicy:\r\n\r\nData center approvals largely fall under state-level jurisdiction. Among the gubernatorial races affected by this election cycle, Texas, Georgia, and Ohio are particularly important data center markets. The incoming governor of Texas has advocated a pause in new approvals.\r\n\r\nThe immediate constraint is therefore the deployment timeline, not necessarily the order pipeline.\r\n\r\nCredit:\r\n\r\nAI-related companies have issued nearly US$500 billion in debt this year. Oracle's five-year CDS spread has reached approximately 263 basis points, while CoreWeave's has climbed to around 982 basis points. Oracle bonds maturing in 2060 have also traded at approximately 63 cents on the dollar [3].\r\n\r\nComputing capacity is no longer simply a capital expenditure item. It is increasingly an asset base that requires continuous refinancing. As refinancing costs rise, the effective discount rate rises with them.\r\n\r\nInterest Rates:\r\n\r\nThe 10-year U.S. Treasury yield has climbed to 5.2%–5.3%, while the real yield has reached approximately 2.95%, its highest level since 2008 [4].\r\n\r\nAn upward repricing of discount rates and an expansion in risk premiums are occurring simultaneously. These are precisely the two variables to which long-duration growth equities are most sensitive.\r\n\r\nThese three developments are interconnected rather than independent events:\r\n\r\n- Policy determines whether projects can be built.\r\n- Credit determines whether financing remains accessible.\r\n- Interest rates determine whether that financing remains economically viable.\r\n\r\nWhen all three deteriorate simultaneously, the valuation foundation of the entire AI supply chain comes under pressure. Taiwan, with one of the world's highest concentrations of AI hardware exposure, has therefore experienced a particularly pronounced market reaction.\r\n\r\n3. The Structural Drivers of the Decline: Large-Cap Concentration and ADR Spillover\r\n\r\nTaiwan's equity market is not an independent source of risk. It functions as a high-beta proxy for global AI-related investment positioning.\r\n\r\nOn November 5, the Philadelphia Semiconductor Index (SOX) plunged approximately 12.2%, while Taiwan Semiconductor Manufacturing Company's (TSMC) ADRs fell 10.8%. On November 6, the TAIEX declined 4.50%, with major constituent stocks also under significant pressure:\r\n\r\n- TSMC: −4.69%\r\n- ASE Technology Holding: −7.86%\r\n- Delta Electronics: −6.63%\r\n- Hon Hai Precision Industry (Foxconn): −4.46%\r\n- MediaTek: −2.02% [6]\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FEw5wMzkpVe&scene=im&aid=582478&lang=zh]\r\n\r\nThe decline in ADRs was approximately twice that of the corresponding local shares. Several structural factors help explain the divergence.\r\n\r\nTaiwan's cash equity market operates under a 10% daily price-limit mechanism. Intraday bargain-hunting demand helped pull TSMC shares back from their limit-down level. In addition, selling pressure in the U.S. ADR market had not been fully reflected by the time Taiwan's market closed.\r\n\r\nThe convergence of this valuation gap remains incomplete. If weakness in U.S. equities persists, Taiwan's cash market may face further catch-up pressure next week. This scenario would be supported if the ADR premium or discount remains outside its historical range.\r\n\r\nThe index is heavily influenced by a relatively small number of large-cap stocks. The spillover of limit-down trading to smaller-cap stocks on November 6 suggests that selling pressure has begun spreading into less liquid segments of the market.\r\n\r\n4. Positioning Remains Unsettled: Foreign Selling, Domestic Institutional Buying, and Elevated Leverage\r\n\r\nUltimately, the depth of a market correction is determined not only by prices, but also by positioning.\r\n\r\nThree opposing forces have been operating simultaneously this week, helping explain why the market has struggled to generate a meaningful rebound following the sharp decline.\r\n\r\nForeign Investors: Continued Capital Withdrawal\r\n\r\nForeign investors recorded net selling for six consecutive sessions from October 29 through November 5, totaling NT$614.471 billion. On November 3 alone, net selling reached NT$107.306 billion [1].\r\n\r\nFor October as a whole, foreign investors recorded net sales of US$31.78 billion, while net capital outflows reached US$28 billion. Both figures represented new monthly records [8].\r\n\r\nOver the same period, the New Taiwan dollar weakened to NT$33.036 per U.S. dollar, its lowest level since April 2025.\r\n\r\nDomestic Institutions: Absorbing Selling Pressure\r\n\r\nDomestic investment trusts recorded net buying for 18 consecutive trading sessions, totaling NT$327.040 billion.\r\n\r\nDomestic investors have increased their exposure to large-cap technology stocks through ETFs, making the pattern of foreign investors selling while domestic institutions absorb supply increasingly common.\r\n\r\nHowever, this redistribution of ownership also fragments market positioning and reduces the efficiency with which buying flows can drive a broad-based recovery.\r\n\r\nMargin Financing: Leverage Remains Elevated\r\n\r\nAs of November 5, Taiwan's margin financing balance stood at NT$576.409 billion, only 11.8% below its October 12 peak of NT$653.330 billion.\r\n\r\nCompared with NT$343.410 billion at the beginning of the year, margin financing remained 67.9% higher year to date. The single-day reduction of NT$20.537 billion ranked as the ninth-largest daily decline on record [2].\r\n\r\nThe reduction in margin balances has been significantly smaller than the decline in the index itself, indicating that leveraged positions have not yet been fully unwound.\r\n\r\nA completed deleveraging process should not be defined simply by how far the index has fallen. Rather, it requires three conditions to emerge simultaneously:\r\n\r\n1. A sustained acceleration in the decline of margin financing balances.\r\n2. A meaningful contraction in foreign investors' net selling.\r\n3. A reduction in net short positions in index futures.\r\n\r\nIf any of these conditions remains absent, a market rebound may prove to be an opportunity for investors to exit rather than evidence of a sustainable recovery.\r\n\r\n5. What to Watch Next Week: Can the Market Move From Panic to Rational Pricing?\r\n\r\nFarsight Capital does not seek to predict index levels. Our approach is to identify and monitor the conditions that determine whether market risk is genuinely declining.\r\n\r\nThe following five signals are more informative than the absolute level of the index when assessing whether the market correction is approaching a more stable phase.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FAFun6YCizK&scene=im&aid=582478&lang=zh]\r\n\r\nA sharp decline does not automatically make an asset undervalued.\r\n\r\nUntil discount rates and policy expectations stabilize, treating a steep price correction as a sufficient reason to buy amounts to applying an outdated valuation framework to a fundamentally different risk environment.\r\n\r\nThe analytical priority is to distinguish between two very different sources of valuation destruction:\r\n\r\n- Valuation compression driven by higher discount rates: This may reverse as interest rates stabilize and risk premiums normalize.\r\n- Earnings deterioration driven by weaker underlying demand: This is far less likely to recover simply because interest rates decline.\r\n\r\nThe distinction is critical when evaluating whether current market weakness represents a temporary repricing or a more fundamental deterioration in corporate earnings power.\r\n\r\n## 6. Conclusion: Earnings Have Not Been Downgraded; Valuations Have Been Compressed First\r\n\r\nLooking solely at the real economy, the available data do not appear to justify a 20% market correction.\r\n\r\nTaiwan's cumulative economic growth for the first nine months reached 14.22%, while third-quarter growth alone came in at 14.36%. September exports reached a record US$87.22 billion for a single month [7].\r\n\r\nHowever, October CPI increased 2.33% year over year, marking the sixth consecutive month above the 2% threshold and limiting the scope for monetary easing [9].\r\n\r\nAs of November 5, the market's price-to-earnings ratio stood at 22.46 times, while its price-to-book ratio was 3.41 times [2]. Even after a week of substantial losses, valuations remained relatively elevated.\r\n\r\nWith discount rates rising and risk premiums expanding, a market trading at approximately 22 times earnings requires stronger growth expectations to justify its valuation. The timing and sustainability of that growth are precisely what investors have begun to question.\r\n\r\nThere is no contradiction between saying that fundamentals remain intact and acknowledging that stock prices deserve to fall.\r\n\r\nEarnings estimates have not yet been materially revised downward, but discount rates and risk premiums have already moved higher.\r\n\r\nThe more important question is whether AI capital expenditure will move from being **delayed to being reduced**.\r\n\r\nWhat we are currently observing is postponement and caution, rather than widespread order cancellations. However, if state-level approval freezes persist into 2027, capital expenditure forecasts may eventually be revised downward. Only at that point would the pressure extend more directly into corporate earnings estimates.\r\n\r\nThis distinction represents the most important turning point to monitor in the next phase of the market.\r\n\r\n## 7. The Farsight Perspective: Independent Judgment Matters Most During Periods of Market Volatility\r\n\r\nWhen markets decline rapidly, investors are often drawn toward two extremes: rushing to exit out of fear, or rushing to buy the dip simply because prices have fallen.\r\n\r\nFrom an institutional investment perspective, however, a decline in price alone does not constitute an investment thesis. Nor does a rebound necessarily indicate that underlying risks have been resolved.\r\n\r\nWe place greater emphasis on changes in capital flows and market positioning, the sustainability of corporate fundamentals, and the divergence between market sentiment and valuation.\r\n\r\nThe current correction in Taiwan equities provides an opportunity to reassess both portfolio positioning and the assumptions embedded in market pricing.\r\n\r\nMaintaining independent judgment, particularly when market sentiment becomes extreme, remains essential to disciplined investment decision-making.\r\n\r\n---\r\n\r\nData Sources\r\n\r\n[1] Taiwan Stock Exchange (TWSE): *Daily Closing Prices of Listed Stocks and Trading Activity of Institutional Investors*, retrieved November 6, 2026.\r\n\r\n[2] Goodinfo! Taiwan Stock Market Information: *Taiwan Stock Index and Margin Trading Statistics*, retrieved November 6, 2026.\r\n\r\n[3] Cnyes: *Debt Issuance and CDS Trends Among AI-Related Companies*, retrieved November 6, 2026.\r\n\r\n[4] Fubon Futures: *Taiwan Equity Market Pre-Opening Analysis*, retrieved November 6, 2026.\r\n\r\n[5] Economic Daily News: *Taiwan Stock Market Closing Performance and Intraday Decline on November 6*, November 6, 2026.\r\n\r\n[6] SETN (Sanlih E-Television News): *Daily Performance of Major Constituents and Number of Limit-Down Stocks*, November 6, 2026.\r\n\r\n[7] Directorate-General of Budget, Accounting and Statistics (DGBAS), Executive Yuan: *GDP: Advance Estimate of 2026 Q3*, October 30, 2026.\r\n\r\n[8] Central Bank of the Republic of China (Taiwan): *New Taiwan Dollar Exchange Rates and Foreign Capital Flows*, retrieved November 6, 2026.\r\n\r\n[9] China News Service: *Taiwan's October CPI Rises 2.33% Year Over Year, Exceeding the 2% Threshold for Six Consecutive Months*, November 5, 2026.\r\n\r\n---\r\n\r\nThis report was prepared by the Research Team at Farsight Capital.\r\n\r\nThe information presented is compiled from publicly available sources and is intended solely for research and educational purposes. It does not constitute investment advice, an offer, solicitation, or recommendation to buy or sell any financial instrument. Investment decisions should be made independently, taking into account individual financial circumstances and risk tolerance.\r\n\r\nFarsight Capital\r\n\r\nDeep Insight. Steady Progress."
            },
            "img": "image\/7dc88d6f36a6f19748542866282610275f702c2b.png",
            "video": ""
        },
        {
            "id": "r2",
            "section": "research",
            "cat": "籌碼分析",
            "homeSlot": 2,
            "tag": {
                "zh": "專題研究",
                "en": "Special Research"
            },
            "title": {
                "zh": "外資買賣超與期貨持倉應該如何解讀？",
                "en": "How Should Foreign Investor Net Trading and Futures Positioning Be Interpreted?"
            },
            "desc": {
                "zh": "認識不同指標的意義、局限與相互關係。",
                "en": "Understanding the Meaning, Limitations, and Interrelationships of Different Market Indicators"
            },
            "body": {
                "zh": "外資買賣超與期貨持倉怎麼讀：三個誤讀與一個四象限\r\n涵遠資本研究部｜專題研究｜2026 年 11 月 6 日\r\n\r\n核心問題：外資現貨買賣超與台指期持倉，哪一個才是方向訊號？兩者都不是，合起來才是。\r\n- 流量與存量：買賣超是當日淨成交，未平倉是收盤部位。\r\n- 背離是常態：買現貨同時空期貨，多是套利或避險，不是看空。\r\n- 一致才給權重：現貨賣超、期貨加空、價差轉逆同時出現，才是方向訊號。\r\n- 本週例證：外資六日賣超 6,144.71 億元，期貨淨空單僅增 10,734 口、名目約 893 億元。\r\n\r\n [img:https:\/\/images.openai.com\/static-rsc-4\/LbL165ye1_Bpisimj9lNndl3Frx7_3jI1gK6BTByc3fEVani00at8Q7PbbF1Le10vH5DPUaLdhvVhIeNi3kTxFIaWzd2vuo3ek-Zd1XyXlR_PacgTQ-5mwTp2RBtqATSbX2R3_9GHA4nzDNIkD3g6LRwxF0vkdJXVhQFCjKrfgs?purpose=inline]\r\n\r\n一、兩個數字，三種誤讀\r\n市場最常見的句式是：「外資賣超 1,024 億元、台指期空單 9.5 萬口，外資看空。」這句話把三個不同性質的東西合成一個結論，因而產生三種誤讀。\r\n\r\n把流量當立場。買賣超是當日買進與賣出的差額，只反映當天在集中市場的成交；它不區分避險、換股、被動調倉或主動放空，也不含海外交易。賣超只說明今天有人在賣，不說明為什麼賣、還會不會賣。[3]\r\n\r\n把口數當規模。期交所總表把台指期、電子期、股票期貨、ETF 期貨與選擇權併列，乘數與方向定義各異。11\/6 外資期貨淨空 406,644 口，其中股票期貨 307,179 口、台指期 99,930 口：口數上台指期只占兩成半，名目金額上卻占近九成。口數不能跨契約相加。[1]\r\n\r\n把存量當預測。未平倉是收盤部位，空單高只代表避險或價差交易部位大，不保證指數下跌。若空單是為現貨投保，指數上漲時空單虧損，避險者也不會因此平倉。\r\n\r\n二、口徑先對齊：四個「外資」不是同一個外資\r\n外資及陸資不等於全體外資。證交所與櫃買中心的「外資及陸資」不含外資自營商；後者計入自營商，不重複計入三大法人。統計涵蓋普通股、鉅額、零股、ETF 與綜合帳戶，以當日原始成交為準。[4]\r\n金額是買進減賣出，不是匯出。賣股所得可停在台幣存款，未必換匯，因此賣超與央行外資匯出入數據經常不同步；用賣超推斷資金外逃是常見跳步。\r\n\r\n期貨要指名契約。「外資台指期淨空單」專指 TX，每口 200 元／點；小台 50 元、微台 10 元。媒體若寫「外資期貨空單 45 萬口」，多半已把股票期貨與選擇權算進去。\r\n\r\n三大法人是互抵結果。期交所明示，多空淨額是「眾多法人機構不同看法合計互抵的結果」。它適合描述結構，不適合還原單一機構意圖。[1]\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/AwS0iKDkWadIKvBnDXwRjspnDJBbj285TT2Y_nHTHODAc518Cx7aT5GyUv4AZoEU5YLbPJVlIDc8NjL2LP4kOcPHM4RbjKNLIb90aoBMDzxzp8eg8qS-hKJz7FHibWehfOpapFrjMXUXnbyhQvjFVhJMQV_MbGGT7xwjlExAlcc?purpose=inline]\r\n\r\n三、外資買賣超：五種不是看空的賣超\r\n口徑對齊後，買賣超仍是有效的壓力計，但要先排除五種雜訊。\r\n- ETF 申贖與造市：申贖、造市與套利都計入，賣超 ETF 可能是基金機制，而非看空成分股。\r\n- 指數調整：MSCI 半年度調整落在 5 月與 11 月，生效日尾盤調倉會放大單日金額。\r\n- 除息與換股：除息旺季的賣超不代表撤資；產業遷移也會連續賣超。\r\n- 期現套利與避險：買現貨空期貨、或賣現貨空期貨，兩邊互抵，現貨數字失真。\r\n- 海外與集團調倉：ADR 轉換、可轉債、集團持股調整，未必反映方向判斷。\r\n再看四個標準化指標：賣超占當日成交值（11\/5 約 11%）、占持股市值、連續天數與日均、以及集中度。\r\n背離比同向更值得注意。10\/28 外資買超 159.57 億元，台指期淨空單仍達 84,917 口；買現貨、空期貨同時出現，是典型的期現套利或避險。\r\n\r\n四、期貨持倉：保險還是押注\r\n期貨持倉的核心問題只有一個：這筆空單是保險，還是押注。三個判據足以區分。\r\n換成名目金額。一口台指期 200 元／點；以 11\/5 指數 41,575 點計，每口約 831 萬元。95,651 口約 7,950 億元，對照外資數十兆元持股，仍屬部分避險，而非全面對作。\r\n\r\n看期現價差。11\/5 台指期結算價 41,882 點、現貨 41,574.72 點，正價差 307 點。崩跌日仍正價差，說明期貨沒有領跌、空方未急於追價；若由正轉逆並持續擴大，才代表避險升級或方向性放空。除息造成的「假逆價差」須加回除息點數。[2]\r\n\r\n看結算與轉倉。台指期每月第三個週三結算，11 月契約為 11\/18；結算前後空單增加，可能只是換月，不是新增押注。季月轉倉量更大。\r\n\r\n輔助訊號是選擇權。11\/5 外資買權淨賣 820 口、賣權淨買 2,424 口，方向偏空但規模遠小於台指期，較像尾部保護。[1]\r\n邊界：外資可在 SGX、HKEX 的 MSCI 台灣期貨避險，本地 TX 部位只是總避險的一部分；單看台指期會系統性低估。\r\n\r\n五、四象限：一致才給權重\r\n把現貨買賣超與期貨淨部位變動放在同一張圖，訊號才完整。順序是：先看方向是否一致，再看價差是否配合，最後看是否只是轉倉。\r\n\r\n [img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F05bxYtPSKd&scene=im&aid=582478&lang=zh]\r\n\r\n四象限不是預測表，而是一致性檢查表。①與④權重最高；②與③須先排除機制性因素，否則容易把避險讀成方向、把換股讀成撤資。\r\n時間尺度也要疊加：單日雜訊最大，5 日與 20 日均線較能看出趨勢。\r\n\r\n六、實戰舉例：五步讀完一輪賣壓\r\n以 10\/28 至 11\/6 為例，把框架走一遍。每一步只回答一個問題，結論必須能被下一筆數據推翻。\r\n\r\n [img:https:\/\/aka.doubaocdn.com\/s\/RSB1mneUnC]\r\n\r\n第一步，定位象限。10\/28 外資買超 159.57 億元、淨空單 84,917 口，落在 ③；買超與加空並存，先判為期現套利或避險，不當方向訊號。\r\n第二步，等連續性。10\/29 現貨轉賣超 688.15 億元、空單升至 89,522 口，進入 ①；其後五個交易日現貨均為賣超，方向才確立。\r\n第三步，量比例。10\/29 至 11\/5 現貨累計賣超 6,144.71 億元；淨空單增 10,734 口，以 11\/5 每口約 831 萬元計，名目約 893 億元，僅為現貨賣超的 14.5%。\r\n第四步，驗價差。11\/5 指數重挫 1,374 點，台指期仍正價差 307 點；期貨未領跌，說明空方沒有追價，避險成分大於攻擊。\r\n第五步，寫下可證偽的判斷。「本輪為現貨主導的減碼，期貨只是同步避險。」若現貨賣超收斂、淨空單續增且逆價差擴大，判斷即被推翻。11\/6 淨空單升至 99,930 口、價差收斂至逆價差約 50 點，方向開始一致，但未見空單獨自加速。[5]\r\n\r\n七、結論\r\n結論一：買賣超是流量，持倉是存量，兩者不能互相替代。\r\n結論二：期貨空單的性質取決於名目規模、價差結構與轉倉周期；口數最大者往往不是台指期。\r\n結論三：只有現貨、期貨、價差三者同向，訊號才值得提高權重；背離時先找機制性原因。\r\n\r\n\r\n數據來源\r\n[1] 臺灣期貨交易所，〈三大法人—區分各期貨契約（依日期）〉，2026 年 11 月 6 日。https:\/\/www.taifex.com.tw\/cht\/3\/futContractsDate\r\n[2] 臺灣期貨交易所，〈臺股期貨每日交易行情〉，2026 年 11 月 5 日、11 月 6 日。https:\/\/www.taifex.com.tw\/cht\/3\/futDailyMarketExcel\r\n[3] HiStock 嗨投資，〈上市三大法人買賣超與台股期貨未平倉〉。https:\/\/histock.tw\/stock\/three.aspx\r\n[4] 證券櫃檯買賣中心，〈三大法人買賣明細資訊—說明〉。https:\/\/www.tpex.org.tw\/web\/stock\/3insti\/daily_trade\/3itrade_hedge_result.php\r\n[5] Infobae／EFE，〈La Bolsa de Taiwán cae un 4,51 % y el regulador estudiará medidas para calmar el mercado〉，2026 年 11 月 6 日。https:\/\/www.infobae.com\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "How to Read Foreign Investor Flows and TAIEX Futures Positions: Three Common Misinterpretations and a Four-Quadrant Framework\r\n\r\nFarsight Capital Research Team | Special Research | November 6, 2026\r\n\r\nThe Central Question\r\n\r\nWhich is the more reliable directional signal: foreign investors' net buying and selling in the cash market, or their positions in TAIEX futures?