Using AI to Modify a Volume-Cycle Equity Factor

Summary

This Chinese-language case study describes using an AI coding assistant to reproduce and modify a high-frequency volume factor for Chinese equities. The source factor uses intraday volume’s Fourier spectrum to measure energy in a short-period band, with preprocessing such as spike filtering, mean removal, and a Hann window. The author outlines possible extensions: multiple frequency bands, spectral-shape and time-varying features, alternative spike filters, directional buy and sell volume, price-volume synchronization, cross-sectional ranking, and multi-day stability measures.

The worked example focuses on estimating directional volume from minute bars, calculating buy-side periodicity, and running the factor through a SQL and Python workflow. The article reports that the author computed two years of factor history and compared the result with the original; it gives a turnover and IC figure for the unsmoothed factor but leaves the smoothed performance result incomplete. The example is therefore a workflow demonstration, not conclusive validation: directional volume is inferred from bar prices, and the reported evidence is limited, with further automation and analysis still unfinished.

Key ideas

• The source factor measures the share of intraday volume spectral energy in a selected frequency band.
• Suggested extensions include multiple bands, time-varying spectra, directional volume, and cross-sectional normalization.
• The example estimates buy and sell volume from minute-bar price ranges and computes a Fourier-based factor.
• The author reports a two-year comparison, but the excerpt does not provide complete performance results for the modified factor.

Tags

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