Group 1: Key Findings - The report focuses on the time difference between order placement and cancellation to identify algorithm-driven trading and capture alpha[8] - The algorithmic trading cancel volume ratio (ACVR) and the algorithmic trading cancel counts ratio (ACCR) factors show strong stock selection effectiveness, with ACCR achieving a RankIC of 0.052 and an ICIR of 0.458 across the entire period[11] - The buy algorithm cancel ratio (BABR) factor has a RankIC of 0.059, ICIR of 0.555, and an IC win rate of 70.1%, indicating its robustness even after neutralizing market capitalization and Barra style factors[11] Group 2: Methodology and Analysis - The distribution of order cancellation time differences shows a concentrated impulse pattern, suggesting algorithm-driven cancellations at specific time points[22] - The report identifies key time points for algorithmic cancellations, such as 1 second and 5 seconds, based on statistical analysis of cancellation behavior[29] - The algorithmic trading buy/sell cancellation entropy (ACE) factor has a RankIC of 0.047 and an ICIR of 0.479, reflecting the consistency of buy/sell opinions in algorithmic cancellations[11] Group 3: Composite Factor Performance - The composite factor, derived from combining ACE and BABR, achieves a RankIC of 0.069, ICIR of 0.851, and an IC win rate of 81.5%, significantly improving stability compared to individual factors[11] - A portfolio constructed from the top 100 stocks based on the composite factor yields an annualized return of 13.03%, with an excess annualized return of 13.77% and an information ratio of 1.44[11]