\r\n\r\nNeither indicator is sufficient on its own. The two must be interpreted together.\r\n\r\n- Flows vs. positions: Net buying and selling measure daily transaction flows, while open interest reflects outstanding positions at the close.\r\n- Divergence is normal: Buying cash equities while holding short futures positions often reflects arbitrage or hedging, not necessarily a bearish outlook.\r\n- Consistency determines signal strength: A combination of cash-market net selling, additional futures short positions, and a shift in the futures basis toward backwardation provides a stronger directional signal.\r\n- This week's example: Foreign investors recorded NT$614.471 billion in net selling over six consecutive sessions, while their net short futures positions increased by only 10,734 contracts, representing approximately NT$89.3 billion in notional exposure.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/LbL165ye1_Bpisimj9lNndl3Frx7_3jI1gK6BTByc3fEVani00at8Q7PbbF1Le10vH5DPUaLdhvVhIeNi3kTxFIaWzd2vuo3ek-Zd1XyXlR_PacgTQ-5mwTp2RBtqATSbX2R3_9GHA4nzDNIkD3g6LRwxF0vkdJXVhQFCjKrfgs?purpose=inline]\r\n\r\n1. Two Numbers, Three Common Misinterpretations\r\n\r\nOne of the most frequently heard market statements is: \"Foreign investors sold NT$102.4 billion in equities and held 95,000 short TAIEX futures contracts. Foreign investors are bearish.\"\r\n\r\nThis conclusion combines three indicators with fundamentally different characteristics, leading to three common misinterpretations.\r\n\r\nMistaking Transaction Flows for Market Conviction\r\n\r\nNet buying or selling represents the difference between purchases and sales during a given trading session. It reflects only transactions executed in the centralized cash market on that day.\r\n\r\nIt does not distinguish between hedging, portfolio rotation, passive rebalancing, or active directional short selling. Nor does it include overseas transactions.\r\n\r\nNet selling tells us that investors sold more than they bought that day. It does not explain why they sold or whether the selling will continue. [3]\r\n\r\nMistaking Contract Volume for Exposure Size\r\n\r\nThe Taiwan Futures Exchange's aggregate statistics list TAIEX futures, electronic sector futures, single-stock futures, ETF futures, and options together. These instruments have different contract multipliers and directional definitions.\r\n\r\nOn November 6, foreign investors held a net short position of 406,644 futures contracts, comprising 307,179 single-stock futures contracts and 99,930 TAIEX futures contracts.\r\n\r\nBy contract count, TAIEX futures represented only approximately one-quarter of the total. By notional value, however, they accounted for nearly 90%.\r\n\r\nContract counts across different instruments cannot simply be added together as though they represented equivalent exposure. [1]\r\n\r\nMistaking Outstanding Positions for Market Predictions\r\n\r\nOpen interest represents positions remaining outstanding at the close of trading. A large short position indicates substantial hedging or spread-trading exposure, but does not guarantee that the underlying index will decline.\r\n\r\nIf short futures positions are being used to hedge cash equity holdings, a rising index may generate losses on the futures positions. That does not necessarily mean the hedging investor will close those positions immediately.\r\n\r\n2. Aligning Definitions: Four Different Measures of \"Foreign Investors\" Are Not the Same\r\n\r\nForeign and Mainland Chinese Investors Do Not Represent All Foreign Institutional Activity\r\n\r\nThe category \"Foreign Investors and Mainland Chinese Investors\" reported by the Taiwan Stock Exchange and the Taipei Exchange excludes foreign proprietary trading firms.\r\n\r\nThe latter are classified under proprietary dealers and are not counted again within the three major institutional investor categories.\r\n\r\nThe reported statistics cover common stocks, block trades, odd-lot transactions, ETFs, and omnibus accounts, based on original transactions executed during the trading day. [4]\r\n\r\nNet Selling Does Not Equal Capital Outflows\r\n\r\nNet selling represents purchases minus sales. It does not directly measure capital leaving Taiwan.\r\n\r\nProceeds from equity sales may remain in New Taiwan dollar deposits rather than being converted into foreign currency. Consequently, net selling figures and the central bank's foreign exchange inflow and outflow statistics frequently move independently.\r\n\r\nInterpreting foreign investors' net selling as evidence of capital flight is therefore an unjustified analytical leap.\r\n\r\nFutures Analysis Must Specify the Contract\r\n\r\nThe term \"foreign investors' net short TAIEX futures position\" refers specifically to the TX contract, which has a multiplier of NT$200 per index point.\r\n\r\nBy comparison, Mini-TAIEX futures have a multiplier of NT$50 per point, while Micro-TAIEX futures have a multiplier of NT$10 per point.\r\n\r\nWhen media reports refer to \"450,000 short futures contracts held by foreign investors,\" the figure often includes single-stock futures and options rather than TAIEX futures alone.\r\n\r\nInstitutional Net Positions Are Aggregated Results\r\n\r\nThe Taiwan Futures Exchange explicitly notes that net long and short positions represent the combined, offsetting result of numerous institutional participants with different market views.\r\n\r\nThese statistics are useful for understanding the overall positioning structure, but they cannot reliably reveal the intentions of any single institution. [1]\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/AwS0iKDkWadIKvBnDXwRjspnDJBbj285TT2Y_nHTHODAc518Cx7aT5GyUv4AZoEU5YLbPJVlIDc8NjL2LP4kOcPHM4RbjKNLIb90aoBMDzxzp8eg8qS-hKJz7FHibWehfOpapFrjMXUXnbyhQvjFVhJMQV_MbGGT7xwjlExAlcc?purpose=inline]\r\n\r\n3. Foreign Investor Net Trading: Five Types of Selling That Do Not Necessarily Signal a Bearish View\r\n\r\nEven after aligning statistical definitions, net buying and selling remains a useful measure of market pressure. However, five sources of noise should first be ruled out.\r\n\r\n- ETF subscriptions, redemptions, and market making: Subscription and redemption activity, market making, and arbitrage transactions are all included. Net selling in ETFs may reflect fund mechanics rather than a bearish view on the underlying constituent stocks.\r\n- Index rebalancing: MSCI's semiannual index reviews take place in May and November. Portfolio adjustments near the closing auction on implementation dates can significantly inflate daily trading values.\r\n- Ex-dividend adjustments and portfolio rotation: Net selling during peak dividend periods does not necessarily indicate capital withdrawal. Sector rotation can also generate consecutive sessions of net selling.\r\n- Cash-futures arbitrage and hedging: Investors may buy cash equities while shorting futures, or sell cash equities while maintaining short futures positions. The two legs may offset each other, making the cash-market figures an incomplete reflection of directional intent.\r\n- Overseas and group-level portfolio adjustments: ADR conversions, convertible bonds, and adjustments to group-level shareholdings may affect reported transactions without reflecting a directional market view.\r\n\r\nFour additional standardized indicators should then be examined:\r\n\r\n1. Net selling as a proportion of daily market turnover (approximately 11% on November 5).\r\n2. Net selling relative to the market value of foreign investors' holdings.\r\n3. The number of consecutive net-selling sessions and average daily selling.\r\n4. The concentration of selling across individual stocks and sectors.\r\n\r\nDivergence can be more informative than directional alignment.\r\n\r\nOn October 28, foreign investors recorded net buying of NT$15.957 billion, while their net short TAIEX futures position remained as high as 84,917 contracts.\r\n\r\nThe simultaneous occurrence of cash-market buying and futures short positions is a typical pattern associated with cash-futures arbitrage or hedging.\r\n\r\n4. Futures Positioning: Insurance or Directional Bets?\r\n\r\nThere is one central question when interpreting futures positions: are these short contracts being held as insurance or as directional bets?\r\n\r\nThree criteria help distinguish between the two.\r\n\r\nConvert Contract Counts Into Notional Exposure\r\n\r\nOne TAIEX futures contract has a multiplier of NT$200 per index point.\r\n\r\nUsing the November 5 index level of 41,575, each contract represented approximately NT$8.31 million in notional exposure.\r\n\r\nA position of 95,651 contracts therefore represented approximately NT$795 billion.\r\n\r\nCompared with foreign investors' equity holdings worth tens of trillions of New Taiwan dollars, this still represents partial hedging rather than a comprehensive offset against their entire cash equity exposure.\r\n\r\nExamine the Cash-Futures Basis\r\n\r\nOn November 5, the TAIEX futures settlement price was 41,882, compared with a cash index close of 41,574.72, resulting in a positive basis of 307 points.\r\n\r\nThe futures market remained in contango even on a sharply declining market day. This suggests that futures were not leading the sell-off and that short sellers were not aggressively chasing prices lower.\r\n\r\nA shift from a positive to a negative basis, particularly if the negative spread continues to widen, would provide stronger evidence of intensified hedging or directional short selling.\r\n\r\nHowever, an apparent negative basis caused by ex-dividend adjustments must be corrected by adding back the relevant dividend points. [2]\r\n\r\nMonitor Expiration and Position Rollover\r\n\r\nTAIEX futures expire on the third Wednesday of each month. The November 2026 contract expires on November 18.\r\n\r\nAn increase in short positions around expiration may simply reflect the rollover of positions into the next contract month rather than the initiation of new bearish bets.\r\n\r\nRollover activity is generally more substantial around quarterly contract expirations.\r\n\r\nOptions as a Supplementary Indicator\r\n\r\nOptions positioning provides an additional reference point.\r\n\r\nOn November 5, foreign investors recorded net selling of 820 call options and net buying of 2,424 put options.\r\n\r\nThe positioning leaned bearish, but its scale was considerably smaller than that of TAIEX futures, suggesting that it was more consistent with tail-risk protection than a large-scale directional bet. [1]\r\n\r\nAn Important Limitation\r\n\r\nForeign investors can also hedge through MSCI Taiwan futures traded on overseas exchanges such as SGX and HKEX.\r\n\r\nLocal TX positions therefore represent only part of their overall hedging exposure. Relying exclusively on TAIEX futures may systematically underestimate the extent of their total hedging activity.\r\n\r\n5. The Four-Quadrant Framework: Increase Signal Weight Only When Indicators Align\r\n\r\nCombining cash-market net buying and selling with changes in net futures positions provides a more complete picture of market positioning.\r\n\r\nThe analytical sequence is straightforward:\r\n\r\nFirst, determine whether the cash and futures signals are aligned. Next, examine whether the futures basis supports that interpretation. Finally, determine whether the changes reflect genuine new positioning or merely contract rollover.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F05bxYtPSKd&scene=im&aid=582478&lang=zh]\r\n\r\nThe four-quadrant framework is not a forecasting model. It is a consistency-checking tool.\r\n\r\nQuadrants ① and ④ carry the greatest analytical weight because the cash and futures indicators move in the same direction. Quadrants ② and ③ require further investigation into market mechanics; otherwise, hedging may be mistaken for directional positioning, or portfolio rotation may be mistaken for capital withdrawal.\r\n\r\nTime horizons should also be considered together. Daily data contain the greatest amount of noise, while five-day and 20-day moving averages are generally more useful for identifying sustained trends.\r\n\r\n6. Practical Case Study: A Five-Step Analysis of a Market Sell-Off\r\n\r\nUsing the period from October 28 through November 6 as an example, we can apply the framework step by step.\r\n\r\nEach step addresses one specific question, and every conclusion must remain capable of being disproved by subsequent data.\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/RSB1mneUnC]\r\n\r\nStep 1: Identify the Quadrant\r\n\r\nOn October 28, foreign investors recorded net cash-market buying of NT$15.957 billion while maintaining 84,917 net short TAIEX futures contracts.\r\n\r\nThis placed the market in Quadrant ③.\r\n\r\nThe coexistence of cash buying and additional short futures exposure should initially be interpreted as potential cash-futures arbitrage or hedging, rather than as a directional market signal.\r\n\r\nStep 2: Look for Continuity\r\n\r\nOn October 29, foreign investors shifted to net cash-market selling of NT$68.815 billion, while their net short futures position increased to 89,522 contracts.\r\n\r\nThe market moved into Quadrant ①.\r\n\r\nForeign investors subsequently recorded net selling in each of the following five trading sessions. Only at this stage did the directional pattern become more established.\r\n\r\nStep 3: Compare the Relative Magnitudes\r\n\r\nFrom October 29 through November 5, foreign investors accumulated NT$614.471 billion in net cash-market selling.\r\n\r\nOver the same period, their net short futures position increased by 10,734 contracts.\r\n\r\nUsing an estimated notional value of approximately NT$8.31 million per contract based on the November 5 index level, the increase represented roughly NT$89.3 billion in notional exposure.\r\n\r\nThis was equivalent to only 14.5% of the cumulative cash-market net selling.\r\n\r\nStep 4: Validate Against the Futures Basis\r\n\r\nOn November 5, the TAIEX plunged 1,374 points, yet TAIEX futures maintained a positive basis of 307 points.\r\n\r\nFutures did not lead the decline. This suggests that short sellers were not aggressively chasing the market lower and that hedging activity was more significant than outright directional selling.\r\n\r\nStep 5: Formulate a Falsifiable Conclusion\r\n\r\nThe working assessment was:\r\n\r\n\"This correction was primarily driven by cash-equity exposure reduction, with futures positions serving mainly as a concurrent hedge.\"\r\n\r\nThis conclusion would be invalidated if cash-market net selling began to narrow while net short futures positions continued to increase and the negative futures basis widened.\r\n\r\nOn November 6, net short TAIEX futures positions increased to 99,930 contracts, while the futures basis narrowed to approximately 50 points below the cash index.\r\n\r\nThe cash and futures signals were beginning to align, although there was not yet evidence of an independent acceleration in short positioning. [5]\r\n\r\n7. Conclusion\r\n\r\nConclusion 1: Net buying and selling represent transaction flows, while futures open interest represents outstanding positions. The two measure different things and cannot substitute for one another.\r\n\r\nConclusion 2: The nature of futures short positions depends on notional exposure, the cash-futures basis, and contract rollover cycles. The instrument with the largest contract count is not necessarily TAIEX futures.\r\n\r\nConclusion 3: Only when cash-market activity, futures positioning, and the futures basis move in the same direction should the signal be assigned greater analytical weight. When they diverge, market mechanics should be investigated first.\r\n\r\n---\r\n\r\n## Data Sources\r\n\r\n[1] Taiwan Futures Exchange (TAIFEX), \"Institutional Investors – By Individual Futures Contract (Daily),\" November 6, 2026.\r\nhttps:\/\/www.taifex.com.tw\/cht\/3\/futContractsDate\r\n\r\n[2] Taiwan Futures Exchange (TAIFEX), \"Daily Trading Information for TAIEX Futures,\" November 5 and November 6, 2026.\r\nhttps:\/\/www.taifex.com.tw\/cht\/3\/futDailyMarketExcel\r\n\r\n[3] HiStock, \"Net Buying and Selling by the Three Major Institutional Investors and Open Interest in Taiwan Stock Index Futures.\"\r\nhttps:\/\/histock.tw\/stock\/three.aspx\r\n\r\n[4] Taipei Exchange (TPEx), \"Daily Trading Details of the Three Major Institutional Investors – Explanatory Notes.\"\r\nhttps:\/\/www.tpex.org.tw\/web\/stock\/3insti\/daily_trade\/3itrade_hedge_result.php\r\n\r\n[5] Infobae \/ EFE, \"Taiwan's Stock Market Falls 4.51% as Regulator Considers Measures to Stabilize the Market,\" November 6, 2026.\r\nhttps:\/\/www.infobae.com\r\n\r\n---\r\n\r\nFarsight Capital Research Team\r\n\r\nDeep Insight. Steady Progress."
            },
            "img": "image\/article-r2.png",
            "video": ""
        },
        {
            "id": "r3",
            "section": "research",
            "cat": "風險管理",
            "homeSlot": 3,
            "tag": {
                "zh": "投資教育",
                "en": "Investor Education"
            },
            "title": {
                "zh": "市場波動加劇時，如何理解投資風險？",
                "en": "How Should Investment Risk Be Understood When Market Volatility Intensifies?"
            },
            "desc": {
                "zh": "波動是讀數，風險是結構，從倉位、波動與風險敞口建立分析框架。",
                "en": "A framework built on positioning, volatility and exposure."
