Decomposing Stock Price Drivers into a Composite Intraday Factor

Summary

This Chinese equity research describes a monthly stock-selection factor built from minute-level returns and changes in trading volume. For each stock and day, the authors regress returns on contemporaneous and lagged volume changes. They interpret the volume terms as proxies for sudden stock-specific information, the intercept as a mixture of market and longer-term fundamental influences, and unexplained residual variation as noise. These components are combined into three subfactors intended to capture smoother information arrival, weaker noise influence, and a larger market-related share of the intercept signal.

The three subfactors are equally weighted into the composite, with lower values generally treated as more favorable. The report presents historical rank correlations, long-short results, and tests across broad Chinese equity indexes, including versions adjusted for industry and style exposures. These are backtest findings, not guarantees: the report warns that historical relationships can weaken or fail, and the proposed proxies depend on assumptions about what regression statistics represent. The excerpt does not establish that those interpretations are causal or robust in future markets.

Key ideas

• The method decomposes daily minute-level returns using current and lagged changes in trading volume.
• Regression statistics are treated as proxies for sudden information, market and fundamental influences, and noise.
• Three component signals are averaged into a monthly composite stock-selection factor.
• The report presents historical tests across the broad market and major Chinese equity indexes, including adjusted results.
• The authors warn that historical factor relationships may not persist and that market conditions can change.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.