{"id":542,"date":"2026-04-15T11:36:16","date_gmt":"2026-04-15T11:36:16","guid":{"rendered":"https:\/\/www.tejwin.com\/en\/insights\/how-information-sources-shift-stock-prices-empirical-evidence-from-tcri-wd\/"},"modified":"2026-09-23T18:19:27","modified_gmt":"2026-09-23T18:19:27","slug":"how-information-sources-shift-stock-prices-empirical-evidence-from-tcri-wd","status":"publish","type":"insight","link":"https:\/\/www.tejwin.com\/en\/insights\/how-information-sources-shift-stock-prices-empirical-evidence-from-tcri-wd\/","title":{"rendered":"How Information Sources Shift Stock Prices: Empirical Evidence from TCRI Watchdog"},"content":{"rendered":"
\"\"<\/figure>\n

Preface<\/span><\/h2>\n

TCRI Watchdog (WD)<\/a> converts complex news and announcements into standardized quantitative alternative data. Building on our research into \u201cOfficial Announcements (Source P)<\/a>\u201d and \u201cMedia News (Source N)<\/a>,\u201d we have confirmed that disclosure channels directly dictate the speed and structure of market digestion.<\/p>\n

This chapter moves from macro \u201cEvent Categories\u201d to micro \u201cSource \u00d7 Sub-category\u201d dimensions<\/mark> to capture actionable Alpha within granular events. Focusing on high-sensitivity \u201cCorporate Control Events,\u201d we analyze the signal heterogeneity between Source P and Source N. We further demonstrate how these high-precision signals assist investors in optimizing entry timing and hedging strategies.<\/p>\n

\n
\ud83d\udc49Struggling with unquantifiable market noise? TCRI WD turns raw text into a structured quantitative database<\/a><\/div>\n<\/div>\n

Backtesting Framework\u00a0<\/span><\/h2>\n

This study adopts a standard event study methodology to evaluate abnormal returns (AR) and cumulative abnormal returns (CAAR) surrounding event occurrences.\u00a0<\/p>\n

Research Scope and Event Design\u00a0\u00a0\u00a0<\/span><\/h3>\n
    \n
  • Market coverage<\/strong>: Taiwan-listed and OTC companies (including delisted firms)\u00a0\u00a0<\/li>\n
  • Data Source<\/strong>: TCRI Watchdog (WD),\u00a0focusing on both announcement-based (P) and news-based (N) events\u00a0<\/li>\n
  • Sample period<\/strong>: January 1, 2018 to August 31, 2025\u00a0<\/li>\n
  • Research Scope (Source \u00d7 Subcategory)<\/strong>\uff1aThis chapter focuses on 2 representative event groups selected based on CAAR performance and statistical significance, covering both positive and negative signals:\u00a0\n
      \n
    • P-MO: Corporate governance announcements \u2014 control \/ agency issues\u00a0\u00a0<\/li>\n
    • N-MO: News \u2014 control \/ agency issues\u00a0\u00a0<\/li>\n<\/ul>\n<\/li>\n
    • Event grouping<\/strong>:\u00a0Based on the WD event scoring system, events are grouped as follows:\u00a0<\/li>\n<\/ul>\n
      Group\u00a0<\/th>Event\u00a0\u00a0Score\u00a0<\/th><\/tr><\/thead>
      negative\u00a0<\/td><0\u00a0<\/td><\/tr>
      -1\u00a0<\/td>=-1\u00a0<\/td><\/tr>
      -2\u00a0<\/td>=-2\u00a0<\/td><\/tr>
      -3\u00a0<\/td>=-3\u00a0<\/td><\/tr>
      positive\u00a0<\/td>>0\u00a0<\/td><\/tr>
      1\u00a0<\/td>=1\u00a0<\/td><\/tr>
      2\u00a0<\/td>=2\u00a0<\/td><\/tr>
      3\u00a0<\/td>=3\u00a0<\/td><\/tr>
      neutral\u00a0<\/td>=0\u00a0<\/td><\/tr><\/tbody><\/table><\/figure>\n
        \n
      • Event Study Design\u00a0\n
          \n
        • Estimation Window\uff1at-270 ~ t-21\u00a0\u00a0\u00a0<\/li>\n
        • Gap\u00a0Days\uff1a10 days\u00a0<\/li>\n
        • Event\u00a0Window\uff1a t-10\u00a0to\u00a0t+30\u00a0\u00a0\u00a0<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n

          The post-event window is extended to +30 days to better capture signal persistence and tradability.\u00a0<\/p>\n