            },
            "body": {
                "zh": "涵遠資本研究部｜投資教育｜2026 年 11 月 6 日\r\n\r\n核心問題：波動加劇時，風險究竟變在哪裡？答案不在跌幅，而在結構。\r\n- 波動是讀數：它測量價格變動的幅度，不區分上漲或下跌，也不等於損失。\r\n- 風險是結構：回撤、資金約束與持有期，決定投資人能不能撐到最後。\r\n- 同一點位，兩種結果：年初進場仍 +37.1%，10\/6 追高已 -20.3%。\r\n- 四種誤讀：把波動、回撤、分散與避險當成風險本身。\r\n- 三個假設會失效：常態、分散、槓桿與流動性。\r\n- 判讀順序：先問生存，再問報酬。\r\n\r\n [img:https:\/\/images.openai.com\/static-rsc-4\/ZPSkj8Q3otNi-BGPYjcTECQTZYbNe3aRoPc0-xrF2WokWq8cH4JxJxQPZrHXvJKteXIvciEwcH6CqZall5BehtRW3MAux-WfwqUmTRcQWssBkMCcIQKJQkTwNVcuaZ4jUQwzRxCM0s8Kb2zjPOfwEEWAB9OcY6z9VPL6SxKJF9yY677ceV67GcNGCfXJiZVc?purpose=fullsize]\r\n\r\n一、波動是讀數，風險是結構\r\n波動率衡量的是報酬的離散程度，它把上漲與下跌同樣計入。一個月漲 20% 與跌 20%，在波動率上是同一件事；對投資人而言，兩者卻是相反的結果。把波動當成風險，等於用溫度計判斷病情。\r\n\r\n風險的核心，是「達不到目標」或「被迫退出」。同樣一次 20% 的回撤，對無槓桿、持有期三年的投資人只是路徑上的噪音；對有追繳線的槓桿帳戶，可能是永久損失；對持有期只有一季的資金，則是實質虧損。回撤本身不傷人，被迫在回撤底部賣出才傷人。\r\n\r\n所以波動加劇時，真正改變的是三件事：價格路徑變粗、資產間相關性上升、資金約束收緊。風險的總量沒有突然變大，變大的是它的可見度與約束力。\r\n\r\n二、四種把波動當風險的誤讀\r\n把波動當成風險，通常會導出四種誤讀，而這四種誤讀在壓力期最貴。\r\n誤讀一：波動上升等於下跌。波動是雙向的，但在槓桿與風險平價的世界裡，波動上升確實常伴隨去槓桿與下跌。這是機制，不是自然法則；把機制當因果，會在反彈修復時缺席。\r\n\r\n誤讀二：回撤等於風險。回撤是已實現的過去，風險是未實現的未來。盯著回撤做決策，容易在底部賣出，再於修復段錯過。\r\n\r\n誤讀三：分散等於保險。相關性在壓力時上升，分散給的是平常的安心，不是危機的保護。\r\n\r\n誤讀四：避險等於免費。期權有時間價值，期貨有轉倉成本與基差風險，避險比率過高會把報酬一起避掉。\r\n\r\n三、同一點位，兩種敘事\r\n2026 年的台股提供了一個乾淨的對照。2025 年底加權指數收 28,963.60 點；10 月 6 日收 49,822.55 點，盤中最高 49,968.92 點；11 月 6 日收 39,701.64 點。[1][2][3]\r\n\r\n對年初進場的人，這個點位仍是 +37.1%；對 10 月 6 日收盤進場的人，則是 -20.3%。同一個 39,701.64，兩種完全不同的風險敘事。指數沒有變，變的是進場點、持有期與資金結構。\r\n\r\n把幣別放進來，差異更大。11 月台幣兌美元貶破 33，為 2025 年 4 月以來首見；外資的美元計價報酬同時承受股價與匯率兩層變動，本地投資人面對的則是進口成本與通膨，而不是匯兌損失。風險不是市場的屬性，是「誰在看、用什麼幣別、能撐多久」的函數。[5]\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/Oic-RM4ZcMnSRzrx7AOd_xwnsqHF18K2EW4lWiDytbwwBjHOWJ905FnLyoE9MJZGZCtM8lCUdrVcgLZOnIgddkP3CrhaE6Poeecwpu4FCqWk-JbciFwmKOhbuDpkNVJ21SJlZMEyhc-JMebQdi7WIHAcvrBbeKU8IW602Uo7Iu7-nJzLneLhf9AUN-oVH2d8?purpose=fullsize]\r\n\r\n四、壓力下，三個假設會失效\r\n多數風險模型建立在三個隱含假設上，而波動加劇時，正是這三個假設同時鬆動的時刻。\r\n\r\n常態分布。以年化波動 20% 計，單日 95% 的 VaR 約 2.1%；11 月 6 日單日 -4.51% 已是它的兩倍，開盤跳空 -6.4% 則是三倍。厚尾與跳空不是例外，而是壓力期的常態，VaR 與固定停損會同時低估。\r\n\r\n分散化。壓力時相關性趨同，股票、債券、區域與因子一起下跌。不同的代號，不等於不同的風險來源；真正的分散，是持有不同驅動因素。\r\n\r\n槓桿與流動性。保證金與買賣價差同時惡化，波動目標與風險平價策略被動去槓桿，賣壓在同一個方向疊加。這時的下跌，有一部分是機械性的，不是觀點改變。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FqO2iodiaQs&scene=im&aid=582478&lang=zh]\r\n\r\n五、四層透鏡：把風險拆開看\r\n與其預測波動，不如按順序檢查四個層次：價格、流動性、資金、行為。順序不能反，因為後三層決定第一層的損失會不會變成永久損失。\r\n\r\n價格層問最壞一天是多少，看的是跳空與路徑，不是日均波動。流動性層問需要賣時賣不賣得掉，看的是價差、深度與 ETF 折溢價。資金層問會不會被迫離場，看的是槓桿、追繳線與期限。行為層問決策是否被波動綁架，看的是新聞密度與自己的反應速度。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FsYU3ZCFDSU&scene=im&aid=582478&lang=zh]\r\n\r\n四層之中，只有價格層可以靠工具對沖；資金層只能靠結構，降低槓桿、拉長期限；行為層只能靠規則，事先寫好的再平衡與檢視頻率。這也解釋了為什麼「買保險」解決不了所有問題。\r\n\r\n六、本週讀數：-20% 的回撤，+37% 的年度\r\n11 月 5 日加權指數收 41,574.72 點、跌 3.20%；11 月 6 日開盤 38,915.76 點，較前一日收盤跳空 -6.4%，盤中最低 37,837.05 點、約 -9.0%，終場收 39,701.64 點、跌 4.51%，盤中最大跌點 3,737.6 點創紀錄。[4]\r\n\r\n這組數字的意義不在跌幅，而在路徑：開盤即跳空、盤中再破底、收盤收斂。任何以收盤價設計的停損，都無法在跳空那一刻執行；任何以日均波動估算的風險，都會低估當日 9% 的瞬時暴露。\r\n\r\n同日監管層表示研議穩定措施、不排除限制放空；台幣續貶破 33。[6]更值得留意的是反彈的時點：市場稱融資盤強制平倉壓力大致釋放後，指數才自盤中低點回升。決定當日低點的，不是估值，而是誰被迫賣。外資現貨六日賣超 6,144.71 億元、期貨同步避險，也應讀成資金預算調整，而非方向判斷。[7]\r\n\r\n七、結論：三個問題贯穿策略\r\n結論一：波動是讀數，風險是結構；回撤、資金約束與持有期，才決定損失是否永久。\r\n結論二：壓力期要檢查三個假設：常態、分散、槓桿與流動性。它們同時失效，才是波動被放大的原因。\r\n結論三：判讀順序應是價格、流動性、資金、行為；先確保不被強制離場，再談報酬。\r\n三個問題：持有期多長？最大可承受回撤多少？資金會不會在回撤底部被迫退出？答不出來，任何進出都只是猜測。\r\n\r\n\r\n數據來源\r\n[1] 臺灣證券交易所，〈發行量加權股價指數歷史資料〉。https:\/\/wwwc.twse.com.tw\/zh\/indices\/taiex\/mi-5min-hist.html\r\n[2] 臺灣證券交易所，〈2025 年報〉。https:\/\/wwwc.twse.com.tw\/downloads\/en\/about\/company\/annual\/2025\/P1.pdf\r\n[3] 台視新聞，〈台股 10 月 6 日盤中攻上 49,968.92 點〉，2026 年 10 月 6 日。https:\/\/news.ttv.com.tw\r\n[4] Infobae／EFE，〈La Bolsa de Taiwán cae un 4,51 % y el regulador estudiará medidas para calmar el mercado〉，2026 年 11 月 6 日。https:\/\/www.infobae.com\r\n[5] 新浪財經，〈台幣兌美元貶破 33 元大關，創 2025 年 4 月以來首見〉，2026 年 11 月 6 日。https:\/\/finance.sina.com.cn\r\n[6] 格隆匯，〈台股重挫 4.5% 收復年線，當地政府或將出手穩市〉，2026 年 11 月 6 日。https:\/\/finance.sina.com.cn\r\n[7] HiStock 嗨投資，〈上市三大法人買賣超與台股期貨未平倉〉。https:\/\/histock.tw\/stock\/three.aspx\r\n\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "Farsight Capital Research Team | Investor Education | November 6, 2026\r\n\r\nCore question: When volatility intensifies, what exactly changes about risk? The answer lies not in the magnitude of the decline, but in the underlying structure.\r\n\r\n- Volatility is a measurement: it measures the magnitude of price movements, without distinguishing between gains and losses, and does not in itself represent a loss.\r\n- Risk is structural: drawdowns, funding constraints, and investment horizons determine whether investors can stay invested through the end of a market cycle.\r\n- The same index level, two different outcomes: investors who entered at the beginning of the year are still up 37.1%, while those who chased the market on October 6 are down 20.3%.\r\n- Four common misinterpretations: treating volatility, drawdowns, diversification, and hedging as risk itself.\r\n- Three assumptions can break down: normal market conditions, diversification, and the assumptions surrounding leverage and liquidity.\r\n- The order of analysis: survival first, returns second.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/ZPSkj8Q3otNi-BGPYjcTECQTZYbNe3aRoPc0-xrF2WokWq8cH4JxJxQPZrHXvJKteXIvciEwcH6CqZall5BehtRW3MAux-WfwqUmTRcQWssBkMCcIQKJQkTwNVcuaZ4jUQwzRxCM0s8Kb2zjPOfwEEWAB9OcY6z9VPL6SxKJF9yY677ceV67GcNGCfXJiZVc?purpose=fullsize]\r\n\r\n1. Volatility Is a Measurement; Risk Is Structural\r\n\r\nVolatility measures the dispersion of returns, treating gains and losses equally. A 20% gain in one month and a 20% decline over the same period are equivalent from a volatility perspective. For investors, however, they represent opposite outcomes. Treating volatility as risk is like using a thermometer to diagnose an illness.\r\n\r\nThe essence of risk lies in the possibility of failing to achieve an objective or being forced to exit an investment. The same 20% drawdown may be no more than noise along the investment path for an unleveraged investor with a three-year horizon. For a leveraged account subject to margin calls, it may result in a permanent loss. For capital with a holding period of only one quarter, it represents a realized loss. A drawdown itself does not necessarily cause lasting damage; being forced to sell at the bottom of a drawdown does.\r\n\r\nWhen volatility intensifies, three things actually change: price movements become more extreme, correlations between assets increase, and funding constraints tighten. The overall amount of risk does not suddenly expand. What increases is its visibility and its power to constrain investors.\r\n\r\n2. Four Misinterpretations That Confuse Volatility with Risk\r\n\r\nTreating volatility as risk typically leads to four common misinterpretations. During periods of market stress, each can prove costly.\r\n\r\nMisinterpretation 1: Rising volatility means the market will fall.\r\n\r\nVolatility is bidirectional. However, in a world of leverage and risk-parity strategies, rising volatility is indeed often accompanied by deleveraging and market declines. This is a mechanism, not a natural law. Mistaking a mechanism for an inevitable causal relationship can leave investors on the sidelines when markets rebound.\r\n\r\nMisinterpretation 2: A drawdown is equivalent to risk.\r\n\r\nA drawdown is a realized event in the past; risk concerns what has yet to happen. Making decisions based solely on the size of a drawdown can lead investors to sell near the bottom and miss the subsequent recovery.\r\n\r\nMisinterpretation 3: Diversification is equivalent to insurance.\r\n\r\nCorrelations tend to rise under stress. Diversification provides a degree of comfort under normal market conditions, but it does not guarantee protection during a crisis.\r\n\r\nMisinterpretation 4: Hedging is free.\r\n\r\nOptions carry time value, while futures involve rollover costs and basis risk. An excessively high hedge ratio can eliminate potential returns along with risk.\r\n\r\n3. The Same Index Level, Two Different Narratives\r\n\r\nTaiwan's equity market in 2026 provides a clear comparison. The TAIEX closed at 28,963.60 points at the end of 2025. On October 6, it closed at 49,822.55 points, after reaching an intraday high of 49,968.92. On November 6, it closed at 39,701.64 points. [1][2][3]\r\n\r\nFor investors who entered at the beginning of the year, the index remains up 37.1%. For those who entered at the October 6 closing level, it represents a loss of 20.3%. The same index level of 39,701.64 tells two entirely different risk stories. The index has not changed; what differs is the entry point, investment horizon, and funding structure.\r\n\r\nThe difference becomes even more pronounced when currency exposure is taken into account. In November, the New Taiwan dollar weakened beyond 33 against the US dollar for the first time since April 2025. Foreign investors measuring returns in US dollars face the combined impact of equity-price movements and exchange-rate fluctuations. Local investors, by contrast, are more directly exposed to higher import costs and inflation rather than foreign-exchange losses. Risk is not an inherent property of the market. It is a function of who is assessing it, which currency they use, and how long they can remain invested. [5]\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/Oic-RM4ZcMnSRzrx7AOd_xwnsqHF18K2EW4lWiDytbwwBjHOWJ905FnLyoE9MJZGZCtM8lCUdrVcgLZOnIgddkP3CrhaE6Poeecwpu4FCqWk-JbciFwmKOhbuDpkNVJ21SJlZMEyhc-JMebQdi7WIHAcvrBbeKU8IW602Uo7Iu7-nJzLneLhf9AUN-oVH2d8?purpose=fullsize]\r\n\r\n4. Under Stress, Three Assumptions Can Break Down\r\n\r\nMost risk models are built on three implicit assumptions. When volatility intensifies, these are precisely the assumptions that begin to weaken simultaneously.\r\n\r\nNormal distribution.\r\n\r\nAssuming annualized volatility of 20%, the one-day 95% VaR is approximately 2.1%. The TAIEX's 4.51% single-day decline on November 6 was already more than twice that level, while the 6.4% opening gap-down was approximately three times as large. Fat tails and price gaps are not merely exceptions; they become characteristic features of stressed markets. Both VaR and fixed stop-loss strategies can underestimate the potential scale of losses under these conditions.\r\n\r\nDiversification.\r\n\r\nCorrelations tend to converge under stress, causing equities, bonds, regions, and investment factors to decline together. Different ticker symbols do not necessarily represent different sources of risk. Genuine diversification requires exposure to different underlying return drivers.\r\n\r\nLeverage and liquidity.\r\n\r\nMargin requirements and bid-ask spreads can deteriorate simultaneously. Volatility-targeting and risk-parity strategies may be forced to deleverage, causing selling pressure to accumulate in the same direction. Part of the resulting decline is mechanical rather than the consequence of a change in investment views.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FqO2iodiaQs&scene=im&aid=582478&lang=zh]\r\n\r\n5. Four Analytical Lenses: Breaking Risk Down into Its Components\r\n\r\nRather than attempting to predict volatility, investors should examine risk in a defined sequence across four layers: price, liquidity, funding, and behavior. The order matters because the latter three determine whether a loss at the price level becomes permanent.\r\n\r\nThe price layer asks: How severe could the worst trading day be? The focus is on opening gaps and price trajectories, not average daily volatility.\r\n\r\nThe liquidity layer asks: If an asset needs to be sold, can it actually be sold? The relevant indicators include bid-ask spreads, market depth, and ETF premiums or discounts to net asset value.\r\n\r\nThe funding layer asks: Could the investor be forced to exit? The focus is on leverage, margin-call thresholds, and investment horizons.\r\n\r\nThe behavioral layer asks: Are investment decisions being driven by volatility? Relevant considerations include the intensity of news flow and the investor's own speed of reaction.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FsYU3ZCFDSU&scene=im&aid=582478&lang=zh]\r\n\r\nOf these four layers, only price risk can be directly addressed through hedging instruments. Funding risk must be managed through structural adjustments, including reducing leverage and extending investment horizons. Behavioral risk requires predefined rules, such as scheduled rebalancing and review intervals. This explains why buying insurance cannot solve every risk-management problem.\r\n\r\n6. This Week's Market Readings: A 20% Drawdown and a 37% Year-to-Date Gain\r\n\r\nOn November 5, the TAIEX closed at 41,574.72 points, down 3.20%. On November 6, it opened at 38,915.76 points, representing a 6.4% gap-down from the previous day's close. It subsequently fell to an intraday low of 37,837.05 points, approximately 9.0% below the previous close, before recovering to close at 39,701.64 points, down 4.51%. The maximum intraday point decline of 3,737.6 points set a record. [4]\r\n\r\nThe significance of these figures lies not simply in the size of the decline, but in the price trajectory: a gap-down at the open, a further decline to a new intraday low, and a partial recovery by the close. Any stop-loss strategy based on closing prices could not have been executed at the moment the market gapped lower. Any risk estimate based on average daily volatility would have underestimated the day's intraday exposure of approximately 9%.\r\n\r\nOn the same day, regulators indicated that they were considering stabilization measures and did not rule out restrictions on short selling. The New Taiwan dollar continued to weaken beyond 33 against the US dollar. [6] More importantly, the timing of the rebound deserves attention. According to market commentary, the index began recovering from its intraday low only after much of the forced-liquidation pressure from margin-financed positions had been released. The factor determining the day's low was not valuation, but who was forced to sell.\r\n\r\nForeign investors' net selling of NT$614.471 billion in the cash market over the six trading sessions, together with their concurrent hedging in futures, should also be interpreted as an adjustment in capital and risk budgets rather than a definitive directional view. [7]\r\n\r\n7. Conclusion: Three Questions That Should Guide Investment Strategy\r\n\r\nConclusion 1: Volatility is a measurement; risk is structural. Drawdowns, funding constraints, and investment horizons determine whether losses become permanent.\r\n\r\nConclusion 2: During periods of stress, three assumptions must be examined: normal market conditions, diversification, and the assumptions surrounding leverage and liquidity. Their simultaneous breakdown is what amplifies market volatility.\r\n\r\nConclusion 3: Risk should be assessed in the following order: price, liquidity, funding, and behavior. The first priority is to ensure that investors are not forced out of the market before considering potential returns.\r\n\r\nThree questions investors should ask:\r\n\r\nHow long is the investment horizon? What is the maximum tolerable drawdown? Could funding constraints force an exit at the bottom of a drawdown?\r\n\r\nIf these questions cannot be answered, any decision to enter or exit the market is little more than speculation.\r\n\r\nData Sources\r\n\r\n[1] Taiwan Stock Exchange, “Historical Data of the TAIEX.”  \r\nhttps:\/\/wwwc.twse.com.tw\/zh\/indices\/taiex\/mi-5min-hist.html\r\n\r\n[2] Taiwan Stock Exchange, “2025 Annual Report.”  \r\nhttps:\/\/wwwc.twse.com.tw\/downloads\/en\/about\/company\/annual\/2025\/P1.pdf\r\n\r\n[3] TTV News, “TAIEX Climbs to an Intraday High of 49,968.92 Points on October 6,” October 6, 2026.  \r\nhttps:\/\/news.ttv.com.tw\r\n\r\n[4] Infobae \/ EFE, “Taiwan's Stock Market Falls 4.51% as Regulator Considers Measures to Stabilize the Market,” November 6, 2026.  \r\nhttps:\/\/www.infobae.com\r\n\r\n[5] Sina Finance, “New Taiwan Dollar Weakens Beyond NT$33 Against the US Dollar, First Time Since April 2025,” November 6, 2026.  \r\nhttps:\/\/finance.sina.com.cn\r\n\r\n[6] Gelonghui, “Taiwan Stocks Plunge 4.5% but Recover Above the Annual Moving Average; Local Government May Step In to Stabilize the Market,” November 6, 2026.  \r\nhttps:\/\/finance.sina.com.cn\r\n\r\n[7] HiStock, “Net Buying and Selling by the Three Major Institutional Investors and Open Interest in TAIEX Futures.”  \r\nhttps:\/\/histock.tw\/stock\/three.aspx\r\n\r\nFarsight Capital Research Team | Deep Insight. Steady Progress."
            },
            "img": "image\/article-r3.png",
            "video": ""
        },
        {
            "id": "r4",
            "section": "research",
            "cat": "市場評論",
            "homeSlot": 0,
            "tag": {
                "zh": "市場評論",
                "en": "Market Commentary"
            },
            "title": {
                "zh": "宏觀預期變化如何影響市場定價？",
                "en": "How Do Changes in Macroeconomic Expectations Affect Market Pricing?"
            },
            "desc": {
                "zh": "結合經濟數據、利率預期與市場反應展開討論。",
                "en": "An analysis of how economic data, interest rate expectations, and market reactions interact."
