金融工程定期：开源交易行为因子绩效月报（2026年2月）-20260227

Quantitative Models and Construction Methods Barra Style Factors - Model Name: Barra Style Factors - Construction Idea: The model tracks the performance of common style factors such as size, value, growth, and profitability[3][13] - Specific Construction Process: The factors are calculated based on predefined metrics. For example: - Size Factor: Measured by market capitalization - Value Factor: Measured by book-to-market ratio - Growth Factor: Measured by growth-related metrics - Profitability Factor: Measured by earnings expectations[3][13] - Evaluation: The model provides a comprehensive view of style factor performance across different dimensions, aiding in understanding market trends[3][13] Open-Source Trading Behavior Factors - Factor Name: Ideal Reversal Factor - Construction Idea: Identifies trading days with the strongest reversal attributes based on large transaction sizes[4][13] - Specific Construction Process: 1. Retrieve the past 20 days of data for a stock 2. Calculate the average transaction size per day 3. Identify the 10 days with the highest and lowest transaction sizes 4. Compute the cumulative returns for these days: $M_{\text{high}}$ and $M_{\text{low}}$ 5. Calculate the factor as $M = M_{\text{high}} - M_{\text{low}}$[40][42] - Evaluation: Captures micro-level reversal forces effectively[4][13] - Factor Name: Smart Money Factor - Construction Idea: Tracks institutional trading activity using minute-level price and volume data[4][13] - Specific Construction Process: 1. Retrieve the past 10 days of minute-level data 2. Calculate the indicator $S_t = \frac{|R_t|}{V_t^{0.25}}$, where $R_t$ is the return and $V_t$ is the volume for minute $t$ 3. Sort minutes by $S_t$ and select the top 20% by cumulative volume 4. Compute the volume-weighted average price (VWAP) for these minutes ($\text{VWAP}_{\text{smart}}$) and for all minutes ($\text{VWAP}_{\text{all}}$) 5. Calculate the factor as $Q = \frac{\text{VWAP}_{\text{smart}}}{\text{VWAP}_{\text{all}}}$[41][43] - Evaluation: Effectively identifies institutional trading patterns[4][13] - Factor Name: APM Factor - Construction Idea: Measures the difference in trading behavior between morning and afternoon sessions[4][13] - Specific Construction Process: 1. Retrieve the past 20 days of data 2. Calculate daily overnight and afternoon returns for both the stock and the index 3. Perform regression to obtain residuals for overnight ($\epsilon_{\text{overnight}}$) and afternoon ($\epsilon_{\text{afternoon}}$) returns 4. Compute the daily difference $\delta_t = \epsilon_{\text{overnight}} - \epsilon_{\text{afternoon}}$ 5. Calculate the statistic $\text{stat} = \frac{\mu(\delta_t)}{\sigma(\delta_t) / \sqrt{N}}$, where $\mu$ is the mean, $\sigma$ is the standard deviation, and $N$ is the sample size 6. Regress $\text{stat}$ against a momentum factor and use the residual as the APM factor[42][44][45] - Evaluation: Captures intraday behavioral differences effectively[4][13] - Factor Name: Ideal Amplitude Factor - Construction Idea: Measures the structural differences in amplitude information between high and low price states[4][13] - Specific Construction Process: 1. Retrieve the past 20 days of data 2. Calculate daily amplitude as $\text{Amplitude} = \text{High Price} / \text{Low Price} - 1$ 3. Compute the average amplitude for the top 25% (high price) and bottom 25% (low price) trading days 4. Calculate the factor as $V = V_{\text{high}} - V_{\text{low}}$[47] - Evaluation: Highlights structural differences in price amplitude effectively[4][13] - Factor Name: Composite Trading Behavior Factor - Construction Idea: Combines the above factors using ICIR-based weights to enhance overall performance[31] - Specific Construction Process: 1. Standardize and winsorize individual factors within industries 2. Use the past 12 months' ICIR values as weights to compute the composite factor[31] - Evaluation: Provides a robust and comprehensive measure of trading behavior[31] --- Model Backtesting Results Barra Style Factors - Size Factor: Return of -0.44% in February 2026[3][13] - Value Factor: Return of 0.16% in February 2026[3][13] - Growth Factor: Return of -0.15% in February 2026[3][13] - Profitability Factor: Return of 0.00% in February 2026[3][13] Open-Source Trading Behavior Factors - Ideal Reversal Factor: - IC: -0.048 - RankIC: -0.060 - IR: 2.39 - Monthly win rate: 77.5% (historical), 58.3% (last 12 months) - February 2026 return: -0.40%[5][14] - Smart Money Factor: - IC: -0.037 - RankIC: -0.062 - IR: 2.69 - Monthly win rate: 80.4% (historical), 66.7% (last 12 months) - February 2026 return: -0.76%[5][19] - APM Factor: - IC: 0.028 - RankIC: 0.034 - IR: 2.25 - Monthly win rate: 75.8% (historical), 41.7% (last 12 months) - February 2026 return: -0.45%[5][23] - Ideal Amplitude Factor: - IC: -0.053 - RankIC: -0.073 - IR: 2.99 - Monthly win rate: 82.6% (historical), 66.7% (last 12 months) - February 2026 return: -0.67%[5][26] - Composite Trading Behavior Factor: - IC: 0.065 - RankIC: 0.093 - IR: 3.23 - Monthly win rate: 79.1% (historical), 58.3% (last 12 months) - February 2026 return: -0.60%[5][31]