新价量相关性因子绩效月报20250430-20250506

Quantitative Models and Construction 1. Model Name: RPV (Renewed Correlation of Price and Volume) - Model Construction Idea: The RPV factor integrates intraday and overnight price-volume correlations. It identifies the best representatives for intraday and overnight correlations, combining their information to enhance the factor's effectiveness. The model leverages the reversal and momentum effects in price-volume relationships, incorporating turnover data to strengthen these effects [6][7]. - Model Construction Process: - The RPV factor is derived by combining the intraday price-volume correlation (CCOIV) and the overnight price-volume correlation (COV). - Intraday correlations enhance reversal effects, while overnight correlations capture momentum effects through the "mismatch" in time dimensions, where "yesterday's volume" strengthens overnight returns. - These components are harmonized to serve a unified goal, resulting in the RPV factor [6]. - Model Evaluation: The RPV factor is described as "novel and effective," capable of capturing both reversal and momentum effects in price-volume relationships [6]. 2. Model Name: SRV (Smart Renewed Correlation of Price and Volume) - Model Construction Idea: The SRV factor refines the RPV factor by introducing "smart" metrics. It splits intraday price movements into morning and afternoon segments, identifying informed trading periods. It also replaces turnover with a higher-informed trading proportion for overnight correlations, combining the improved intraday and overnight factors [6][7]. - Model Construction Process: - Intraday price movements are divided into morning and afternoon segments. The "smart" metric identifies the top 20% most informed trading minutes in the afternoon (24 minutes). - The afternoon "smart" turnover rate is correlated with afternoon price movements. - For overnight correlations, the turnover rate is replaced with the turnover from the last 30 minutes of the previous day, which has a higher proportion of informed trading. - The improved intraday and overnight correlations are combined to form the SRV factor [6][7]. - Model Evaluation: The SRV factor is considered superior to the RPV factor, with better performance metrics and lower maximum drawdown [6][7]. --- Model Backtesting Results 1. RPV Model - Annualized Return: 14.97% - Annualized Volatility: 7.72% - Information Ratio (IR): 1.94 - Monthly Win Rate: 73.33% - Maximum Drawdown: 10.63% [7][10] 2. SRV Model - Annualized Return: 17.84% - Annualized Volatility: 6.44% - Information Ratio (IR): 2.77 - Monthly Win Rate: 76.30% - Maximum Drawdown: 3.74% [7][10] --- Quantitative Factors and Construction 1. Factor Name: RPV (Renewed Correlation of Price and Volume) - Factor Construction Idea: The RPV factor combines intraday and overnight price-volume correlations to capture reversal and momentum effects. It incorporates turnover data to enhance these effects [6][7]. - Factor Construction Process: - Intraday price-volume correlation (CCOIV) strengthens reversal effects. - Overnight price-volume correlation (COV) captures momentum effects through time-dimension mismatches. - These components are combined to form the RPV factor [6]. - Factor Evaluation: The RPV factor is described as "novel and effective," capable of leveraging both reversal and momentum effects [6]. 2. Factor Name: SRV (Smart Renewed Correlation of Price and Volume) - Factor Construction Idea: The SRV factor enhances the RPV factor by introducing "smart" metrics for intraday and overnight correlations, improving its ability to capture informed trading signals [6][7]. - Factor Construction Process: - Intraday price movements are split into morning and afternoon segments. - The "smart" metric identifies the top 20% most informed trading minutes in the afternoon. - The afternoon "smart" turnover rate is correlated with afternoon price movements. - For overnight correlations, the turnover rate is replaced with the turnover from the last 30 minutes of the previous day. - The improved intraday and overnight correlations are combined to form the SRV factor [6][7]. - Factor Evaluation: The SRV factor is considered superior to the RPV factor, with better performance metrics and lower maximum drawdown [6][7]. --- Factor Backtesting Results 1. RPV Factor - Annualized Return: 14.97% - Annualized Volatility: 7.72% - Information Ratio (IR): 1.94 - Monthly Win Rate: 73.33% - Maximum Drawdown: 10.63% [7][10] 2. SRV Factor - Annualized Return: 17.84% - Annualized Volatility: 6.44% - Information Ratio (IR): 2.77 - Monthly Win Rate: 76.30% - Maximum Drawdown: 3.74% [7][10]