2025 年 08 月 29 日 金融工程研究团队 魏建榕（首席分析师） 证书编号：S0790519120001 张 翔（分析师） 证书编号：S0790520110001 傅开波（分析师） 证书编号：S0790520090003 高 鹏（分析师） 证书编号：S0790520090002 苏俊豪（分析师） 证书编号：S0790522020001 胡亮勇（分析师） 证书编号：S0790522030001 王志豪（分析师） 盛少成（分析师） 证书编号：S0790523060003 苏 良（分析师） 证书编号：S0790523060004 何申昊（研究员） 证书编号：S0790122080094 蒋 韬（研究员） 证书编号：S0790123070037 相关研究报告 开源交易行为因子绩效月报（2025 年 8 月） ——金融工程定期 | 魏建榕（分析师） | 高鹏（分析师） | 盛少成（分析师） | | --- | --- | --- | | weijianrong@kysec.cn | gaopeng@kysec.cn | shengshaocheng@kysec.cn | | 证书编号：S079051912000 ...

Quantitative Models and Construction Methods Barra Style Factors - **Model Name**: Barra Style Factors - **Construction Idea**: The Barra style factors are designed to capture the performance of different market styles, such as size, value, growth, and profitability, through specific factor definitions[4][14] - **Construction Process**: - **Size Factor**: Measures the market capitalization of stocks - **Value Factor**: Captures the book-to-market ratio of stocks - **Growth Factor**: Reflects the growth potential of stocks - **Profitability Factor**: Based on earnings expectations[4][14] - **Evaluation**: These factors are widely used in the industry to analyze market trends and style rotations[4][14] --- Open-source Trading Behavior Factors - **Factor Name**: Ideal Reversal Factor - **Construction Idea**: Identifies the strongest reversal days by analyzing the average transaction size of large trades[5][15] - **Construction Process**: 1. Retrieve the past 20 trading days' data for a stock 2. Calculate the average transaction size per day (transaction amount/number of transactions) 3. Identify the 10 days with the highest transaction sizes and sum their returns (M_high) 4. Identify the 10 days with the lowest transaction sizes and sum their returns (M_low) 5. Compute the factor as $M = M_{high} - M_{low}$[43] - **Evaluation**: Captures the microstructure of reversal forces in the A-share market[5][15] - **Factor Name**: Smart Money Factor - **Construction Idea**: Tracks institutional trading activity by analyzing minute-level price and volume data[5][15] - **Construction Process**: 1. Retrieve the past 10 days' minute-level data for a stock 2. Construct the indicator $S_t = |R_t| / V_t^{0.25}$, where $R_t$ is the return at minute $t$, and $V_t$ is the trading volume at minute $t$ 3. Sort minute-level data by $S_t$ in descending order and select the top 20% of minutes by cumulative trading volume 4. Calculate the volume-weighted average price (VWAP) for smart money trades ($VWAP_{smart}$) and all trades ($VWAP_{all}$) 5. Compute the factor as $Q = VWAP_{smart} / VWAP_{all}$[42][44] - **Evaluation**: Effectively identifies institutional trading patterns[5][15] - **Factor Name**: APM Factor - **Construction Idea**: Measures the difference in trading behavior between morning (or overnight) and afternoon sessions[5][15] - **Construction Process**: 1. Retrieve the past 20 days' data for a stock 2. Calculate daily overnight and afternoon returns for both the stock and the index 3. Perform a regression of stock returns on index returns to obtain residuals 4. Compute the difference between overnight and afternoon residuals for each day 5. Calculate the statistic $\mathrm{stat} = \frac{\mu(\delta_t)}{\sigma(\delta_t) / \sqrt{N}}$, where $\mu$ is the mean, $\sigma$ is the standard deviation, and $N$ is the sample size 6. Regress the statistic on momentum factors and use the residual as the APM factor[43][45][46] - **Evaluation**: Captures intraday trading behavior differences[5][15] - **Factor Name**: Ideal Amplitude Factor - **Construction Idea**: Measures the structural differences in amplitude information between high and low price states[5][15] - **Construction Process**: 1. Retrieve the past 20 trading days' data for a stock 2. Calculate the daily amplitude as $(\text{High Price}/\text{Low Price}) - 1$ 3. Compute the average amplitude for the top 25% of days with the highest closing prices ($V_{high}$) 4. Compute the average amplitude