          Event Categories and Backtesting Results\u00a0<\/span><\/h2>\n

          Compared to the\u00a0previous\u00a0chapter\u2019s aggregate analysis at the category level, this section further decomposes events into specific subcategories to\u00a0identify\u00a0how different information sources and event types affect market reactions.\u00a0<\/p>\n

          The analytical approach is adaptive:\u00a0<\/p>\n

            \n
          • When subcategories\u00a0exhibit\u00a0consistent return directions, results are aggregated\u00a0to preserve both sample size and statistical robustness\u00a0<\/li>\n
          • When subcategories show heterogeneous behavior, the analysis focuses on representative and predictive subgroups to enhance precision\u00a0\u00a0<\/li>\n<\/ul>\n

            Based on this framework, 2 representative event groups are examined across governance, fundamentals, and capital structure dimensions.\u00a0<\/p>\n

            P-MO: Announcement-Based Governance Events (Agency Issues)\u00a0<\/span><\/h3>\n

            P-MO events originate from material disclosures (P), including ownership changes, control transfers, and governance disputes.\u00a0<\/p>\n

            These disclosures are released via the MOPS system, including both mandatory regulatory filings and exchange- or firm-initiated announcements, characterized by high standardization\u00a0and rapid information transmission.\u00a0\u00a0<\/p>\n

            Based on this principle, this study focuses on four representative event combinations across three major dimensions\u2014corporate governance, fundamentals, and capital structure\u2014to analyze their individual impacts on stock price performance.<\/p>\n

            This group consists of seven subcategories:\u00a0<\/p>\n

            Subcategory\u00a0<\/th>Description\u00a0<\/th><\/tr><\/thead>
            MO01\u00a0<\/td>Insider \/ major shareholder\u00a0share\u00a0transfers\u00a0<\/td><\/tr>
            MO02\u00a0<\/td>Waiver of subscription rights\u00a0<\/td><\/tr>
            MO03\u00a0<\/td>Insufficient insider shareholding\u00a0<\/td><\/tr>
            MO04\u00a0<\/td>External intervention \/ control disputes\u00a0<\/td><\/tr>
            MO05\u00a0<\/td>Control transfer\u00a0<\/td><\/tr>
            MO06\u00a0<\/td>Share pledging\u00a0<\/td><\/tr>
            MO07\u00a0<\/td>Financial distress of insiders \/ major shareholders\u00a0<\/td><\/tr><\/tbody><\/table><\/figure>\n

            Event Distribution\u00a0<\/strong><\/span><\/h4>\n

            Figure\u00a01 presents the distribution of event scores.\u00a0A total of 849 events\u00a0are\u00a0observed:\u00a0<\/p>\n

              \n
            • Negative: 52.8%\u00a0\u00a0<\/li>\n
            • Neutral: 42.9%\u00a0\u00a0<\/li>\n
            • Positive: 4.4%\u00a0<\/li>\n<\/ul>\n

              Figure 1\uff1aEvent Sample Distribution (P-MO)\u00a0\u00a0<\/strong>\u00a0<\/p>\n

              \"\"<\/figure>\n

              Table 2\uff1aDetailed Statistical Results (P-MO)<\/strong>\u00a0<\/p>\n

              Group\u00a0<\/th>Event Count\u00a0<\/th>Mean AAR (%)\u00a0<\/th>Final CAAR (%)\u00a0<\/th><\/tr><\/thead>
              1\u00a0<\/td>37\u00a0<\/td>0.50\u00a0<\/td>+19.05***<\/strong>\u00a0<\/td><\/tr>
              positive\u00a0<\/td>37\u00a0<\/td>0.50\u00a0<\/td>+19.05***<\/strong>\u00a0<\/td><\/tr>
              neutral\u00a0<\/td>364\u00a0<\/td>0.12\u00a0<\/td>+4.64***<\/strong>\u00a0<\/td><\/tr>
              negative\u00a0<\/td>448\u00a0<\/td>\u22120.06\u00a0<\/td>\u22122.41**\u00a0<\/td><\/tr>
              \u22121\u00a0<\/td>320\u00a0<\/td>\u22120.10\u00a0<\/td>\u22124.03***<\/strong>\u00a0<\/td><\/tr>
              \u22122\u00a0<\/td>83\u00a0<\/td>0.03\u00a0<\/td>+1.27\u00a0<\/td><\/tr>
              \u22123\u00a0<\/td>45\u00a0<\/td>0.03\u00a0<\/td>+1.16\u00a0<\/td><\/tr><\/tbody><\/table><\/figure>\n

              CAAR Analysis\u00a0<\/strong><\/span><\/h4>\n

              CAAR patterns\u00a0exhibit\u00a0clear directional separation,\u00a0indicating\u00a0strong discriminatory power of event intensity.\u00a0<\/p>\n