            },
            "body": {
                "zh": "宏觀預期如何進入價格：預期差、三條鏈與已定價程度\r\n涵遠資本研究部｜2026 年 10 月 11 日\r\n\r\n核心問題：宏觀預期變化，究竟透過什麼進入價格？答案不在數據好壞，而在鏈路與預期差。\r\n- 定價的是預期差：10 月 6 日台股創高，年初至今漲約 72%，好消息多已反映。\r\n- 三條鏈：現金流、貼現率、風險溢酬；三個時代各由一條主導。\r\n- 三個案例：2020 風險溢酬、2022 貼現率、2023–2026 現金流。\r\n- 前瞻讀數：利率、匯率、信用價差、波動率。\r\n- 判讀順序：哪條鏈、舊共識、傳導路徑、已定價多少。\r\n\r\n一、價格定的是預期差，不是數據水準\r\n10 月 6 日，加權指數盤中觸及 49,968.92 點、收 49,822.55 點；相對 2025 年底的 28,963.60 點，年初至今上漲約 72%。[1][2]\r\n支撐這段漲勢的是實體數據：8 月外銷訂單年增 71.4%，出口連月創高，AI 伺服器與先進製程需求仍在擴張。[3]\r\n但價格評價的是未來的現金流，不是已實現的訂單。當好消息連續三季被驗證，它已不再是「新信息」；能推動下一段行情的，是增速能不能維持、折現假設會不會變。這就是預期差：新預期減舊共識。\r\n還有一層時間差：估值對預期的反應以日計，訂單與產能對現金流的反應以季計。價格先動、數據後到，不是矛盾，而是更新頻率不同。等到訂單真的下滑，估值通常已修正過一輪。\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/TH7RHER4Qa3e_PgZ34IxQdQWalk65qfLNWpxW1wMRFDNuIaU1uFhNDTSO4iik9fARjo7wY-KvDGZzJCjQnIzqNs7sB41BmiFWkqAWv3e1WryPi0lq7swez8e7umwc0mu2_Yp7D5gO1eS62BgX1OL8OomVo3CLxpMvE2mkeUfWsY?purpose=inline]\r\n\r\n二、宏觀進入價格的三條鏈\r\n宏觀變數不會直接落在股價，必須經過三條鏈之一，而多數時候只有一條主導。\r\n現金流鏈：資本開支預期→雲廠訂單→晶圓需求→電子權值 EPS，決定「分子」。貼現率鏈：政策利率與長端殖利率→折現率→長久期資產估值，決定「分母」。風險溢酬鏈：監管、地緣與政策不確定性→要求更高補償→外資流向、匯率與信用價差，決定「折現率之上的額外折扣」。\r\n三條鏈的相對強弱會隨周期輪換。認錯主導鏈，就會用對的數據解釋錯的行情。\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/LDZGHRBW2nTSacKwHxu2IlBWMUIl2OdFP4OczgoTEcl2QYsOhKU8CjDvrOfi0I3UNYpJyhZzrvH8RpJfmXAZjymQeeFs-l5TbsF6V6H1T8-ITNkwt8PsGWA9SIMbkqJSH4-Nh5nmnjreTyBirv_1Pf5dX3a0Ej7MhgR0r2lLr1I?purpose=inline]\r\n\r\n三、三個時代，三條主導鏈\r\n2020：風險溢酬與現金流同時惡化。疫情讓經濟活動停擺，需求與供應鏈同時斷裂；Fed 緊急降息至 0–0.25%，貼現率大幅下降，仍無法阻止台股由 2020 年 1 月的 12,197 點，跌至 3 月 19 日盤中的 8,681 點，跌幅約 29%。當分子被按下暫停，分母再便宜也沒有用。[4]\r\n2022：貼現率單獨主導。企業獲利仍在成長，但 Fed 快速升息、10 年期美債殖利率突破 4%，長久期資產的估值先被壓縮。台股由 1 月的 18,619 點跌至 10 月的 12,619 點，跌幅約 32%，幾乎全來自乘數，而非 EPS。[5][6]\r\n2023–2026：現金流重新主導。四大雲廠商資本開支由 4,100 億美元增至 7,250 億美元，訂單與出口連月上修，台股由 12,619 點升至 10 月 6 日的 49,968 點，漲幅約 296%，即使貼現率仍處高位。[7]\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2Fidlncut5o4&scene=im&aid=582478&lang=zh]\r\n\r\n三個時代的共同點，是每一輪都有一個「被重新定價的假設」：2020 是需求能不能存在，2022 是資金成本，2023–2026 是資本開支能不能續強。行情方向由那條鏈決定，其餘兩條只是背景。\r\n四、預期的四個前瞻讀數\r\n宏觀預期真正的前瞻訊號不在股價，而在其他市場：利率定價政策路徑，匯率定價相對成長與資金流向，信用價差定價違約與風險偏好，波動率定價不確定性本身。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FfgbiFfUgQK&scene=im&aid=582478&lang=zh]\r\n\r\n用四個讀數交叉驗證，可以避免「新聞很壞、但價格早已反映」的重複交易，也能提早看到「新聞還很好、價格已先轉」的背離。2022 年的殖利率與 2020 年的信用價差，都是比股市更早轉向的例子。\r\n\r\n五、已定價程度：比方向更重要的問題\r\n10 月 6 日的 49,968 點，是「好消息已充分定價」的位置：訂單、出口與 AI 需求都已被三季驗證，估值也反映了高成長假設。此時該問的不是「AI 好不好」，而是「還有多少預期可以上修」。\r\n三個交叉驗證。一看債市：長端殖利率是否已反映寬鬆；二看匯率：資金是否仍在流入；三看信用價差：最邊際的借款人是否還拿得到錢。三者同時轉弱，才是預期真正轉向；只有股價下跌而三者未動，多半是估值修正，不是周期反轉。\r\n這也是「已定價程度」的實用價值：同一個利空，在預期低時是買點，在預期高時是起跌點。方向判斷只對一半，定價程度決定另一半。\r\n\r\n六、判讀順序：四問\r\n一問哪個預期變了：現金流、貼現率，還是風險溢酬？二問舊共識在哪：賣方預估、期貨與選擇權隱含、債市定價。三問傳到哪條鏈：哪類資產、多長時間。四問已定價多少：用殖利率、匯率、價差與波動率交叉驗證。\r\n\r\n順序不能反：先確認鏈路，再談方向；先確認已定價程度，再談交易。以 2022 年為例，第一問的答案是貼現率，第二問的共識是「升息只是暫時」，第三問傳到長久期成長股，第四問用殖利率與美元交叉檢查——四個答案拼起來，才是一句可證偽的判斷。\r\n\r\n七、結論\r\n結論一：宏觀預期透過三條鏈進入價格，多數時候只有一條主導；認錯鏈路，數據再多也解釋不了行情。\r\n結論二：定價的是預期差與持續性，不是數據水準；好消息被驗證三次之後，就不再是新的催化劑。\r\n結論三：已定價程度比方向更重要；債市、匯市與信用價差是最直接的交叉驗證。\r\n待驗證：資本開支指引是否再上修、利率路徑是否轉向、外資流入是否延續。三者決定 2026 年的現金流鏈還能主導多久。\r\n\r\n數據來源\r\n[1] 台視新聞，〈台股 10 月 6 日盤中攻上 49,968.92 點〉，2026 年 10 月 6 日。https:\/\/news.ttv.com.tw\r\n[2] 臺灣證券交易所，〈2025 年報〉。https:\/\/wwwc.twse.com.tw\r\n[3] TradingEconomics，〈台灣外銷訂單〉，2026 年 10 月。https:\/\/zh.tradingeconomics.com\r\n[4] 臺灣證券交易所，〈2020 年報〉（3 月 19 日盤中低點 8,681.34 點）。https:\/\/wwwc.twse.com.tw\r\n[5] ETtoday 財經雲，〈台股還會回落？富邦投顧董座估明年下半年上看萬六〉，2022 年 11 月 28 日。https:\/\/finance.ettoday.net\r\n[6] FRED，〈10-Year Treasury Constant Maturity Rate〉。https:\/\/fred.stlouisfed.org\r\n[7] 證券時報，〈全球 AI 概念股暴跌，AI「算力稀缺」信仰開始動搖？〉，2026 年 6 月 24 日。https:\/\/www.stcn.com\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "How Macroeconomic Expectations Are Reflected in Prices: Expectation Gaps, Three Transmission Channels, and the Extent to Which Expectations Are Already Priced In\r\n\r\nFarsight Capital Research Team | October 11, 2026\r\n\r\nCore question: Through what mechanisms do changes in macroeconomic expectations enter market prices? The answer lies not in whether the data are good or bad, but in the transmission channels and the gap between expectations and reality.\r\n\r\n- Markets price changes in expectations, not simply data outcomes: Taiwan's stock market reached a record high on October 6, up approximately 72% year to date, suggesting that much of the positive news had already been priced in.\r\n- Three transmission channels: cash flows, discount rates, and risk premiums. Each has dominated a different market cycle.\r\n- Three historical cases: risk premiums in 2020, discount rates in 2022, and cash flows from 2023 to 2026.\r\n- Forward-looking indicators: interest rates, exchange rates, credit spreads, and volatility.\r\n- Analytical sequence: identify the dominant channel, establish the previous consensus, trace the transmission path, and assess how much has already been priced in.\r\n\r\n1. Markets Price Changes in Expectations, Not the Absolute Level of Data\r\n\r\nOn October 6, the TAIEX reached an intraday high of 49,968.92 points and closed at 49,822.55 points. Compared with its year-end 2025 level of 28,963.60 points, the index had gained approximately 72% year to date. [1][2]\r\n\r\nThe rally was supported by tangible economic data. Taiwan's export orders increased 71.4% year on year in August, exports continued to reach new highs, and demand for AI servers and advanced semiconductor manufacturing remained on an expansionary trajectory. [3]\r\n\r\nHowever, market prices reflect expectations of future cash flows, not orders that have already been realized. When positive developments have been confirmed over three consecutive quarters, they are no longer new information. What can drive the next leg of the market is whether growth can be sustained and whether discount-rate assumptions will change. This is the expectation gap: new expectations minus the previous consensus.\r\n\r\nThere is also a timing mismatch. Valuations respond to changes in expectations over days, while orders and production capacity translate into cash flows over quarters. Prices move first and the data follow later. This is not a contradiction; it reflects different update frequencies. By the time orders actually decline, valuations have often already undergone a round of adjustment.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/TH7RHER4Qa3e_PgZ34IxQdQWalk65qfLNWpxW1wMRFDNuIaU1uFhNDTSO4iik9fARjo7wY-KvDGZzJCjQnIzqNs7sB41BmiFWkqAWv3e1WryPi0lq7swez8e7umwc0mu2_Yp7D5gO1eS62BgX1OL8OomVo3CLxpMvE2mkeUfWsY?purpose=inline]\r\n\r\n2. Three Channels Through Which Macroeconomic Factors Enter Market Prices\r\n\r\nMacroeconomic variables do not translate directly into stock prices. They must pass through one of three transmission channels, and in most market environments, only one is dominant.\r\n\r\nThe cash flow channel:\r\n\r\nCapital expenditure expectations → cloud service providers' orders → wafer demand → EPS of major electronics constituents.\r\n\r\nThis channel determines the numerator in valuation.\r\n\r\nThe discount rate channel:\r\n\r\nPolicy rates and long-term bond yields → discount rates → valuations of long-duration assets.\r\n\r\nThis channel determines the denominator.\r\n\r\nThe risk premium channel:\r\n\r\nRegulatory, geopolitical, and policy uncertainty → higher required compensation → foreign capital flows, exchange rates, and credit spreads.\r\n\r\nThis channel determines the additional discount applied on top of the discount rate.\r\n\r\nThe relative importance of these three channels rotates over the course of an economic cycle. Misidentifying the dominant channel means using the right data to explain the wrong market movement.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/LDZGHRBW2nTSacKwHxu2IlBWMUIl2OdFP4OczgoTEcl2QYsOhKU8CjDvrOfi0I3UNYpJyhZzrvH8RpJfmXAZjymQeeFs-l5TbsF6V6H1T8-ITNkwt8PsGWA9SIMbkqJSH4-Nh5nmnjreTyBirv_1Pf5dX3a0Ej7MhgR0r2lLr1I?purpose=inline]\r\n\r\n3. Three Market Eras, Three Dominant Transmission Channels\r\n\r\n2020: Risk premiums and cash flows deteriorated simultaneously.\r\n\r\nThe pandemic brought economic activity to a standstill, disrupting both demand and supply chains. The Federal Reserve (Fed) cut interest rates urgently to 0–0.25%. Despite the sharp decline in discount rates, the TAIEX fell from 12,197 points in January 2020 to an intraday low of 8,681 points on March 19, a decline of approximately 29%.\r\n\r\nWhen the numerator is effectively put on hold, a cheaper denominator is not enough to support valuations. [4]\r\n\r\n2022: The discount rate became the dominant factor.\r\n\r\nCorporate earnings were still growing, but the Fed raised interest rates rapidly, and the 10-year US Treasury yield moved above 4%. Valuations of long-duration assets were compressed first.\r\n\r\nThe TAIEX declined from 18,619 points in January to 12,619 points in October, a fall of approximately 32%. The decline was driven almost entirely by valuation multiples rather than EPS. [5][6]\r\n\r\n2023–2026: Cash flows regained dominance.\r\n\r\nCapital expenditure by the four major cloud service providers increased from US$410 billion to US$725 billion. Orders and exports were repeatedly revised upward, and the TAIEX rose from 12,619 points to 49,968 points on October 6, an increase of approximately 296%, even though discount rates remained elevated. [7]\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2Fidlncut5o4&scene=im&aid=582478&lang=zh]\r\n\r\nThe common thread across these three periods is that each cycle involved an assumption being repriced:\r\n\r\n- 2020: Whether demand could continue to exist.\r\n- 2022: The cost of capital.\r\n- 2023–2026: Whether capital expenditure could continue to strengthen.\r\n\r\nThe direction of the market was determined by the dominant transmission channel, while the other two remained part of the background.\r\n\r\n4. Four Forward-Looking Indicators of Macroeconomic Expectations\r\n\r\nThe most useful forward-looking signals of macroeconomic expectations are not found in stock prices alone, but across other markets.\r\n\r\nInterest rate pricing reflects expectations for the policy path.\r\n\r\nExchange rates reflect relative growth prospects and capital flows.\r\n\r\nCredit spreads reflect default risk and risk appetite.\r\n\r\nVolatility reflects uncertainty itself.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FfgbiFfUgQK&scene=im&aid=582478&lang=zh]\r\n\r\nCross-checking these four indicators helps investors avoid repeatedly trading on negative headlines that have already been reflected in prices. It can also provide an early warning when the news remains positive but market prices have already begun to turn. The movement in bond yields in 2022 and credit spreads in 2020 are examples of indicators that turned ahead of the equity market.\r\n\r\n5. The Extent to Which Expectations Are Priced In: A More Important Question Than Direction\r\n\r\nThe TAIEX level of 49,968 points on October 6 represented a market in which positive expectations had been substantially priced in. Orders, exports, and AI demand had all been validated over three consecutive quarters, while valuations already reflected assumptions of strong growth.\r\n\r\nAt this stage, the relevant question was no longer whether AI fundamentals were positive, but how much further expectations could be revised upward.\r\n\r\nThree cross-checks are essential.\r\n\r\nFirst, examine the bond market: Have long-term yields already priced in monetary easing?\r\n\r\nSecond, examine the foreign exchange market: Is capital still flowing in?\r\n\r\nThird, examine credit spreads: Can even the marginal borrowers still access financing?\r\n\r\nOnly when all three indicators weaken simultaneously is there stronger evidence that expectations are genuinely turning. If stock prices decline while these three indicators remain unchanged, the move is more likely to reflect a valuation adjustment than a reversal of the economic cycle.\r\n\r\nThis is the practical value of assessing how much has already been priced in. The same negative development may represent a buying opportunity when expectations are low, but the starting point of a decline when expectations are elevated. A directional view only tells half the story; the extent to which expectations are priced in determines the other half.\r\n\r\n6. Analytical Sequence: Four Questions\r\n\r\nQuestion 1: Which expectation has changed? Is it cash flow, the discount rate, or the risk premium?\r\n\r\nQuestion 2: Where was the previous consensus? Examine sell-side forecasts, expectations implied by futures and options, and bond market pricing.\r\n\r\nQuestion 3: Through which transmission channel will the change be felt? Identify the affected asset classes and the time horizon over which the impact is likely to emerge.\r\n\r\nQuestion 4: How much has already been priced in? Cross-check bond yields, exchange rates, credit spreads, and volatility.\r\n\r\nThe sequence should not be reversed. Identify the transmission channel before forming a directional view, and assess how much has already been priced in before considering a trade.\r\n\r\nTake 2022 as an example. The answer to the first question was the discount rate. The prevailing consensus under the second question was that rate hikes would be temporary. The third question identified long-duration growth stocks as the primary transmission point. The fourth question required cross-checking Treasury yields and the US dollar.\r\n\r\nOnly by putting all four answers together can an investor arrive at a judgment that can be tested and potentially disproved.\r\n\r\n## 7. Conclusion\r\n\r\nConclusion 1: Macroeconomic expectations enter market prices through three transmission channels, but in most periods only one is dominant. Misidentifying the channel means that even extensive data may fail to explain market movements.\r\n\r\nConclusion 2: Markets price expectation gaps and the sustainability of developments, not simply the absolute level of economic data. Once positive news has been validated three times, it is no longer a new catalyst.\r\n\r\nConclusion 3: The extent to which expectations are already priced in matters more than direction alone. Bond markets, foreign exchange markets, and credit spreads provide the most direct cross-checks.\r\n\r\nWhat remains to be verified:\r\n\r\nWhether capital expenditure guidance will be revised upward again, whether the interest rate path will turn, and whether foreign capital inflows will continue. These three factors will determine how much longer the cash flow channel can remain dominant in 2026.\r\n\r\nData Sources\r\n\r\n[1] TTV News, “TAIEX Climbs to an Intraday High of 49,968.92 Points on October 6,” October 6, 2026.  \r\nhttps:\/\/news.ttv.com.tw\r\n\r\n[2] Taiwan Stock Exchange, “2025 Annual Report.”  \r\nhttps:\/\/wwwc.twse.com.tw\r\n\r\n[3] Trading Economics, “Taiwan Export Orders,” October 2026.  \r\nhttps:\/\/zh.tradingeconomics.com\r\n\r\n[4] Taiwan Stock Exchange, “2020 Annual Report” (including the intraday low of 8,681.34 points on March 19).  \r\nhttps:\/\/wwwc.twse.com.tw\r\n\r\n[5] ETtoday Finance Cloud, “Will Taiwan Stocks Fall Further? Fubon Securities Investment Services Chairman Sees the Index Reaching 16,000 in the Second Half of Next Year,” November 28, 2022.  \r\nhttps:\/\/finance.ettoday.net\r\n\r\n[6] FRED, “10-Year Treasury Constant Maturity Rate.”  \r\nhttps:\/\/fred.stlouisfed.org\r\n\r\n[7] Securities Times, “Global AI Concept Stocks Plunge: Is the Belief in AI Compute Scarcity Beginning to Waver?” June 24, 2026.  \r\nhttps:\/\/www.stcn.com\r\n\r\nFarsight Capital Research Team | Deep Insight. Steady Progress."