for the bottom 25% of days with the lowest closing prices ($V_{low}$) 5. Compute the factor as $V = V_{high} - V_{low}$[48] - **Evaluation**: Highlights amplitude differences across price states[5][15] - **Factor Name**: Composite Trading Behavior Factor - **Construction Idea**: Combines the above trading behavior factors using ICIR-based weights to enhance predictive power[31] - **Construction Process**: 1. Standardize and winsorize the individual factors within industries 2. Use the past 12 periods' ICIR values as weights to compute the composite factor[31] - **Evaluation**: Demonstrates superior performance in small-cap stock pools[32] --- Backtesting Results of Models and Factors Barra Style Factors - **Size Factor**: Return of 0.64% in July 2025[4][14] - **Value Factor**: Return of 0.59% in July 2025[4][14] - **Growth Factor**: Return of 0.16% in July 2025[4][14] - **Profitability Factor**: Return of -0.32% in July 2025[4][14] Open-source Trading Behavior Factors - **Ideal Reversal Factor**: - IC: -0.050 - RankIC: -0.061 - IR: 2.52 - Long-short monthly win rate: 78.3% (historical), 66.7% (last 12 months) - July 2025 long-short return: 0.47%[6][16] - **Smart Money Factor**: - IC: -0.037 - RankIC: -0.061 - IR: 2.76 - Long-short monthly win rate: 82.2% (historical), 91.7% (last 12 months) - July 2025 long-short return: 1.78%[6][19] - **APM Factor**: - IC: 0.029 - RankIC: 0.034 - IR: 2.30 - Long-short monthly win rate: 77.4% (historical), 58.3% (last 12 months) - July 2025 long-short return: 1.42%[6][23] - **Ideal Amplitude Factor**: - IC: -0.054 - RankIC: -0.073 - IR: 3.03 - Long-short monthly win rate: 83.6% (historical), 75.0% (last 12 months) - July 2025 long-short return: 3.86%[6][28] - **Composite Trading Behavior Factor**: - IC: 0.067 - RankIC: 0.092 - IR: 3.30 - Long-short monthly win rate: 82.6% (historical), 83.3% (last 12 months) - July 2025 long-short return: 2.13%[6][31]

Quantitative Models and Construction Methods Model Name: DBD-GRU Model - **Model Construction Idea**: Combining factor slicing theory with deep learning to enhance information extraction and prediction capabilities[4][25] - **Model Construction Process**: - Assume input data x contains features A and B, where feature A is the slicing indicator[4] - Use the median of feature A in the time series as the threshold to construct two masks: mask_Ahigh and mask_Alow[4] - Input the masked data into two branch networks: GRU_high and GRU_low[4] - Take the difference between the outputs of the last time step of the two networks as the input to the output layer[4] - Formula: $ \text{DBD-GRU} = \text{GRU}_\text{high} - \text{GRU}_\text{low} $[4] - **Model Evaluation**: The DBD-GRU model provides significant information increment compared to the original slicing theory factors and baseline GRU model factors[5][55] - **Model Testing Results**: - Ideal Amplitude-DBD RankIC: -10.33%[5] - Ideal Reversal-DBD RankIC: -10.31%[5] - Active Buy-Sell-DBD RankIC: -9.81%[5] Quantitative Factors and Construction Methods Factor Name: Ideal Reversal Factor - **Factor Construction Idea**: Improve traditional reversal factors by slicing the time series data based on transaction amount[14][15] - **Factor Construction Process**: - Retrieve past 20 days of data for selected stocks[17] - Calculate the average transaction amount per trade for each day[17] - Sum the price changes for the 10 days with the highest transaction amounts, denoted as M_high[18] - Sum the price changes for the 10 days with the lowest transaction amounts, denoted as M_low[18] - Ideal Reversal Factor M = M_high - M_low[18] - **Factor Evaluation**: The slicing process effectively distinguishes between strong and weak reversal periods, enhancing the factor's stability and predictive power[24] - **Factor Testing Results**: - RankIC: -6.06%[40] - RankICIR: -2.98[40] - Annualized long-short return: 24.26%[40] - Annualized long-short volatility: 9.38%[40] - Long-short return volatility ratio: 2.59[40] - Maximum long-short