机构行为的十大伪规律

1. Report Industry Investment Rating No relevant content provided. 2. Core Views of the Report In the current low - interest - rate and trading - oriented bond market environment in 2025, the research on institutional behavior has greater significance for interest rate guidance. The report focuses on ten pseudo - laws of institutional behavior, including data caliber, trading nature misjudgments, and model limitations [2]. 3. Summary According to the Directory 3.1 Why We Focus on Pseudo - Laws of Institutional Behavior In the low - interest - rate and trading - oriented bond market environment this year, the research on institutional behavior has increased significance for interest rate guidance, and there are many pseudo - laws in institutional behavior that need to be explored [14]. 3.2 Ten Pseudo - Laws of Institutional Behavior 3.2.1 Are Spot Bond Trading Data T + 0 or T + 1? —— It Should Be T + 1 Spot bond trading data should be analyzed with a one - day lag from interest rate trends because the transactions are usually negotiated the day before [16]. 3.2.2 Do Banks and Brokerages Only Sell Bonds for Distribution? —— There Is Also a Lot of Proprietary Trading Banks and brokerages often show net selling, which may be due to "primary subscription and secondary distribution." Observing old - bond transactions of banks and bond lending of brokerages can help understand their real trading intentions [20][21]. 3.2.3 Is Selling "Smashing" and Buying "Snatching"? —— Focus on the Active Party's Behavior In the secondary spot bond market, it is a zero - sum game. Market trends are driven by the active party. For example, the relationship between rural commercial banks and funds should be analyzed in combination with the active party's behavior [33][34]. 3.2.4 Is Fund Selling a "Redemption Wave"? —— Not Necessarily Historically The current market's panic about fund redemptions may be more of a preventive redemption by wealth management. Measuring the liability - side pressure of funds through the comprehensive holding - cost indicator shows that the current redemption pressure on bond funds is controllable [39][43]. 3.2.5 Are Insurance Companies All Allocation - Oriented? —— The Proportion of Trading - Oriented Has Significantly Increased The trading - oriented proportion of insurance companies has increased, as shown by the fact that they may buy 30Y Treasury bonds for trading purposes and the increase in their 30Y bond selling volume [47]. 3.2.6 Can't We Observe Foreign Investment in Bonds at a High Frequency? —— Most of the "Other" Institutions May Be Foreign Investors Foreign investors' trading behavior can be observed through the "other" type of institutions in high - frequency secondary spot bond trading data. For example, in April, the net buying of "other" institutions was roughly equivalent to the net buying of foreign investors mentioned by the SAFE [53]. 3.2.7 Does Institutional Allocation Increase in the Custody Caliber Mean Buying? —— It May Just Be Due to High Bond Supply An increase in institutional custody may be due to more primary - market bond underwriting or bond maturity, rather than active buying [6]. 3.2.8 High - Winning - Rate Institutional Behavior Indicators May Have Errors —— Interference from the Bull - Market Model Some high - winning - rate institutional behavior indicators may have asymmetric winning rates for buying and selling signals, which may be affected by the previous bull market. When constructing models, the weight of volatile or bear markets should be increased [6]. 3.2.9 Are Non - Linearly Extrapolatable Institutional Behavior Indicators Useless? —— "Direction Sense" Can Also Improve Investment Winning Rates Although some institutional behavior indicators have poor "prediction" effects, they can help investors find the core driving factors of market changes and provide a "direction sense" in investment [7]. 3.2.10 Boundaries of Institutional Behavior Quantitative Models —— Some Errors Such as Short - Term Emotions and Event Impacts Are Still Difficult to Eliminate The existing institutional behavior models have common errors, such as overfitting, being affected by short - term events, and errors in weight assignment. These errors can be reduced but cannot be completely eliminated, so dynamic adjustment and other subjective and objective research are needed [7].