            },
            "img": "image\/835dfb93b19bedb278e791995c2539e470f6718d.png",
            "video": ""
        },
        {
            "id": "k1",
            "section": "knowledge",
            "cat": "市場觀察",
            "homeSlot": 0,
            "tag": {
                "zh": "01",
                "en": "01"
            },
            "title": {
                "zh": "認識市場指標",
                "en": "Understanding Market Indicators"
            },
            "desc": {
                "zh": "外資買賣超、融資餘額、VIX 等指標如何解讀？",
                "en": "Indicators Are Coordinates, Not Answers"
            },
            "body": {
                "zh": "涵遠資本研究部｜2026 年 10 月 11 日\r\n指標滿屏，該看哪一個、又該怎麼看？\r\n\r\n- 四層結構：價格、流動性、資金、基本面，每一層回答不同問題。\r\n- 只能定位：單一指標指出位置，不能單獨定案。\r\n- 背離才有信息：三層同向只是趨勢，不同向才是訊號。\r\n- 先問用途：它回答什麼問題、口徑怎麼定義、多久更新一次。\r\n- 多數滯後：指標以同步或落後為主，很少真正領先。\r\n\r\n一、指標是座標，不是答案\r\n10 月 6 日，加權指數盤中觸及 49,968.92 點、收 49,822.55 點，相對 2025 年底的 28,963.60 點，年初至今上漲約 72%。[1][2]同一組事實，有人讀成多頭確認，有人讀成末段加速。差別不在數據，而在讀法。\r\n指標的功能是定位，不是預測。它指出三件事：價格現在隱含什麼假設、資金用什麼成本在買、分子還在不在長。把指標當預測器，就會在每一次轉折時同時失去方向和信任。\r\n最常見的三種錯位是：把水位當趨勢、把單日當結構、把同步當領先。三者都源自同一個誤會——以為指標的數值本身就是結論。\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/-WSd_gJOuPng6wFtVRknaj4lk0ZOi41KhmjrjQK0bVRi7v7dDfGH7kOR612PoXvqHQrCHAiMNGRfKpYzvny-GbU6qQDyT2zLipCPJSCfJ_SnpEThOWwy3OiC_aElFT0eA04FBKxv9Lhpz5LArMp7IdYA2hvnIXYliUQuARl99_Bdc3CDfY072LMFkX8175uz?purpose=fullsize]\r\n\r\n二、指標的四層地圖\r\n價格層回答「市場已定價什麼」，包括指數、類股輪動與波動率。它最快，也最容易被誤認為原因。\r\n\r\n流動性層回答「這個價位能不能被推動」，看成交金額、週轉率與買賣盤深度。量能萎縮時，價格訊號會先失真。\r\n\r\n資金層回答「誰在買、用什麼錢買」，看外資買賣超、融資餘額與匯率。它記錄的是已成交的事實，因此同步甚至滯後。\r\n\r\n基本面層回答「分子能不能長出來」，看營收、外銷訂單與資本開支。台灣 8 月外銷訂單年增 71.4%，是這一層最直接的讀數。[3]\r\n\r\n四層的節奏不同：基本面慢、資金居中、價格最快。快慢錯配是誤讀的主要原因——用慢資料解釋快價格，或反之。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FXmz3eycDDb&scene=im&aid=582478&lang=zh]\r\n\r\n這張地圖的用法不是「全部都要看」，而是「先確認現在的問題屬於哪一層」。問「為什麼跌」時，答案若在基本面層，價格層的指標再漂亮也沒有意義。\r\n同一層內也有替代關係：成交金額之於週轉率、外資買賣超之於匯率、營收之於訂單。替代指標的選擇標準是同頻率、同口徑、且能互相驗證；若兩個指標永遠同向，只需留一個。\r\n\r\n三、同一指標的三種誤讀\r\n誤讀一：把水位當趨勢。49,968 點是高是低，取決於分母——EPS 與折現率。同一點位，在 EPS 上修與折現率上升兩種情境下，含義完全相反。指數本身不帶估值信息。\r\n\r\n誤讀二：把單日當結構。單日賣超可能只是調整；連續賣超、且期貨部位同步增加，才是結構。判讀要問持續性與部位，而不是單日金額。\r\n\r\n誤讀三：把同步當領先。外資買賣超是成交紀錄，不是意圖；它與價格同日變動，用它預測隔日行情，等於看著後照鏡開車。\r\n三種誤讀的共同解法，是把指標放回它所屬的層，並問一句：這個數字，是別人已經知道的，還是還沒被定價的？\r\n\r\n還有一個容易忽略的維度：指標的有效期會隨市場狀態改變。同一組資金指標，在多頭趨勢裡幾乎不觸發訊號，在流動性緊縮時卻天天發出警告。判讀之前先問「現在是什麼狀態」，比記住指標的歷史準確率更有用。\r\n\r\n四、背離才有信息\r\n價格、資金、基本面三者的組合，比任何單一指標更能說明性質。\r\n2020 年 3 月，價格下跌、基本面預期崩壞、資金撤出，三層同時惡化，最終指數由 12,197 點跌至 8,681 點。[4]2022 年則是價格下跌、資金撤出，但企業獲利仍在上修：跌幅幾乎全來自乘數，指數由 18,619 點跌至 12,619 點。[5]\r\n\r\n2026 年 10 月則相反：價格創高、基本面強、資金仍在流入。三者同向，趨勢成立；但同向也意味著好消息已被定價，該問的是還剩多少空間。三種組合、三種處理，不能用同一套倉位應對。\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/H0Y05P5Up8]\r\n\r\n背離的價值在於排除解釋：當價格與基本面背道而馳，只用基本面解釋行情一定失敗；此時該往資金層與貼現率找原因。\r\n\r\n五、口徑與時效\r\n同一指標常有不同口徑：外資含不含自營、買賣超是金額還是張數、訂單是美元還是台幣、殖利率是名目還是實質。口徑不對齊，比較就沒有意義。\r\n\r\n時效同樣關鍵：訂單每月公布、GDP 每季且會修訂、指數即時。用季頻資料做日頻判斷，是把慢變數當快變數；用初值建立結論，則要為後續修訂留餘量。\r\n\r\n實務上的順序是：先確定問題的頻率，再選同頻率的指標，最後才談判讀。選錯頻率，再精確的數字也是噪音。\r\n\r\n六、結論\r\n結論一：指標是座標而非答案，四層各答一個問題；先定位問題屬於哪一層，再選指標。\r\n結論二：同向只是趨勢，背離才提供信息；三層交叉驗證比單一指標可靠。\r\n結論三：口徑與時效決定指標能不能用；慢資料不做快判斷，初值不當結論。\r\n\r\n數據來源\r\n[1] 台視新聞，〈台股 10 月 6 日盤中攻上 49,968.92 點〉，2026 年 10 月 6 日。https:\/\/news.ttv.com.tw\r\n[2] 臺灣證券交易所，〈2025 年報〉。https:\/\/wwwc.twse.com.tw\r\n[3] TradingEconomics，〈台灣外銷訂單〉，2026 年。https:\/\/zh.tradingeconomics.com\r\n[4] 臺灣證券交易所，〈2020 年報〉（3 月 19 日盤中低點 8,681.34 點）。https:\/\/wwwc.twse.com.tw\r\n[5] ETtoday 財經雲，〈台股還會回落？富邦投顧董座估明年下半年上看萬六〉，2022 年 11 月 28 日。https:\/\/finance.ettoday.net\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "Farsight Capital Research Team | October 11, 2026\r\n\r\nWith indicators filling the screens, which ones should investors watch, and how should they interpret them?\r\n\r\n- A four-layer framework: price, liquidity, capital flows, and fundamentals. Each layer answers a different question.\r\n- Indicators locate; they do not decide: an individual indicator identifies a position or condition, but cannot provide a definitive conclusion on its own.\r\n- Divergences carry information: when all three layers move in the same direction, they indicate a trend; when they diverge, they generate signals worth investigating.\r\n- Start with the purpose: What question does the indicator answer? How is it defined? How frequently is it updated?\r\n- Most indicators are lagging: the majority are coincident or lagging indicators, while genuinely leading indicators are relatively rare.\r\n\r\n1. Indicators Are Coordinates, Not Answers\r\n\r\nOn October 6, the TAIEX reached an intraday high of 49,968.92 points and closed at 49,822.55 points. Compared with its year-end 2025 level of 28,963.60 points, the index had gained approximately 72% year to date. [1][2] The same set of facts led some investors to interpret the market as confirming its bullish trend, while others saw it as an acceleration in the final stage of the rally. The difference lies not in the data, but in how it is interpreted.\r\n\r\nThe function of an indicator is to establish a reference point, not to predict the future. It helps answer three questions: What assumptions are currently embedded in prices? At what cost is capital being deployed? Is the earnings numerator still growing? Treating indicators as forecasting tools means risking both direction and confidence whenever the market turns.\r\n\r\nThree common interpretive errors are confusing a level with a trend, a single trading day with a structural change, and coincident indicators with leading indicators. All three stem from the same misconception: assuming that the numerical value of an indicator is itself the conclusion.\r\n\r\n[img:https:\/\/images.openai.com\/static-rsc-4\/-WSd_gJOuPng6wFtVRknaj4lk0ZOi41KhmjrjQK0bVRi7v7dDfGH7kOR612PoXvqHQrCHAiMNGRfKpYzvny-GbU6qQDyT2zLipCPJSCfJ_SnpEThOWwy3OiC_aElFT0eA04FBKxv9Lhpz5LArMp7IdYA2hvnIXYliUQuARl99_Bdc3CDfY072LMFkX8175uz?purpose=fullsize]\r\n\r\n2. The Four-Layer Map of Market Indicators\r\n\r\nThe price layer answers: “What has the market already priced in?”\r\n\r\nIt includes market indices, sector rotation, and volatility. This layer reacts the fastest and is also the one most easily mistaken for the underlying cause of market movements.\r\n\r\nThe liquidity layer answers: “Can the market be moved at this price level?”\r\n\r\nIt focuses on trading value, turnover, and bid-ask depth. When trading activity contracts, price signals can become distorted first.\r\n\r\nThe capital flow layer answers: “Who is buying, and what capital are they using?”\r\n\r\nIt examines foreign investor net buying and selling, margin financing balances, and exchange rates. These indicators record transactions that have already taken place, making them coincident or even lagging indicators.\r\n\r\nThe fundamental layer answers: “Can the earnings numerator continue to grow?”\r\n\r\nIt examines revenue, export orders, and capital expenditure. Taiwan's export orders increased 71.4% year on year in August, providing one of the most direct readings at this level. [3]\r\n\r\nThe four layers operate at different speeds: fundamentals move slowly, capital flows occupy the middle ground, and prices move fastest. This mismatch in timing is a major source of misinterpretation—using slow-moving data to explain fast-moving prices, or vice versa.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FXmz3eycDDb&scene=im&aid=582478&lang=zh]\r\n\r\nThe purpose of this map is not to suggest that investors must monitor everything. It is to first identify which layer the current question belongs to. When asking why the market is falling, if the answer lies in fundamentals, even the most favorable indicators at the price level may be irrelevant.\r\n\r\nSubstitution is also possible within each layer: trading value can be considered alongside turnover, foreign investor net flows alongside exchange rates, and revenue alongside orders. Substitute indicators should be selected based on matching frequencies, consistent definitions, and the ability to cross-validate one another. If two indicators always move in the same direction, one may be sufficient.\r\n\r\n3. Three Ways to Misread the Same Indicator\r\n\r\nMisinterpretation 1: Confusing a level with a trend.\r\n\r\nWhether 49,968 points is high or low depends on the denominator—EPS and the discount rate. At the same index level, the implications are entirely different under two scenarios: rising EPS estimates or increasing discount rates. The index level itself does not contain valuation information.\r\n\r\nMisinterpretation 2: Confusing a single day with a structural change.\r\n\r\nOne day of net selling may simply reflect portfolio adjustments. Persistent net selling accompanied by a simultaneous increase in futures positions is more indicative of a structural shift. Analysis should focus on persistence and positioning, not merely the net amount traded on a single day.\r\n\r\nMisinterpretation 3: Confusing coincident indicators with leading indicators.\r\n\r\nForeign investor net buying and selling is a record of completed transactions, not an expression of intent. When it moves on the same day as prices, using it to predict the following day's market is like driving while looking in the rearview mirror.\r\n\r\nThe common solution to all three errors is to put each indicator back into its proper layer and ask one question: Is this information already known to the market, or has it yet to be priced in?\r\n\r\nThere is another dimension that is easily overlooked: the effectiveness of an indicator changes with market conditions. The same set of capital flow indicators may generate almost no signals during a sustained bull market, yet issue warnings every day during a liquidity squeeze. Before interpreting an indicator, asking “What kind of market environment are we in?” is more useful than simply remembering its historical accuracy.\r\n\r\n4. Divergences Are Where Information Emerges\r\n\r\nThe interaction between price, capital flows, and fundamentals provides a more meaningful picture than any single indicator.\r\n\r\nIn March 2020, prices fell, fundamental expectations collapsed, and capital flowed out. All three layers deteriorated simultaneously, ultimately driving the TAIEX down from 12,197 points to 8,681 points. [4]\r\n\r\nIn 2022, prices declined and capital flowed out, but corporate earnings estimates were still being revised upward. Almost the entire decline came from valuation multiple compression, with the index falling from 18,619 points to 12,619 points. [5]\r\n\r\nOctober 2026 presented the opposite configuration: prices reached new highs, fundamentals remained strong, and capital continued to flow in. All three moved in the same direction, confirming the prevailing trend. However, their alignment also meant that positive developments had already been priced in. The relevant question was how much upside remained.\r\n\r\nThree different configurations require three different responses. They cannot be managed with the same positioning strategy.\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/H0Y05P5Up8]\r\n\r\nThe value of divergence lies in eliminating incomplete explanations. When prices move in the opposite direction to fundamentals, relying on fundamentals alone to explain market movements will inevitably fall short. At that point, the analysis should turn to capital flows and discount rates for further explanations.\r\n\r\n5. Definitions and Timeliness\r\n\r\nThe same indicator can be reported using different definitions. For example: Does the foreign investor category include proprietary trading? Is net buying or selling measured in monetary value or number of shares? Are orders denominated in US dollars or New Taiwan dollars? Is the yield nominal or real?\r\n\r\nComparisons are meaningless when definitions are not aligned.\r\n\r\nTimeliness is equally important. Export orders are released monthly, GDP is reported quarterly and subject to revision, while market indices update in real time. Using quarterly data to make daily trading decisions means treating slow-moving variables as fast-moving ones. Building conclusions on preliminary data also requires allowing for subsequent revisions.\r\n\r\nThe practical sequence is straightforward: first determine the frequency of the question, then select indicators with a matching frequency, and only then proceed to interpretation. If the frequency is wrong, even the most precise figures become noise.\r\n\r\n6. Conclusion\r\n\r\nConclusion 1: Indicators are coordinates, not answers. Each of the four layers addresses a different question. First identify which layer the issue belongs to, then select the appropriate indicators.\r\n\r\nConclusion 2: Alignment indicates a trend; divergence provides information. Cross-validation across the three layers is more reliable than relying on a single indicator.\r\n\r\nConclusion 3: Definitions and timeliness determine whether an indicator is suitable for use. Slow-moving data should not be used to make fast-moving judgments, and preliminary figures should not be treated as final conclusions.\r\n\r\nData Sources\r\n\r\n[1] TTV News, “TAIEX Climbs to an Intraday High of 49,968.92 Points on October 6,” October 6, 2026.  \r\nhttps:\/\/news.ttv.com.tw\r\n\r\n[2] Taiwan Stock Exchange, “2025 Annual Report.”  \r\nhttps:\/\/wwwc.twse.com.tw\r\n\r\n[3] Trading Economics, “Taiwan Export Orders,” 2026.  \r\nhttps:\/\/zh.tradingeconomics.com\r\n\r\n[4] Taiwan Stock Exchange, “2020 Annual Report” (including the intraday low of 8,681.34 points on March 19).  \r\nhttps:\/\/wwwc.twse.com.tw\r\n\r\n[5] ETtoday Finance Cloud, “Will Taiwan Stocks Fall Further? Fubon Securities Investment Services Chairman Sees the Index Reaching 16,000 in the Second Half of Next Year,” November 28, 2022.  \r\nhttps:\/\/finance.ettoday.net\r\n\r\nFarsight Capital Research Team | Deep Insight. Steady Progress."
            },
            "img": "image\/article-k1.png",
            "video": ""
        },
        {
            "id": "k2",
            "section": "knowledge",
            "cat": "投資教育",
            "homeSlot": 0,
            "tag": {
                "zh": "02",
                "en": "02"
            },
            "title": {
                "zh": "建立分析框架",
                "en": "Build an Analytical Framework"
            },
            "desc": {
                "zh": "框架是證偽工具，不是預測器；依序定義問題、對齊口徑、寫下否證條件、設定追蹤。",
                "en": "Combining macro, fundamentals and price action to find opportunities?"
            },
            "body": {
                "zh": "涵遠資本研究部｜2026 年 10 月 11 日\r\n\r\n如何把資料變成可執行、可檢查的判斷，而不是又多一種說法？\r\n- 框架不是預測器：它的作用是約束自己，不是增加解釋。\r\n- 四步流程：定義問題、對齊口徑、寫下否證條件、設定追蹤。\r\n- 三種誤用：指標堆疊、事後歸因、框架移植。\r\n- 一句話標準：說不出「什麼情況下我錯了」，就不是判斷。\r\n\r\n一、框架不是預測器，是證偽工具\r\n市場每天產出大量解釋，缺的從來不是解釋，而是約束。框架的價值在於事前把問題縮小到可回答的範圍，並在事後能指出錯在哪一步。\r\n一個可用的判斷有三個特徵：有明確變數、有時間範圍、有否證條件。三者缺一，就退化成觀點。「看好 AI，因為需求很強」不是判斷，是氣氛；「未來兩季若雲廠資本開支增速低於兩成，電子權值 EPS 將下修，因此估值應壓縮」才是判斷。\r\n這也解釋了為什麼多數研究無法累積：判斷沒有否證條件，錯了也找不到錯在哪，只能再補一個新解釋。\r\n\r\n二、框架的四步\r\n第一步，定義問題。把「市場怎麼了」改成可回答的問句：哪個變數、哪類資產、多長時間。問題越具體，答案越可檢查；問題越宏大，答案越只能含糊。\r\n\r\n第二步，對齊口徑。定義、期間、幣別、修訂方式先統一。用不同尺規比較，結論一定失真；用不同期間比較，趨勢會憑空出現或消失。\r\n\r\n第三步，寫下否證條件。事前指定什麼數據出現就代表判斷錯了。這一步最常被跳過，卻恰恰是唯一能防止事後歸因的步驟。\r\n\r\n否證條件要寫到三個層次：數值（哪個指標到哪個水位）、時間（多久之內沒發生就算失效）、機制（若走的是另一條鏈，原判斷即作廢）。只寫數值，容易在期限內被雜訊反覆觸發；只寫時間，則等於沒有門檻。\r\n\r\n第四步，設定追蹤。把關鍵讀數放進固定頻率的檢查表，而不是每天重新想一遍。追蹤的頻率要和資料的頻率匹配。\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/dP6TeytHKM]\r\n\r\n四步之中，第一步決定上限，第三步決定下限。問題定義錯了，再嚴謹的推論也在錯的方向上；否證條件沒寫，框架就只剩裝飾功能。\r\n\r\n三、三種框架誤用\r\n指標堆疊。用二十個指標拼出一個結論，哪個支持就留下哪個。這種做法事後永遠說得通，事前卻無法執行，因為它沒有事先指定權重與優先順序。\r\n事後歸因。收盤後為漲跌找理由，把相關當因果。今天的下跌可以歸因於利率、油價、匯率或情緒，但若同一組原因在昨天對應的是上漲，這套歸因就沒有信息量。\r\n框架移植。把美股的估值框架套到流動性較薄的市場，或把總經框架套個股。市場結構、交易成本與資訊傳遞速度都不同，同一套參數會系統性偏誤。\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/BEWrMEHqqs]\r\n\r\n三者共同點，是把框架從「約束」變成「裝飾」：看起來很有結構，實際上不對任何結論負責。\r\n\r\n四、跑一遍：2022 年的例子\r\n2022 年初的市場共識是「升息只是暫時，獲利成長足以抵消」。把這句話寫成可證偽形式：若 10 年期美債殖利率突破 4%，且估值壓縮幅度超過 EPS 成長，則共識失效。[1]\r\n\r\n結果是：殖利率在 2022 年突破 4%，台股由 1 月的 18,619 點跌至 10 月的 12,619 點，跌幅約 32%，而企業獲利並未同步惡化——失效的是乘數假設，不是現金流假設。[2]\r\n\r\n這個例子的價值不在猜對，而在錯得清楚：可以只修正貼現率假設，保留現金流假設。框架的產出不是一個結論，而是一組可以分別修正的假設。\r\n\r\n反過來說，「升息就會跌」這種判斷即使事後成立，也沒有信息量——它無法區分跌幅來自估值還是獲利，下一次遇到同樣情境仍然只能重猜。可證偽的價值，在於把「猜對」轉換成「知道為什麼對」。\r\n\r\n五、框架的維護\r\n框架需要保養，否則會退化成事後解釋的藉口。可行的做法是三段檢視：每月檢查否證條件有沒有觸發；每季檢查假設是否還成立，例如訂單、資本開支與利率路徑；每年檢查框架本身——哪些判斷錯了，錯在口徑還是錯在鏈路。\r\n其中最有價值的是錯判紀錄。成功案例只能證明運氣，錯判紀錄才能指出框架的盲區。把每一次修正寫下來，框架才會累積；否則每年都從零開始，只是換一批指標。\r\n\r\n六、結論\r\n結論一：框架的目的是讓判斷可被證偽，不是增加解釋；沒有否證條件的判斷不構成判斷。\r\n結論二：四步之中最容易被跳過的是「寫下否證條件」，它也是唯一能約束事後歸因的一步。\r\n結論三：框架要維護，錯判紀錄比成功案例更有價值。\r\n\r\n\r\n數據來源\r\n[1] FRED，〈10-Year Treasury Constant Maturity Rate〉。https:\/\/fred.stlouisfed.org\r\n[2] ETtoday 財經雲，〈台股還會回落？富邦投顧董座估明年下半年上看萬六〉，2022 年 11 月 28 日。https:\/\/finance.ettoday.net\r\n[3] 臺灣證券交易所，〈2025 年報〉。https:\/\/wwwc.twse.com.tw\r\n\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "Farsight Capital Research Team | October 11, 2026\r\n\r\nHow can information be transformed into actionable, testable judgments rather than simply becoming another explanation?\r\n\r\n- A framework is not a forecasting tool: its purpose is to constrain one's own judgment, not generate more explanations.\r\n- A four-step process: define the question, align data definitions, specify falsification criteria, and establish a monitoring process.\r\n- Three common misuses: indicator stacking, post hoc attribution, and framework transplantation.\r\n- A one-sentence standard: If you cannot explain under what circumstances you would be wrong, you have not formed a proper investment judgment.\r\n\r\n1. A Framework Is a Tool for Falsification, Not Prediction\r\n\r\nMarkets generate an abundance of explanations every day. What is missing has never been more explanations, but rather constraints on how those explanations are formed.\r\n\r\nThe value of a framework lies in narrowing a question, in advance, to a scope that can be answered, and then identifying precisely where the reasoning failed afterward.\r\n\r\nA useful investment judgment has three characteristics: clearly defined variables, a specified time horizon, and explicit falsification criteria. If any one of these is missing, the judgment deteriorates into an opinion.\r\n\r\n“Bullish on AI because demand is strong” is not an investment judgment; it is a sentiment. By contrast, “If cloud providers' capital expenditure growth falls below 20% over the next two quarters, EPS estimates for major electronics constituents will be revised downward, putting pressure on valuations” is a testable judgment.\r\n\r\nThis also explains why much investment research fails to build on itself. Without falsification criteria, even an incorrect judgment cannot be traced to a specific point of failure. The only option is to add another explanation after the fact.\r\n\r\n2. The Four Steps of an Analytical Framework\r\n\r\nStep 1: Define the question.\r\n\r\nTurn “What is happening in the market?” into a question that can actually be answered: Which variable? Which asset class? Over what time horizon?