drawdown: 7.80%[40] - Monthly win rate: 78.57%[40] Factor Name: Ideal Amplitude Factor - **Factor Construction Idea**: Measure the difference in amplitude information between high and low price states of stocks[78] - **Factor Construction Process**: - Retrieve past 20 days of data for selected stocks[80] - Calculate the daily amplitude (highest price/lowest price - 1)[80] - Calculate the average amplitude for the 25% of days with the highest closing prices, denoted as Amplitude_high[80] - Calculate the average amplitude for the 25% of days with the lowest closing prices, denoted as Amplitude_low[80] - Ideal Amplitude Factor = Amplitude_high - Amplitude_low[80] - **Factor Evaluation**: The factor effectively captures the difference in amplitude information between high and low price states, providing a stable and predictive measure[24] - **Factor Testing Results**: - RankIC: -7.00%[40] - RankICIR: -3.47[40] - Annualized long-short return: 21.02%[40] - Annualized long-short volatility: 10.53%[40] - Long-short return volatility ratio: 2.00[40] - Maximum long-short drawdown: 17.67%[40] - Monthly win rate: 76.98%[40] Factor Name: Active Buy-Sell Factor - **Factor Construction Idea**: Measure retail investors' trading behavior in a declining market environment[79] - **Factor Construction Process**: - Retrieve past 20 days of data for selected stocks[79] - Calculate the daily stock price change and small order inflow intensity[79] - Formula for small order inflow intensity: $ \frac{\text{Active Buy Amount (small orders)} - \text{Active Sell Amount (small orders)}}{\text{Active Buy Amount (small orders)} + \text{Active Sell Amount (small orders)}} $[79] - Calculate the average small order inflow intensity for the 25% of days with the lowest closing prices to obtain the Active Buy-Sell Factor[79] - **Factor Evaluation**: The factor effectively captures retail investors' trading behavior in a declining market, providing a stable and predictive measure[24] - **Factor Testing Results**: - RankIC: -3.39%[40] - RankICIR: -1.27[40] - Annualized long-short return: 10.20%[40] - Annualized long-short volatility: 12.57%[40] - Long-short return volatility ratio: 0.81[40] - Maximum long-short drawdown: 25.26%[40] - Monthly win rate: 70.63%[40] Factor Backtesting Results DBD-GRU Model Factors - **Ideal Amplitude-DBD**: - RankIC: -10.33%[47] - RankICIR: -3.68[47] - Annualized long-short return: 34.31%[52] - Annualized long-short volatility: 15.17%[52] - Long-short return volatility ratio: 2.26[52] - Maximum long-short drawdown: 17.98%[52] - Monthly win rate: 76.98%[52] - **Ideal Reversal-DBD**: - RankIC: -10.31%[47] - RankICIR: -3.57[47] - Annualized long-short return: 37.62%[52] - Annualized long-short volatility: 12.55%[52] - Long-short return volatility ratio: 3.00[52] - Maximum long-short drawdown: 8.96%[52] - Monthly win rate: 80.95%[52] - **Active Buy-Sell-DBD**: - RankIC: -9.81%[47] - RankICIR: -3.63[47] - Annualized long-short return: 33.33%[52] - Annualized long-short volatility: 13.32%[52] - Long-short return volatility ratio: 2.50[52] - Maximum long-short drawdown: 13.82%[52] - Monthly win rate: 75.40%[52] DBD-Combine Factor - **Performance in Major Broad-Based Indices**: - **CSI 300**: - RankIC: -5.76%[63] - RankICIR: -1.87[61] - Annualized long-short return: 14.9%[63] - Annualized excess return: 7.64%[67] - Excess IR: 1.84[61] - Maximum excess drawdown: 3.37%[61] - **CSI 500**: - RankIC: -7.40%[68] - RankICIR: -2.58[65] - Annualized long-short return: 17.5%[69] - Annualized excess return: 7.23%[70] - Excess IR: 1.37[65] - Maximum excess drawdown: 6.43%[65] - **CSI 1000**: - RankIC: -9.84%[75] - RankICIR: -3.48[71] - Annualized long-short return: 30.8%[72] - Annualized excess return: 11.8%[76] - Excess IR: 2.21[71] - Maximum excess drawdown: 3.94%[71]

