量价淘金”选股因子系列研究(十二)：高频数据+离散化构建方式”在因子研究中的重要性

Key Points - The report emphasizes the importance of "high-frequency data + discretization factor construction" for future research on price-volume factors, suggesting that this combination is crucial for obtaining incremental information with low correlation to existing factors [1][10][9] - Continuous factor construction methods are discussed, highlighting that while minute and tick data can enhance performance, their contribution at the portfolio level is limited, with only a 0.5%-0.6% increase in excess annualized returns when added to daily frequency factors [2][48][41] - Discretization methods are shown to provide significant incremental value at the portfolio level, with a specific example indicating an excess annualized return of 8.15% for a strategy using "continuous + tick discretization" factors compared to a baseline [3][48][4] - The performance of the "tick discretization" factor is analyzed, revealing a monthly IC average of 0.073 and an annualized ICIR of 2.87, indicating its effectiveness even after controlling for common style and industry influences [4][58][17] - The report outlines the construction of various factor clusters, including daily, minute, and tick factors, and their respective performances, demonstrating that minute and tick factors can enhance the overall factor performance when combined with daily factors [2][38][41] - The findings suggest that while continuous factors show strong individual performance, the addition of minute and tick factors does not significantly enhance portfolio performance, indicating a potential saturation of information from these sources [39][41][48]