\r\n\r\nThe more specific the question, the more testable the answer. The broader the question, the more likely the answer will remain vague.\r\n\r\nStep 2: Align data definitions.\r\n\r\nStandardize definitions, measurement periods, currencies, and revision methodologies before making comparisons.\r\n\r\nComparing data measured using different standards will inevitably distort conclusions. Comparing different time periods can make trends appear or disappear when they do not actually exist.\r\n\r\nStep 3: Specify falsification criteria.\r\n\r\nDetermine in advance which data outcomes would invalidate the original judgment.\r\n\r\nThis is the step most frequently skipped, yet it is the only one that can prevent post hoc attribution.\r\n\r\nFalsification criteria should be defined at three levels:\r\n\r\n- Numerical: Which indicator must reach what level?\r\n- Time-based: If the expected development does not occur within what period, should the judgment be considered invalid?\r\n- Mechanism-based: If the market moves through a different transmission channel, does the original judgment become invalid?\r\n\r\nSpecifying only numerical thresholds can lead to repeated false triggers caused by short-term noise within the designated period. Specifying only a time limit is equivalent to having no meaningful threshold at all.\r\n\r\nStep 4: Establish a monitoring process.\r\n\r\nPut key indicators into a checklist with a fixed review frequency, rather than reconsidering the entire analysis every day. The monitoring frequency should match the frequency at which the underlying data are released.\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/dP6TeytHKM]\r\n\r\nOf the four steps, the first determines the upper limit of the framework, while the third determines its lower limit.\r\n\r\nIf the question is defined incorrectly, even the most rigorous reasoning will proceed in the wrong direction. Without explicit falsification criteria, the framework becomes little more than decoration.\r\n\r\n3. Three Common Misuses of Analytical Frameworks\r\n\r\nIndicator stacking.\r\n\r\nUsing twenty indicators to construct a conclusion, then retaining whichever ones support the desired outcome. This approach can always be justified after the fact, but it cannot be executed consistently in advance because no weights or priorities were specified beforehand.\r\n\r\nPost hoc attribution.\r\n\r\nFinding reasons for market movements after the closing bell and mistaking correlation for causation. A market decline today can be attributed to interest rates, oil prices, exchange rates, or sentiment. But if the same set of factors was used to explain a market rally yesterday, the attribution provides no meaningful information.\r\n\r\nFramework transplantation.\r\n\r\nApplying a US equity valuation framework to a market with thinner liquidity, or applying a macroeconomic framework directly to an individual stock. Market structures, transaction costs, and the speed of information transmission differ. Using identical parameters across these environments creates systematic biases.\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/BEWrMEHqqs]\r\n\r\nWhat these three misuses have in common is that they turn a framework from a tool of constraint into a form of decoration. The analysis may appear highly structured, but in practice, it is not accountable for any conclusion.\r\n\r\n4. Applying the Framework: The 2022 Example\r\n\r\nAt the beginning of 2022, the prevailing market consensus was that “interest rate hikes would be temporary and earnings growth would be sufficient to offset their impact.”\r\n\r\nTo make this view falsifiable, it could have been stated as follows: If the 10-year US Treasury yield rises above 4%, and the extent of valuation multiple compression exceeds EPS growth, the prevailing consensus is invalidated. [1]\r\n\r\nWhat happened?\r\n\r\nThe 10-year Treasury yield moved above 4% in 2022. The TAIEX fell from 18,619 points in January to 12,619 points in October, a decline of approximately 32%, while corporate earnings did not deteriorate at the same pace. What failed was the valuation multiple assumption, not the cash flow assumption. [2]\r\n\r\nThe value of this example does not lie in having predicted the outcome correctly, but in being able to identify precisely what was wrong. The discount rate assumption could be revised while the cash flow assumption remained intact.\r\n\r\nThe output of a framework is therefore not a single conclusion, but a set of assumptions that can be reviewed and adjusted independently.\r\n\r\nConversely, a statement such as “rising interest rates will cause stocks to fall” provides little information, even if it proves correct afterward. It cannot distinguish whether the decline was driven by valuation compression or weaker earnings. When the same situation occurs again, the investor is left guessing all over again.\r\n\r\nThe value of falsifiability lies in turning “getting the prediction right” into “understanding why it was right.”\r\n\r\n5. Maintaining the Framework\r\n\r\nAn analytical framework requires regular maintenance. Otherwise, it can gradually deteriorate into an excuse for explaining events after they happen.\r\n\r\nA practical approach involves three review intervals:\r\n\r\nMonthly: Check whether any falsification criteria have been triggered.\r\n\r\nQuarterly: Review whether the underlying assumptions remain valid, including orders, capital expenditure, and the interest rate path.\r\n\r\nAnnually: Review the framework itself. Which judgments were wrong? Did the errors originate from inconsistent data definitions or from misidentifying the transmission channel?\r\n\r\nThe most valuable record is the log of incorrect judgments.\r\n\r\nSuccessful cases may simply demonstrate good fortune. Records of incorrect judgments, however, reveal the blind spots in a framework. Only by documenting every revision can the framework accumulate knowledge over time. Otherwise, each year begins from zero, with nothing more than a different set of indicators.\r\n\r\n6. Conclusion\r\n\r\nConclusion 1: The purpose of a framework is to make judgments falsifiable, not to generate more explanations. A judgment without falsification criteria does not constitute a proper analytical judgment.\r\n\r\nConclusion 2: The step most likely to be skipped is specifying falsification criteria. It is also the only step that meaningfully constrains post hoc attribution.\r\n\r\nConclusion 3: Frameworks require ongoing maintenance. Records of incorrect judgments are more valuable than records of successful cases.\r\n\r\nData Sources\r\n\r\n[1] FRED, “10-Year Treasury Constant Maturity Rate.”  \r\nhttps:\/\/fred.stlouisfed.org\r\n\r\n[2] ETtoday Finance Cloud, “Will Taiwan Stocks Fall Further? Fubon Securities Investment Services Chairman Sees the Index Reaching 16,000 in the Second Half of Next Year,” November 28, 2022.  \r\nhttps:\/\/finance.ettoday.net\r\n\r\n[3] Taiwan Stock Exchange, “2025 Annual Report.”  \r\nhttps:\/\/wwwc.twse.com.tw\r\n\r\nFarsight Capital Research Team | Deep Insight. Steady Progress."
            },
            "img": "image\/article-k2.png",
            "video": ""
        },
        {
            "id": "k3",
            "section": "knowledge",
            "cat": "風險管理",
            "homeSlot": 0,
            "tag": {
                "zh": "03",
                "en": "03"
            },
            "title": {
                "zh": "理解風險管理",
                "en": "Understanding Risk Management"
            },
            "desc": {
                "zh": "倉位、波動與風險承受能力之間有什麼關係？",
                "en": "How do positioning, volatility and risk tolerance interact?"
            },
            "body": {
                "zh": "涵遠資本研究部｜2026 年 10 月 11 日\r\n\r\n波動加劇時，風險管理究竟在管什麼？\r\n- 三層結構：部位層、組合層、尾部層，各有不同的參數。\r\n- 四個錯覺：分散等於安全、波動等於風險、停損等於風控、槓桿等於效率。\r\n- 順序：先定損失上限，再談報酬。\r\n- 壓力下：相關性趨近 1，流動性先消失。\r\n- 邊界：風控降低破產概率，不提高勝率。\r\n\r\n一、風控不是止損\r\n止損是執行細節，風控是事前的決定：錯的時候虧多少、能錯幾次、什麼情況下必須停手。同一筆交易，期望值相同，部位大小不同，結果可能是複利，也可能是出場。\r\n這個區別在市場高位時格外重要。台股在 10 月 6 日觸及 49,968 點，相對 2025 年底上漲約 72%。[1]同一套策略，在年初與現在面對的回撤承受度完全不同：漲幅越大，同樣的百分比回撤對應的金額越大。\r\n換句話說，風險管理的對象不是價格，而是自己的資金曲線。價格由市場決定，資金曲線由部位與紀律決定。\r\n\r\n二、風險的三層\r\n部位層管單筆：單筆風險上限、停損距離、部位大小。三者互相牽制——停損設得越遠，部位就必須越小，否則單筆風險會超過上限。\r\n組合層管整體：相關性、集中度、產業與因子暴露。十檔股票若同屬一個因子，本質上是一檔；分散在同一因子裡，等於沒有分散。\r\n尾部層管極端：流動性、槓桿、現金緩衝、跳空風險。極端情境下，能不能按自己的價格出場，比成本高低更重要。\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/rV01hFGnge]\r\n\r\n三層的關係是遞進的：部位層決定錯一次虧多少，組合層決定同時錯會怎樣，尾部層決定還能不能留在場內。多數人只做第一層，因為它最容易量化。\r\n三層衝突時，以尾部層為優先。若壓力測試顯示極端情境下無法出場，那麼部位層算得再精細也沒有意義；反過來，尾部有足夠緩衝，才有資格談部位的最佳化。\r\n\r\n三、四個常見錯覺\r\n分散等於安全。壓力下相關性趨近 1，平常看似無關的資產會同時下跌。2020 年 3 月，股票、信用與商品同步走弱，分散策略在最需要它的時候失效。\r\n\r\n波動等於風險。波動率只量價格擺動，不量流動性與跳空。2022 年流動性緊縮時，同樣的波動率對應更大的滑價；成交量薄的時候，波動率是低估的。\r\n\r\n停損等於風控。跳空與停牌會讓停損無法在設定價成交。極端日的開盤可能直接跳過停損價，成交在更差的位置；停損是計畫，不是保證。\r\n\r\n槓桿等於效率。槓桿放大的是回撤，也壓縮了等待的時間。保證金追繳會在最低點強制出場，把「看對但撐不到」變成實際虧損。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FyZgad8NbA5&scene=im&aid=582478&lang=zh]\r\n\r\n四個錯覺的共同點，是把平常時期的統計當成壓力時期的保證。風控要問的從來不是「平常會怎樣」，而是「最壞的時候我還能不能活著」。\r\n\r\n四、歷史檢驗\r\n2020 年，台股由 1 月的 12,197 點跌至 3 月 19 日盤中的 8,681 點，約兩個月跌掉 29%。[2]若部位滿倉且沒有現金緩衝，即使方向判斷正確，也未必能撐到後來的反轉。\r\n2022 年，台股由 18,619 點跌至 12,619 點，跌幅約 32%，且跌幅主要來自估值而非獲利。[3]若同時持有高估值成長股與高槓桿，兩者會在同一時間放大回撤。\r\n兩次的共同點不是「看錯方向」致命，而是錯的時候沒有退出機制：滿倉、無現金、無停損、有槓桿，四者任一都足以把小錯變成不可逆。\r\n\r\n五、可執行的參數\r\n參數不需要複雜，但必須寫下來。單筆風險上限（例如總資產 1%）；組合相關性上限（避免五檔持倉同屬一個因子）；現金緩衝（覆蓋六個月支出或兩成部位）；壓力測試（指數 -30%、匯率 -10%、利率 +200bp 同時發生時的損失）。\r\n\r\n換算方式很直接：十萬元本金、單筆風險 1%（即 1,000 元）、停損設在 8%，部位上限就是 1.25 萬元。停損放寬到 16%，部位必須減半；不想減部位，就只能縮小停損距離。\r\n\r\n壓力測試的價值在於把「感覺很危險」變成「具體虧多少」。若測試結果超出可承受範圍，要調整的是部位，而不是信心。\r\n寫下來的參數才是風控，記在腦子裡的只是決心。決心會在壓力下變形，參數不會。\r\n\r\n六、結論\r\n結論一：風控的第一問是「錯的時候虧多少」，而不是「會漲到哪裡」。\r\n結論二：三層之中，尾部層最常被忽略，因為它在平常時期幾乎沒有成本。\r\n結論三：參數化與壓力測試比感覺可靠；寫下來的紀律才能穿越壓力。\r\n\r\n參考來源\r\n[1] 台視新聞，〈台股 10 月 6 日盤中攻上 49,968.92 點〉，2026 年 10 月 6 日。https:\/\/news.ttv.com.tw\r\n[2] 臺灣證券交易所，〈2020 年報〉（3 月 19 日盤中低點 8,681.34 點）。https:\/\/wwwc.twse.com.tw\r\n[3] ETtoday 財經雲，〈台股還會回落？富邦投顧董座估明年下半年上看萬六〉，2022 年 11 月 28 日。https:\/\/finance.ettoday.net\r\n[4] 臺灣證券交易所，〈2025 年報〉。https:\/\/wwwc.twse.com.tw\r\n\r\n涵遠資本研究部｜深涵洞見，行穩致遠",
                "en": "Farsight Capital Research Team | October 11, 2026\r\n\r\nWhat Exactly Does Risk Management Control When Market Volatility Intensifies?\r\n\r\n- Three layers: position-level, portfolio-level, and tail-risk management, each governed by different parameters.\r\n- Four misconceptions: diversification equals safety, volatility equals risk, stop-losses equal risk control, and leverage equals efficiency.\r\n- The sequence: establish the maximum acceptable loss before considering returns.\r\n- Under stress: correlations converge toward 1, while liquidity disappears first.\r\n- The boundary: risk management reduces the probability of ruin; it does not increase the probability of winning.\r\n\r\n1. Risk Management Is Not the Same as Stop-Loss Execution\r\n\r\nA stop-loss is an execution detail. Risk management is a decision made in advance: how much can be lost when a trade goes wrong, how many times can a loss be sustained, and under what circumstances must trading stop?\r\n\r\nFor the same trade with the same expected value, different position sizes can produce very different outcomes: one may support compounding, while the other may force an exit from the market.\r\n\r\nThis distinction becomes particularly important when the market is trading at elevated levels. On October 6, the Taiwan stock market reached 49,968 points, up approximately 72% from the end of 2025.[1] The same strategy faces a very different drawdown tolerance at the beginning of the year compared with today. The greater the preceding gain, the larger the monetary loss represented by the same percentage drawdown.\r\n\r\nIn other words, risk management is not primarily concerned with price. It is concerned with the investor's own capital curve. The market determines prices; position sizing and discipline determine the trajectory of capital.\r\n\r\n2. The Three Layers of Risk\r\n\r\nThe position layer manages individual trades: the maximum risk per trade, stop-loss distance, and position size. These three variables constrain one another. The wider the stop-loss distance, the smaller the position must be; otherwise, the risk on a single trade will exceed the predefined limit.\r\n\r\nThe portfolio layer manages overall exposure: correlations, concentration, and sector and factor exposures. Ten stocks belonging to the same underlying factor are, in substance, one position. Diversifying within the same factor does not constitute meaningful diversification.\r\n\r\nThe tail-risk layer manages extreme scenarios: liquidity, leverage, cash buffers, and gap risk. Under extreme conditions, the ability to exit at an acceptable price matters more than the cost basis.\r\n\r\n[img:https:\/\/aka.doubaocdn.com\/s\/rV01hFGnge]\r\n\r\nThese three layers form a hierarchy. The position layer determines how much one mistake costs. The portfolio layer determines what happens when several positions are wrong at the same time. The tail-risk layer determines whether an investor can remain in the market at all. Most investors focus only on the first layer because it is the easiest to quantify.\r\n\r\nWhen the three layers conflict, tail-risk management takes priority. If stress testing shows that positions cannot be exited under extreme conditions, even the most precise position-level calculations become meaningless. Conversely, only with sufficient tail-risk buffers does it make sense to optimize position sizing.\r\n\r\n3. Four Common Misconceptions\r\n\r\nDiversification equals safety. Under stress, correlations can converge toward 1, causing assets that normally appear unrelated to decline simultaneously. In March 2020, equities, credit, and commodities weakened at the same time. Diversification strategies failed precisely when investors needed them most.\r\n\r\nVolatility equals risk. Volatility measures price fluctuations, but it does not capture liquidity conditions or gap risk. During the liquidity tightening of 2022, the same level of volatility could translate into greater slippage. When trading volume is thin, volatility can understate the actual risk.\r\n\r\nStop-losses equal risk control. Price gaps and trading suspensions can prevent stop-loss orders from being executed at their designated prices. On extreme trading days, the market may open beyond the stop-loss level, resulting in execution at a substantially worse price. A stop-loss is a plan, not a guarantee.\r\n\r\nLeverage equals efficiency. Leverage magnifies drawdowns and reduces the time available to wait for a position to recover. Margin calls can force liquidation near market lows, turning a position that was directionally correct but could not be maintained into a realized loss.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2FyZgad8NbA5&scene=im&aid=582478&lang=zh]\r\n\r\nThe common thread behind these four misconceptions is the assumption that statistics observed under normal conditions offer guarantees during periods of stress. Risk management should never be limited to asking, \"What usually happens?\" The more important question is, \"Can I survive when things are at their worst?\"\r\n\r\n4. Lessons from Historical Market Stress\r\n\r\nIn 2020, the Taiwan stock market fell from 12,197 points in January to an intraday low of 8,681 points on March 19, losing approximately 29% in about two months.[2] Investors who were fully invested without a cash buffer might not have been able to withstand the decline, even if their market direction assessment was ultimately correct.\r\n\r\nIn 2022, the Taiwan stock market declined from 18,619 points to 12,619 points, a drop of approximately 32%. The decline was driven primarily by valuation contraction rather than lower corporate earnings.[3] Investors holding high-valuation growth stocks alongside highly leveraged positions would have seen both exposures amplify drawdowns at the same time.\r\n\r\nThe common lesson from these two episodes is that being wrong about market direction was not necessarily what caused the greatest damage. The critical failure was having no exit mechanism when positions moved against expectations. Being fully invested, holding no cash, having no stop-loss discipline, and using leverage—any one of these could turn a manageable mistake into an irreversible loss.\r\n\r\n5. Parameters That Can Be Put into Practice\r\n\r\nRisk parameters do not need to be complicated, but they must be documented. Examples include:\r\n\r\n- Maximum risk per trade: for example, 1% of total assets.\r\n- Portfolio correlation limits: avoid having five holdings exposed to the same underlying factor.\r\n- Cash buffer: enough to cover six months of expenses or 20% of the portfolio.\r\n- Stress testing: calculate potential losses if the index falls 30%, the exchange rate moves 10% adversely, and interest rates rise by 200 basis points simultaneously.\r\n\r\nThe calculation is straightforward. With initial capital of NT$100,000, a 1% risk limit per trade (NT$1,000), and a stop-loss set at 8%, the maximum position size is NT$12,500. If the stop-loss distance widens to 16%, the position size must be cut in half. If the investor does not want to reduce the position, the only alternative is to narrow the stop-loss distance.\r\n\r\nThe value of stress testing lies in turning the vague feeling that \"the situation is dangerous\" into a concrete estimate of how much could be lost. If the results exceed the investor's tolerance, the variable to adjust is position size, not confidence.\r\n\r\nWritten parameters constitute risk control. Parameters kept only in one's head amount to little more than determination. Determination can change under pressure; predefined parameters provide a more consistent basis for action.\r\n\r\n6. Conclusion\r\n\r\nConclusion 1: The first question in risk management is, \"How much will I lose if I am wrong?\"—not, \"How high will the market go?\"\r\n\r\nConclusion 2: Of the three layers, tail-risk management is the most frequently overlooked because it carries almost no visible cost under normal market conditions.\r\n\r\nConclusion 3: Quantified parameters and stress testing are more reliable than intuition. Only discipline that has been documented can withstand periods of market stress.\r\n\r\nReferences\r\n\r\n[1] TTV News, \"Taiwan Stocks Reach an Intraday High of 49,968.92 Points on October 6,\" October 6, 2026. https:\/\/news.ttv.com.tw\r\n\r\n[2] Taiwan Stock Exchange, \"2020 Annual Report\" (intraday low of 8,681.34 points on March 19). https:\/\/wwwc.twse.com.tw\r\n\r\n[3] ETtoday Finance, \"Will Taiwan Stocks Pull Back Further? Fubon Securities Investment Consulting Chairman Estimates Market Could Reach 16,000 in the Second Half of Next Year,\" November 28, 2022. https:\/\/finance.ettoday.net\r\n\r\n[4] Taiwan Stock Exchange, \"2025 Annual Report.\" https:\/\/wwwc.twse.com.tw\r\n\r\nFarsight Capital Research Team | Deep Insight. Steady Progress."
