Constructing a Stock Factor from Volume Entropy and Liquidity Elasticity

Summary

This post describes a Chinese equity factor inspired by a research report. Its volume component divides the trading session into five-minute intervals, measures each stock’s share of market volume, and uses Shannon entropy to characterize how that activity is distributed. Its liquidity component compares price-range volatility during volume surges with volatility during ordinary minutes; the proposed elasticity measure is one minus that sensitivity ratio. Both components are transformed by distance from the daily cross-sectional mean, smoothed over twenty days using a mean and standard deviation, and combined with equal weight.

The author reports that the factor’s long-side predictive ability over the analyzed period was not especially strong, while a five-day-return information coefficient of 0.0328 is also stated. The post supplies implementation code, but the reported result is limited and does not establish robust performance. The code and description should be checked carefully before replication, including data alignment, definitions, and the way the smoothing window interacts with batch processing.

Key ideas

• The volume component measures the distribution of each stock’s relative trading volume across intraday intervals using entropy.
• The liquidity component compares price volatility during volume surges with volatility during normal periods.
• Both components are converted to distances from their daily cross-sectional means and smoothed over twenty days.
• The two smoothed components are combined with equal weight.
• The author describes long-side prediction as modest and reports a five-day-return information coefficient of 0.0328.

Tags

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