- Model Name: Barra Style Factors; Model Construction Idea: Measure the performance of common Barra style factors in April 2025; Model Construction Process: Calculate the returns of various factors such as market capitalization, book-to-market ratio, growth, and earnings expectations; Model Evaluation: Provides insights into the performance of different style factors in the market[4][14] - Factor Name: Ideal Reversal Factor; Factor Construction Idea: Identify the strongest reversal days based on the average transaction amount per trade; Factor Construction Process: 1. Retrieve the past 20 days of data for the selected stock 2. Calculate the average transaction amount per trade for each day 3. Sum the returns of the top 10 days with the highest average transaction amount, denoted as M_high 4. Sum the returns of the bottom 10 days with the lowest average transaction amount, denoted as M_low 5. Calculate the Ideal Reversal Factor as M = M_high - M_low[5][46][49] - Factor Name: Smart Money Factor; Factor Construction Idea: Identify the participation of smart money in trading based on minute-level price and volume data; Factor Construction Process: 1. Retrieve the past 10 days of minute-level data for the selected stock 2. Construct the indicator $ S_t = \frac{|R_t|}{V_t^{0.25}} $, where $ R_t $ is the return at minute t and $ V_t $ is the volume at minute t 3. Sort the minute data by $ S_t $ in descending order and select the top 20% of minutes by cumulative volume as smart money trades 4. Calculate the volume-weighted average price (VWAP) of smart money trades, denoted as VWAP_smart 5. Calculate the VWAP of all trades, denoted as VWAP_all 6. Calculate the Smart Money Factor as $ Q = \frac{VWAP_{smart}}{VWAP_{all}} $[5][47] - Factor Name: APM Factor; Factor Construction Idea: Measure the difference in stock price behavior between morning (or overnight) and afternoon sessions; Factor Construction Process: 1. Retrieve the past 20 days of data for the selected stock 2. Record the overnight and afternoon returns for both the stock and the index 3. Perform a regression of the form $ r_t = \alpha + \beta R_t + \epsilon_t $ to obtain residuals 4. Calculate the difference between overnight and afternoon residuals 5. Construct the statistic $ \text{stat} = \frac{\mu(\delta_t)}{\sigma(\delta_t)/\sqrt{N}} $ 6. Regress the statistic against the momentum factor to obtain the APM Factor[5][48][50] - Factor Name: Ideal Amplitude Factor; Factor Construction Idea: Measure the difference in amplitude information between high and low price states; Factor Construction Process: 1. Retrieve the past 20 days of data for the selected stock 2. Calculate the daily amplitude as (highest price/lowest price - 1) 3. Calculate the average amplitude for the top 25% of days with the highest closing prices, denoted as V_high 4. Calculate the average amplitude for the bottom 25% of days with the lowest closing prices, denoted as V_low 5. Calculate the Ideal Amplitude Factor as V = V_high - V_low[5][51] Model and Factor Performance - Barra Style Factors: Market Capitalization Factor return: 0.09%, Book-to-Market Ratio Factor return: 0.11%, Growth Factor return: -0.19%, Earnings Expectations Factor return: -0.02%[4][14] - Ideal Reversal Factor: IC: -0.051, rankIC: -0.061, IR: 2.55, Long-Short Monthly Win Rate: 78.5%, April 2025 Long-Short Return: 0.89%, Last 12 Months Long-Short Monthly Win Rate: 66.7%[6][16] - Smart Money Factor: IC: -0.038, rankIC: -0.061, IR: 2.78, Long-Short Monthly Win Rate: 82.5%, April 2025 Long-Short Return: 0.89%, Last 12 Months Long-Short Monthly Win Rate: 100.0%[6][21] - APM Factor: IC: 0.030, rankIC: 0.034, IR: 2.32, Long-Short Monthly Win Rate: 77.6%, April 2025 Long-Short Return: -0.27%, Last 12 Months Long-Short Monthly Win Rate: 75.0%[6][25] - Ideal Amplitude Factor: IC: -0.054, rankIC: -0.073, IR: 3.04, Long-Short Monthly Win Rate: 83.9%, April 2025 Long-Short Return: 2.52%, Last 12 Months Long-Short Monthly Win Rate: 83.3%[6][30] - Composite Trading Behavior Factor: IC: 0.068, rankIC: 0.092, IR: 3.36, Long-Short Monthly Win Rate: 82.2%, April 2025 Long-Short Return: 0.99%, Last 12 Months Long-Short Monthly Win Rate: 83.3%[6][35]