            },
            "img": "image\/article-k3.png",
            "video": ""
        },
        {
            "id": "c1",
            "section": "column",
            "cat": "市場評論",
            "homeSlot": 0,
            "tag": {
                "zh": "慕遠專欄",
                "en": "Miles Chen's Column"
            },
            "title": {
                "zh": "陳慕遠：在波動中尋找秩序——機構研究的思考方式",
                "en": "Miles Chen: Finding Order Amid Volatility — The Institutional Approach to Market Research"
            },
            "desc": {
                "zh": "以機構研究的視角，拆解行情背後的資金、情緒與基本面邏輯。",
                "en": "Decoding the capital, sentiment and fundamentals behind market moves."
            },
            "body": {
                "zh": "涵遠資本 CEO 陳慕遠（Miles Chen）｜人物專欄｜2026 年 11 月 6 日\r\n\r\n一、市場的問題，三十年幾乎沒變\r\n我在 Houlihan Lokey 的近三十年，工作從研究分析、亞洲科技與工業研究，一路做到團隊管理與投資銀行、資本市場相關職務。換過職務，也換過市場，但投資人問的問題幾乎沒有變：為什麼跌、還會不會跌、什麼時候買。\r\n\r\n三個問題都很自然，但性質完全不同。第一個可以回答得比較誠實——把價格變化拆回它的來源；第二個是機率，只能給條件；第三個是紀律，只能談自己的資金曲線。多數討論之所以沒有累積，是因為把三者混在一起談：用一句「AI 泡沫」同時回答為什麼跌與什麼時候買，於是每一次波動都得重新編一套說法。\r\n\r\n本週正好是一個練習題。加權指數由 10 月 30 日的 45,211.83 點跌至 11 月 6 日的 39,701.64 點，五個交易日下跌 5,510.19 點、12.19%，是 2022 年 6 月以來最長的連續七個交易日收黑[1]；自 10 月 6 日的 49,822.55 點起算，回撤 20.31%，已符合技術性熊市定義。11 月 6 日盤中最大跌點 3,737.67 點創台股紀錄，跌停家數一度逾五百檔，終場收斂至 −4.50%[5]。\r\n\r\n同一根 K 線，至少有三種讀法：恐慌、超跌、評價修正。三種讀法用的是同一組數據，差別只在框架。機構研究的工作，就是把框架先立起來，再讓數據進來。\r\n\r\n二、秩序不是預測出來的，是拆出來的\r\n我習慣把任何一段行情拆成三條線：資金、情緒、基本面。三條線問的問題不同，因此不能互相替代，也不能用一條線的答案去回答另一條線的問題。\r\n\r\n資金線問「誰還在場」：賣方走了多少、買方接了什麼、槓桿清到哪裡。它讀的是存量與流速，不是動機。\r\n\r\n情緒線問「用什麼價格承接」：量能、寬度、價差。它量的是市場當下的溫度，不是明天的方向。\r\n\r\n基本面線問「盈利還撐不撐得住這個價格」：分子有沒有動，分母有沒有動。它回答的是值多少，而不是現在貴不貴。\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F2FUJvrZoRL&scene=im&aid=582478&lang=zh]\r\n\r\n三線的關係不是並列，而是遞進。基本面決定「值多少」，資金決定「現在誰在定價」，情緒決定「成交在哪個價位」。價格是三者交會處，不是任何一條線單獨的輸出。\r\n\r\n也因此，單線只能定位，不能定案。看到外資賣超就說「外資看空」，看到跌幅夠大就說「超跌」，都是把定位當結論——這在事後檢討時最容易發現，在當下最難察覺。\r\n\r\n三、資金線：問「還有多少籌碼沒走」\r\n資金線最常被誤讀成動機。外資賣超不等於看空，可能只是調節、換股，或追蹤指數的被動調整。資金線能回答的只有兩件事：還有多少部位在場內，以及離開的速度有多快。\r\n\r\n本週這條線給的答案是「還沒清完」，因為三個方向相反的力量同時存在。外資 10 月 29 日至 11 月 5 日連六日賣超 6,144.71 億元，11 月 3 日單日 1,073.06 億元[1]；10 月賣超 317.8 億美元、淨匯出 280 億美元，雙雙創單月新高[6]。同期投信連 18 個交易日買超 3,270.40 億元，內資透過 ETF 承接電子權值，籌碼由集中轉為分散；代價是反彈時推升效率下降，同一筆買盤能推高的點數變少。\r\n真正決定修正深度的是槓桿。11 月 5 日融資餘額 5,764.09 億元，距 10 月 12 日高點 6,533.30 億元只減 11.8%；年初為 3,434.10 億元，年內仍增加 67.9%[2]。融資減幅明顯小於指數跌幅，說明這輪出場的是現股，不是槓桿。\r\n\r\n所以資金線的結論不是「賣壓很重」，而是「賣壓還沒走完」。清洗完成的條件可以寫得很具體：融資餘額連續且加速下降、外資現貨賣超收斂、期貨淨空單回補，三者同時出現。任一項缺席，反彈都容易變成逃命波。\r\n\r\n四、情緒線：量溫度，不報方向\r\n情緒是每天被談論最多、卻最少被正確使用的一條線。情緒指標的功能是量「市場現在願意用什麼價格成交」，不是「明天會漲還是跌」。前者可以觀測，後者只是推測。\r\n\r\n本週三個讀數指向同一件事：賣壓是主動的，不是被迫的。量能上，11 月 3 日與 5 日成交值均接近 1 兆元，價格下跌但量沒有萎縮——這是投資人主動降低曝險，不是流動性抽離。寬度上，11 月 6 日跌停家數一度逾五百檔，賣壓由權值股外溢到流動性較差的中小型標的[7]。價差上，11 月 5 日美股費半重挫約 12.2%、台積電 ADR 下跌 10.8%，隔日現貨台積電 −4.69%、日月光投控 −7.86%、台達電 −6.63%、鴻海 −4.46%、聯發科 −2.02%[7]；ADR 跌幅約為現貨兩倍，價差尚未收斂。\r\n\r\n三個讀數合起來，情緒線說的是「承接價格下移、但承接的量還在」，而不是「恐慌到極點」。恐慌與否，要看寬度是否擴散、量能是否萎縮、價差是否收斂，不是看跌幅夠不夠大。\r\n\r\n情緒線最危險的誤用，是把跌深當便宜。跌幅只說明價格離高點多遠，不說明價格離價值多遠；後者要回到第三條線。\r\n\r\n五、基本面線：分子與分母，多數人只看分子\r\n基本面在機構研究裡常被簡化成「盈利好不好」，但價格是分子除以分母：分子是盈利，分母是折現率。多數爭論只談分子，於是在利率變動時完全失去座標。\r\n\r\n本週分子的讀數並不差。前 9 月累計經濟成長 14.22%，第 3 季單季 14.36%；9 月出口 872.2 億美元創單月新高[8]。這正是「基本面沒壞」與「股價該跌」可以同時成立的原因。\r\n\r\n分母則同時被三件事推高。利率上，10 年期美債殖利率升至 5.2%–5.3%，實質利率約 2.95%，為 2008 年以來最高[3]。信用上，AI 相關企業今年發債近 5,000 億美元，甲骨文 5 年期 CDS 約 263bp、CoreWeave 約 982bp[4]；算力不再只是資本支出科目，而是需要持續再融資的資產。政策上，資料中心審批權主要落在州政府，本次期中選舉翻轉的州長席位中，德州、喬治亞、俄亥俄為裝機重鎮，其中德州州長當選人主張暫停核准。受限的是時間表，不是訂單。\r\n\r\n還有一個本地變數：10 月 CPI 年增 2.33%，連續 6 個月突破 2%，壓抑貨幣寬鬆的空間[9]；11 月 5 日加權指數本益比 22.46 倍、股價淨值比 3.41 倍[2]，修正一週之後，評價仍在偏高區間。\r\n\r\n把這些放在一起，結論是：這一次跌的是評價中樞，不是盈利。這個區分之所以重要，是因為兩者的後續路徑完全不同——評價的修正可以隨利率與風險溢酬回穩而部分修復，盈利的下修不會自己回來。\r\n\r\n六、三線交會：把這一週的讀數放在一起\r\n圖 2 把三條線在本週的讀數並排。並排之後會發現，三條線給出的不是同一種答案，而是三種不同性質的答案。\r\n\r\n [img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F2FUJvrZoRL&scene=im&aid=582478&lang=zh]\r\n\r\n資金線說「還沒清完」，情緒線說「價格下移但量還在」，基本面線說「分子沒壞、分母變了」。三句話沒有矛盾，但組合起來只有一個結論：這是評價修正加上籌碼未清，不是景氣轉折。\r\n\r\n這個結論決定接下來看什麼。若是評價修正，要追蹤的是利率、信用利差與政策時間表；若是景氣轉折，要追蹤的是訂單與資本支出。兩者對應的指標完全不同，混用就會在下一次數據公布時再度誤判。台股本週的跌幅由 AI 供應鏈的評價基礎下移所帶動——台股不是獨立的風險源，而是全球 AI 交易的高 beta 代理，這也是跌幅為何集中在權值股與其衛星股。\r\n\r\n七、四個紀律\r\n方法之所以是方法，是因為它可以被重複執行，也可以被檢驗。以下四條，是我在三十年裡反覆校正出來的，寫下來的目的不是為了正確，而是為了在壓力下仍然做得到。\r\n\r\n [img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F4O9yCFXD9e&scene=im&aid=582478&lang=zh]\r\n\r\n四條之中，最難的是第三條。人在波動中最想找的是確認，不是反駁；但研究的價值，恰恰在於先替自己的結論找出可能被推翻的地方。第四條解決的是另一個常見問題：如果不留下當時的讀數與理由，事後檢討只會變成重新編故事，同樣的錯會再犯一次。\r\n\r\n八、秩序是可以被建立的工作方法\r\n回到最初的問題：一根長黑 K 出現時，機構研究在問什麼。\r\n一、波動是結果，不是原因。把價格變化拆成資金、情緒、基本面三條線，才有可檢驗的解釋。\r\n二、三條線的答案性質不同，不能互相替代。單線只能定位，交會才能定案。\r\n三、本週的讀數指向評價中樞下移與籌碼未清，而非盈利衰退。這個區分決定接下來該追蹤利率與信用，還是訂單與資本支出。\r\n\r\n本文由涵遠資本研究部整理，觀點以方法論為主，不針對個別標的。資料來源為公開資訊，僅供研究與教育參考，不構成任何投資建議、要約或招攬。投資決策應自行評估風險與自身財務狀況。\r\n\r\n涵遠資本——深涵洞見，行穩致遠",
                "en": "An institutional research perspective on the capital flows, market sentiment, and fundamental forces behind market movements.\r\n\r\nMiles Chen, CEO of Farsight Capital | Executive Profile | November 6, 2026\r\n\r\n1. The Questions Investors Ask Have Barely Changed in 30 Years\r\n\r\nDuring my nearly three decades at Houlihan Lokey, my responsibilities evolved from research and analysis, including coverage of the Asian technology and industrial sectors, to team management and roles related to investment banking and capital markets. My positions changed, and the markets I covered changed, but the questions investors asked remained largely the same: Why is the market falling? Will it fall further? When should I buy?\r\n\r\nAll three questions are natural, but they are fundamentally different in nature. The first can be answered with a reasonable degree of objectivity: break down price movements and identify their underlying drivers. The second is a matter of probability and can only be addressed through conditional scenarios. The third is a matter of discipline and can only be assessed against one's own capital curve.\r\n\r\nMany market discussions fail to build on previous analysis because these three questions are treated as though they were interchangeable. A single phrase such as \"AI bubble\" is used to explain both why the market is falling and when to buy. As a result, every bout of volatility requires an entirely new narrative.\r\n\r\nThis week provides a useful case study. The TAIEX fell from 45,211.83 points on October 30 to 39,701.64 points on November 6, declining 5,510.19 points, or 12.19%, over five trading sessions. It also recorded its longest streak of seven consecutive declining sessions since June 2022.[1] From the October 6 high of 49,822.55 points, the index had retraced 20.31%, meeting the technical definition of a bear market. On November 6, the intraday decline reached 3,737.67 points, a record for the Taiwan stock market. At one point, more than 500 stocks were limit-down, before the market recovered somewhat to close at a decline of 4.50%.[5]\r\n\r\nThe same candlestick can be interpreted in at least three ways: panic, oversold conditions, or valuation correction. All three interpretations use the same set of data. The difference lies in the analytical framework. The work of institutional research is to establish the framework first, then bring in the data.\r\n\r\n2. Order Is Not Predicted; It Is Revealed Through Decomposition\r\n\r\nI approach any market movement by breaking it down into three dimensions: capital flows, sentiment, and fundamentals. Each addresses a different question. They cannot replace one another, nor can an answer from one dimension be used to resolve a question belonging to another.\r\n\r\nThe capital-flow dimension asks, \"Who is still in the market?\" How much have sellers exited? What have buyers absorbed? How much leverage has been unwound? This dimension examines outstanding positions and the speed of capital movement, not investor motives.\r\n\r\nThe sentiment dimension asks, \"At what prices is the market willing to transact?\" It examines trading volume, market breadth, and price spreads. These indicators measure the market's current temperature, not its direction tomorrow.\r\n\r\nThe fundamental dimension asks, \"Can earnings continue to support the current price?\" Has the numerator changed? Has the denominator changed? It addresses what an asset is worth, not whether it is currently expensive or cheap.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F2FUJvrZoRL&scene=im&aid=582478&lang=zh]\r\n\r\nThe relationship among these three dimensions is hierarchical rather than parallel. Fundamentals determine what an asset is worth. Capital flows determine who is setting prices at a given moment. Sentiment determines the prices at which transactions take place. Price emerges where all three intersect; it is not the output of any single dimension.\r\n\r\nThat is why any one dimension can help establish context, but cannot deliver a definitive conclusion on its own. Seeing foreign investors as net sellers and immediately concluding that \"foreign investors are bearish,\" or seeing a sufficiently large decline and declaring the market \"oversold,\" both amount to mistaking an observation for a conclusion. Such errors are easiest to identify in hindsight and hardest to recognize in real time.\r\n\r\n3. Capital Flows: Asking How Much Positioning Remains in the Market\r\n\r\nCapital flows are most often misinterpreted as a reflection of investor motives. Foreign investors' net selling does not necessarily mean they are bearish. It may reflect portfolio rebalancing, rotation into other stocks, or passive adjustments to index-tracking positions. The capital-flow dimension can answer only two questions: how much positioning remains in the market, and how quickly it is being unwound.\r\n\r\nThis week's data point to one conclusion: the unwinding process is not yet complete. Three opposing forces are operating simultaneously.\r\n\r\nForeign investors recorded six consecutive sessions of net selling from October 29 to November 5, totaling NT$614.471 billion. On November 3 alone, net selling reached NT$107.306 billion.[1] In October, foreign investors recorded net selling of US$31.78 billion and net capital outflows of US$28 billion, both monthly records.[6]\r\n\r\nOver the same period, domestic investment trusts recorded 18 consecutive trading sessions of net buying, totaling NT$327.040 billion. Domestic capital absorbed large-cap technology stocks through ETFs, shifting market holdings from concentrated positions toward a broader distribution. The trade-off is reduced efficiency in driving a rebound: the same amount of buying can now move the index by fewer points.\r\n\r\nLeverage is the factor that ultimately determines the depth of a correction. On November 5, margin financing balances stood at NT$576.409 billion, only 11.8% below the October 12 peak of NT$653.330 billion. At the beginning of the year, the balance was NT$343.410 billion, meaning it remained 67.9% higher year to date.[2] The decline in margin financing has been significantly smaller than the fall in the index, suggesting that this round of selling has been driven more by cash-equity positions than by leveraged liquidation.\r\n\r\nThe conclusion from the capital-flow dimension, therefore, is not simply that \"selling pressure is heavy,\" but that \"the selling pressure has not yet run its course.\"\r\n\r\nThe conditions for a more complete unwinding can be defined quite specifically: a sustained and accelerating decline in margin financing balances, a moderation in foreign investors' net selling of cash equities, and covering of net short futures positions. All three need to occur together. If any one is absent, a rebound may easily turn into an opportunity for investors to exit rather than a sustainable recovery.\r\n\r\n4. Sentiment: Measuring Market Temperature, Not Predicting Direction\r\n\r\nSentiment is the dimension discussed most frequently every day, yet it is also one of the least accurately applied. The purpose of sentiment indicators is to measure \"the prices at which the market is currently willing to transact,\" not to predict \"whether the market will rise or fall tomorrow.\" The former can be observed; the latter remains speculation.\r\n\r\nThree readings this week point to the same conclusion: selling pressure has been active rather than forced.\r\n\r\nIn terms of trading volume, turnover value on November 3 and November 5 was close to NT$1 trillion on both days. Prices fell, but trading volume did not contract. This suggests that investors were actively reducing exposure rather than liquidity simply disappearing.\r\n\r\nIn terms of market breadth, more than 500 stocks were limit-down at one point on November 6, as selling pressure spread beyond large-cap stocks to smaller companies with weaker liquidity.[7]\r\n\r\nIn terms of price spreads, the Philadelphia Semiconductor Index fell approximately 12.2% on November 5, while TSMC's ADR declined 10.8%. On the following day, the corresponding declines in Taiwan's cash market were 4.69% for TSMC, 7.86% for ASE Technology Holding, 6.63% for Delta Electronics, 4.46% for Hon Hai, and 2.02% for MediaTek.[7] The ADR decline was approximately twice that of the corresponding cash-market move, indicating that the price spread had not yet converged.\r\n\r\nTaken together, these three readings suggest that \"the market is accepting lower prices, but buying volume remains present.\" They do not necessarily indicate that panic has reached an extreme.\r\n\r\nWhether the market is truly experiencing panic should be assessed by examining whether market breadth continues to deteriorate, whether trading volume contracts, and whether price spreads converge—not simply by asking whether the decline has been large enough.\r\n\r\nThe most dangerous misuse of sentiment indicators is treating a sharp decline as evidence of cheap valuations. The size of a decline tells us how far the price has moved from its high; it does not tell us how far the price is from intrinsic value. That question belongs to the third dimension.\r\n\r\n5. Fundamentals: The Numerator and Denominator — Most Investors Only Watch the Numerator\r\n\r\nIn institutional research, fundamentals are often reduced to the question of whether earnings are good or bad. But price is a function of the numerator divided by the denominator: the numerator represents earnings, while the denominator reflects the discount rate. Most debates focus only on the numerator, leaving investors without a proper frame of reference when interest rates change.\r\n\r\nThe numerator's readings this week were not weak. Cumulative economic growth for the first nine months reached 14.22%, while third-quarter growth alone was 14.36%. September exports totaled US$87.22 billion, setting a new monthly record.[8] This is precisely why \"fundamentals have not deteriorated\" and \"stock prices should fall\" can both be true at the same time.\r\n\r\nThe denominator, meanwhile, is being pushed higher by three factors.\r\n\r\nOn interest rates, the 10-year US Treasury yield rose to 5.2%–5.3%, while the real interest rate was approximately 2.95%, its highest level since 2008.[3]\r\n\r\nOn credit, AI-related companies have issued nearly US$500 billion in bonds this year. Oracle's five-year credit default swap (CDS) spread was approximately 263 basis points, while CoreWeave's was around 982 basis points.[4] Computing capacity is no longer merely a capital expenditure item; it has become an asset base that requires ongoing refinancing.\r\n\r\nOn policy, data center approval authority rests primarily with state governments. Among the gubernatorial seats that changed hands in this midterm election, Texas, Georgia, and Ohio are major data center markets. The incoming governor of Texas has advocated suspending approvals. What is constrained is the project timeline, not the order book.\r\n\r\nThere is also a domestic variable. Taiwan's October CPI rose 2.33% year over year, exceeding 2% for six consecutive months and limiting the scope for monetary easing.[9] As of November 5, the TAIEX was trading at a price-to-earnings ratio of 22.46 times and a price-to-book ratio of 3.41 times.[2] Even after a week of correction, valuations remained relatively elevated.\r\n\r\nTaken together, the conclusion is that this correction reflects a lower valuation benchmark, not a deterioration in earnings.\r\n\r\nThis distinction matters because the subsequent paths are fundamentally different. Valuations can partially recover as interest rates and risk premiums stabilize. Earnings downgrades, by contrast, do not reverse on their own.\r\n\r\n6. Where the Three Dimensions Converge: Putting This Week's Readings Together\r\n\r\nFigure 2 places this week's readings across the three dimensions side by side. Once viewed together, it becomes clear that the three dimensions do not provide the same answer. Each offers an answer of a different nature.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F2FUJvrZoRL&scene=im&aid=582478&lang=zh]\r\n\r\nThe capital-flow dimension says, \"The unwinding is not yet complete.\" The sentiment dimension says, \"Prices are moving lower, but trading volume remains present.\" The fundamental dimension says, \"The numerator has held up, but the denominator has changed.\"\r\n\r\nThese statements are not contradictory. Taken together, they lead to one conclusion: this is a valuation correction accompanied by an incomplete unwinding of market positions, rather than an economic cycle turning point.\r\n\r\nThat conclusion determines what should be monitored next. If this is a valuation correction, the focus should be on interest rates, credit spreads, and policy timelines. If it is an economic turning point, the focus should be on orders and capital expenditure. The relevant indicators are entirely different. Confusing the two will lead to another misinterpretation when the next set of data is released.\r\n\r\nThis week's decline in the Taiwan stock market was driven by a downward repricing of the valuation basis of the AI supply chain. Taiwan equities are not an independent source of risk; they are a high-beta proxy for the global AI trade. This also explains why the decline was concentrated in large-cap stocks and their related satellite stocks.\r\n\r\n7. Four Principles of Discipline\r\n\r\nA method is meaningful because it can be repeated and tested. The following four principles are the result of repeated refinement over three decades. Their purpose is not simply to be right, but to make it possible to act consistently even under pressure.\r\n\r\n[img:https:\/\/link.wtturl.cn\/?target=https%3A%2F%2Faka.doubaocdn.com%2Fs%2F4O9yCFXD9e&scene=im&aid=582478&lang=zh]\r\n\r\nOf the four principles, the third is the most difficult to follow. In periods of volatility, people instinctively seek confirmation rather than contradiction. Yet the value of research lies precisely in identifying, before anyone else does, where one's own conclusions could be proven wrong.\r\n\r\nThe fourth principle addresses another common problem. Without preserving the readings and reasoning available at the time, post-market reviews become exercises in rewriting the story. The same mistakes are then repeated.\r\n\r\n## 8. Order Is a Working Method That Can Be Established\r\n\r\nReturning to the original question: when a long bearish candlestick appears, what is institutional research actually asking?\r\n\r\n1. Volatility is an outcome, not a cause. Only by decomposing price movements into capital flows, sentiment, and fundamentals can we develop an explanation that can be tested.\r\n\r\n2. The three dimensions produce different types of answers and cannot replace one another. Any single dimension can establish context; only their convergence can support a definitive conclusion.\r\n\r\n3. This week's readings point to a lower valuation benchmark and an incomplete unwinding of market positions, rather than an earnings contraction. This distinction determines whether the next areas of focus should be interest rates and credit conditions, or orders and capital expenditure.\r\n\r\nThis article was prepared by the Farsight Capital Research Team. Its views focus primarily on analytical methodology and do not address any individual securities. The information is derived from publicly available sources and is provided for research and educational purposes only. It does not constitute investment advice, an offer, or a solicitation. Investment decisions should be made after independently assessing the relevant risks and one's own financial circumstances.\r\n\r\nFarsight Capital | Deep Insight. Steady Progress."
            },
            "img": "image\/c384d3f6a761a371c5231948cb3b88f3f345ce9f.png",
            "video": "video\/hy-column-demo.mp4"
        },
        {
            "id": "c2",
            "section": "column",
            "cat": "市場評論",
            "homeSlot": 0,
            "tag": {
                "zh": "慕遠專欄",
                "en": "Miles Chen's Column"
            },
            "title": {
                "zh": "資金、情緒、基本面：三條線如何互相驗證",
                "en": "Capital Flows, Sentiment, Fundamentals: How the Three Lines Validate Each Other"
            },
            "desc": {
                "zh": "以近期盤面為例，說明三條線交會時該如何判讀。",
                "en": "Using recent market action to show how the three lines should be read together."
            },
            "body": {
                "zh": "本文為慕遠專欄佔位內容，文章正文待更新。\r\n\r\n（此處由後台「專欄」分類編輯，發佈後即取代本段佔位文字。）",
                "en": "This is a placeholder for Miles Chen's Column. The full article will be updated.\r\n\r\n(Edit this entry under the Column section in the admin panel.)"
            },
            "img": "",
            "video": ""
        },
        {
            "id": "c3",
            "section": "column",
            "cat": "市場評論",
            "homeSlot": 0,
            "tag": {
                "zh": "慕遠專欄",
                "en": "Miles Chen's Column"
            },
            "title": {
                "zh": "評價修正與景氣轉折：該追蹤哪些指標",
                "en": "Valuation Correction vs. Cycle Turn: Which Indicators to Track"
            },
            "desc": {
                "zh": "從利率、信用利差到訂單與資本支出，建立可檢驗的追蹤清單。",
                "en": "From rates and credit spreads to orders and capex: a testable checklist."
            },
            "body": {
                "zh": "本文為慕遠專欄佔位內容，文章正文待更新。\r\n\r\n（此處由後台「專欄」分類編輯，發佈後即取代本段佔位文字。）",
                "en": "This is a placeholder for Miles Chen's Column. The full article will be updated.\r\n\r\n(Edit this entry under the Column section in the admin panel.)"
            },
            "img": "",
            "video": ""
        },
        {
            "id": "n1",
            "section": "news",
            "cat": "公司進展",
            "homeSlot": 0,
            "tag": {
                "zh": "公司進展",
                "en": "Company Progress"
            },
            "title": {
                "zh": "涵遠資本正式成立，確立「以研究為核」的服務定位",
                "en": "Farsight Capital founded with research at the core"
            },
            "desc": {
                "zh": "以機構級研究方法服務個人與家族客戶，從第一天起就把研究放在中心。",
                "en": "Bringing institutional-grade research discipline to private and family clients."
            },
            "body": {
                "zh": "涵遠資本（Farsight Capital）正式成立。\n\n我們的定位從一開始就非常清晰：把機構級別的研究紀律帶給個人與家族客戶。市場上不缺觀點，缺的是有紀律、可驗證、長期一致的研究方法。\n\n[img:image\/0ab2c4d92c9997400daf26c86c18fff04f38f959.png]\n\n成立初期我們確立了三件事：\n一、每週對外發布市場觀察，無論行情冷熱；\n二、所有客戶建議必須有研究依據，拒絕「拍腦袋」；\n三、研究與銷售嚴格分離，利益不衝突。",
                "en": "Farsight Capital is officially established.\n\nOur positioning has been clear from day one: bring institutional-grade research discipline to private and family clients. The market is not short of opinions — what is missing is a disciplined, verifiable and consistent research method.\n\n[img:image\/0ab2c4d92c9997400daf26c86c18fff04f38f959.png]\n\nAt launch we set three principles:\n1. Publish a market watch every week, in good markets and bad;\n2. Every client recommendation must rest on research — no guesswork;\n3. Research and sales stay strictly separate, so there is no conflict of interest."
            },
            "img": "image\/66c037a004f682ba909793e7a49491ee561504c7.png"
        },
        {
            "id": "n2",
            "section": "news",
            "cat": "公司進展",
            "homeSlot": 0,
            "tag": {
                "zh": "公司進展",
                "en": "Company Progress"
            },
            "title": {
                "zh": "研究團隊擴編，建立「每週市場觀察」機制",
                "en": "Research team expands; weekly market watch mechanism established"
            },
            "desc": {
                "zh": "研究流程標準化：週報、月度複盤、季度策略會三層節奏。",
                "en": "A standardized three-layer research rhythm: weekly, monthly, quarterly."
            },
            "body": {
                "zh": "研究團隊完成擴編，覆蓋總體經濟、產業與籌碼分析三個方向。\n\n與此同時，我們把研究流程標準化為三層節奏：每週市場觀察跟蹤資金與情緒；月度複盤檢驗判斷得失；季度策略會重新校準配置框架。\n\n固定的節奏比正確的預測更重要——它讓錯誤能被及時發現，讓正確能被穩定重複。",
                "en": "The research team has completed its expansion, now covering macro, industry and positioning analysis.\n\nAt the same time, we standardized the research process into a three-layer rhythm: a weekly market watch that tracks flows and sentiment; a monthly review that tests our calls; and a quarterly strategy meeting that recalibrates the allocation framework.\n\nA fixed rhythm matters more than a correct forecast — it lets mistakes surface in time and lets what works be repeated consistently."
            },
            "img": "image\/4ace8a8496eb65f8b54b1cf49b658443503bec48.png"
        },
        {
            "id": "n3",
            "section": "news",
            "cat": "合作動態",
            "homeSlot": 0,
            "tag": {
                "zh": "合作動態",
                "en": "Partnership Updates"
            },
            "title": {
                "zh": "與本地券商簽署研究數據合作備忘錄",
                "en": "Research data MOU signed with local brokers"
            },
            "desc": {
                "zh": "籌碼與資金流向數據接入研究體系，觀察顆粒度進一步提升。",
                "en": "Positioning and flow data integrated into the research stack."
            },
            "body": {
                "zh": "我們與本地主要券商簽署研究數據合作備忘錄。\n\n合作完成後，外資買賣超、期貨持倉、融資餘額等籌碼數據將以更高頻率接入研究體系，「每週市場觀察」的觀察顆粒度進一步提升。\n\n數據本身不是觀點，但更好的數據能約束觀點的隨意性。",
                "en": "We signed a research data memorandum of understanding with a leading local brokerage.\n\nOnce live, positioning data — foreign net buying and selling, futures positions, margin balances — will feed into our research stack at higher frequency, sharpening the granularity of the weekly market watch.\n\nData is not an opinion, but better data disciplines how freely opinions are formed."
            },
            "img": "image\/7b2ae5a80166a4162dd4ee7c3d99470f51a36a95.png"
        },
        {
            "id": "n4",
            "section": "news",
            "cat": "合作動態",
            "homeSlot": 0,
            "tag": {
                "zh": "合作動態",
                "en": "Partnership Updates"
            },
            "title": {
                "zh": "財富管理服務體系上線，覆蓋顧問諮詢全流程",
                "en": "Wealth management service system launched"
            },
            "desc": {
                "zh": "從需求分析、方案設計到定期檢視，服務流程全部標準化。",
                "en": "Standardized end-to-end advisory: analysis, design, review."
            },
            "body": {
                "zh": "財富管理服務體系正式上線。\n\n服務覆蓋三個環節：需求分析（目標、期限與風險承受度）、方案設計（資產配置與標的選擇）、定期檢視（每季複核與再平衡）。每個環節都有標準交付物，客戶清楚地知道「我們做了什麼、為什麼這樣做」。\n\n透明的流程是信任的基礎。",
                "en": "Our wealth management service system is now live.\n\nThe service spans three stages: needs analysis (goals, horizon and risk tolerance), solution design (asset allocation and instrument selection), and periodic review (quarterly recheck and rebalancing). Each stage has a standard deliverable, so clients always know what we did and why.\n\nA transparent process is the foundation of trust."
            },
            "img": "image\/cfd45d1d75c024f860ad577b2add6a4625eb0d3d.png"
        },
        {
            "id": "n5",
            "section": "news",
            "cat": "線下沙龍",
            "homeSlot": 0,
            "tag": {
                "zh": "線下沙龍",
                "en": "Offline Salons"
            },
            "title": {
                "zh": "首場投資人閉門沙龍：波動市場中的資產配置",
                "en": "First closed-door salon: asset allocation in volatile markets"
            },
            "desc": {
                "zh": "與三十餘位客戶面對面，討論波動加劇環境下的配置思路。",
                "en": "Face-to-face with 30+ clients on allocation under volatility."
            },
            "body": {
                "zh": "首場投資人閉門沙龍在台北舉行，三十餘位客戶到場。\n\n本次主題是「波動市場中的資產配置」：與其預測波動何時結束，不如讓組合先適應波動——透過倉位結構、現金流資產與再平衡紀律，把波動從威脅變成收益的來源之一。\n\n現場問答環節的精華內容，將整理後於專欄發布。",
                "en": "Our first closed-door investor salon was held in Taipei, with more than 30 clients attending.\n\nThe theme was asset allocation in volatile markets: rather than predicting when volatility ends, let the portfolio adapt to it first — through position structure, cash-flow assets and rebalancing discipline, turning volatility from a threat into one of the sources of return.\n\nHighlights from the Q&A will be published as a column piece."
            },
            "img": "image\/d892930c528d8fdefb062e56cf94c19dc7f5c84a.png"
        },
        {
            "id": "n6",
            "section": "news",
            "cat": "線下沙龍",
            "homeSlot": 0,
            "tag": {
                "zh": "線下沙龍",
                "en": "Offline Salons"
            },
            "title": {
                "zh": "分享會回顧：散戶如何閱讀籌碼數據",
                "en": "Salon recap: how retail investors should read positioning data"
            },
            "desc": {
                "zh": "三個最常被誤讀的籌碼指標，與正確的打開方式。",
                "en": "Three commonly misread indicators, and how to read them right."
            },
            "body": {
                "zh": "本期分享會主題為「散戶如何閱讀籌碼數據」。\n\n我們討論了三個最常被誤讀的指標：\n一、外資買賣超——看「連續性」而非單日數字；\n二、融資餘額——它是情緒溫度計，不是方向指示燈；\n三、期貨淨部位——留意結算前的部位調整噪音。\n\n投影片與精華筆記已整理，可向您的顧問索取。",
                "en": "This session covered how retail investors should read positioning data.\n\nWe discussed the three most commonly misread indicators:\n1. Foreign net buying and selling — look at persistence, not a single day's number;\n2. Margin balances — a sentiment thermometer, not a direction indicator;\n3. Net futures positions — beware the noise of position adjustments before settlement.\n\nSlides and summary notes are available from your advisor."
            },
            "img": "image\/fc41161276bb5bd0b3cd44d6688643b1acfc1e3e.png"
        },
        {
            "id": "n7",
            "section": "news",
            "cat": "行業論壇",
            "homeSlot": 0,
            "tag": {
                "zh": "行業論壇",
                "en": "Industry Forums"
            },
            "title": {
                "zh": "參與資產管理行業論壇，分享機構研究方法",
                "en": "Industry forum: how research should drive decisions"
            },
            "desc": {
                "zh": "研究主管受邀出席圓桌討論：研究如何真正影響投資決策。",
                "en": "Research head joins the roundtable on research-to-decision process."
            },
            "body": {
                "zh": "涵遠資本研究主管受邀參與年度資產管理行業論壇圓桌討論。\n\n本次分享的主題是「研究如何真正影響投資決策」：好的研究輸出不是預測報告，而是決策規則——在什麼條件下加倉、什麼條件下減倉、什麼條件下承認錯誤。\n\n與會同業的交流摘要將於近期整理發布。",
                "en": "Farsight Capital's head of research was invited to the roundtable at the annual asset management industry forum.\n\nThe topic was how research should truly drive investment decisions: good research output is not a forecast report but a decision rule — under what conditions to add, when to cut, and when to admit a mistake.\n\nA summary of the exchange with peers will be published shortly."
            },
            "img": "image\/b7eaa02ce69824b2c908af7b50fde05cad273815.png"
        },
        {
            "id": "n8",
            "section": "news",
            "cat": "行業論壇",
            "homeSlot": 0,
            "tag": {
                "zh": "行業論壇",
                "en": "Industry Forums"
            },
            "title": {
                "zh": "論壇觀點摘要：市場結構的幾個中長期變化",
                "en": "Forum takeaways: structural shifts to watch"
            },
            "desc": {
                "zh": "被動資金擴張、散戶機構化、數據平民化——三條主線的影響。",
                "en": "Passive growth, retail institutionalization, data democratization."
            },
            "body": {
                "zh": "論壇觀點摘要：市場結構的幾個中長期變化。\n\n一、被動資金持續擴張：指數成分調整帶來的資金流動，正在成為短線行情的重要變數；\n二、散戶機構化：資訊劣勢縮小，但紀律劣勢仍在；\n三、數據平民化：籌碼與情緒數據門檻降低，關鍵變成「怎麼用」而不是「有沒有」。\n\n我們會把這三條主線納入季度策略會的固定檢視清單。",
                "en": "Forum takeaways: several medium- to long-term shifts in market structure.\n\n1. Passive capital keeps expanding: flows triggered by index rebalancing are becoming a major variable for short-term moves;\n2. Retail institutionalization: the information gap is narrowing, but the discipline gap remains;\n3. Data democratization: positioning and sentiment data are easier to obtain, so the question is how to use them, not whether they exist.\n\nWe will add these three threads to the standing checklist for our quarterly strategy meeting."
            },
            "img": "image\/7eb764870abc13af52bc2d35fe0ff943726c1ca9.png"
        }
    ],
    "collections": [
        {
            "id": "col1",
            "section": "research",
            "cat": "",
            "title": {
                "zh": "涵遠精選",
                "en": "Farsight Picks"
            },
            "desc": {
                "zh": "研究、專欄與企業動態的精華選讀，帶你快速理解涵遠資本的研究脈絡。",
                "en": "A curated read across research, column and company updates."
            },
            "cover": "image\/c384d3f6a761a371c5231948cb3b88f3f345ce9f.png",
            "articles": [
                "c1",
                "r1",
                "k2",
                "n5"
            ],
            "order": 1
        },
        {
            "id": "col2",
            "section": "research",
            "cat": "",
            "title": {
                "zh": "市場研究入門",
                "en": "Market Research Primer"
            },
            "desc": {
                "zh": "從市場觀察、籌碼到風險管理，建立完整的市場研究框架。",
                "en": "From market watch and positioning to risk control — a research framework."
            },
            "cover": "image\/7dc88d6f36a6f19748542866282610275f702c2b.png",
            "articles": [
                "r1",
                "r2",
                "r3",
                "r4"
            ],
            "order": 2
        },
        {
            "id": "col3",
            "section": "news",
            "cat": "公司進展",
            "title": {
                "zh": "公司進展",
                "en": "Company Progress"
            },
            "desc": {
                "zh": "涵遠資本的重要發展與里程碑。",
                "en": "Key developments and milestones of Farsight Capital."
            },
            "cover": "image\/66c037a004f682ba909793e7a49491ee561504c7.png",
            "articles": [],
            "order": 3
        },
        {
            "id": "col4",
            "section": "news",
            "cat": "合作動態",
            "title": {
                "zh": "合作動態",
                "en": "Partnership Updates"
            },
            "desc": {
                "zh": "合作夥伴與服務體系進展。",
                "en": "Partners and service network progress."
            },
            "cover": "image\/7b2ae5a80166a4162dd4ee7c3d99470f51a36a95.png",
            "articles": [],
            "order": 4
        },
        {
            "id": "col5",
            "section": "news",
            "cat": "線下沙龍",
            "title": {
                "zh": "線下沙龍",
                "en": "Offline Salons"
            },
            "desc": {
                "zh": "投資人沙龍與分享會回顧。",
                "en": "Investor salons and sharing sessions."
            },
            "cover": "image\/d892930c528d8fdefb062e56cf94c19dc7f5c84a.png",
            "articles": [],
            "order": 5
        },
        {
            "id": "col6",
            "section": "news",
            "cat": "行業論壇",
            "title": {
                "zh": "行業論壇",
                "en": "Industry Forums"
            },
            "desc": {
                "zh": "行業論壇參與與觀點摘要。",
                "en": "Forum participation and takeaway views."
            },
            "cover": "image\/b7eaa02ce69824b2c908af7b50fde05cad273815.png",
            "articles": [],
            "order": 6
        }
    ]
}