Testing Momentum Factors in Chinese A-Shares with Regression, IC, and Portfolios

Summary

This report tests thirteen momentum and modified-momentum signals in China’s A-share market. The signals include returns over several lookback periods, turnover-weighted returns, exponentially weighted turnover-adjusted returns, and a long-horizon regression intercept. It describes three evaluation methods: industry-neutral portfolio sorts, cross-sectional regressions, and information coefficient (IC) analysis. The regression process uses monthly observations, exposure cleaning and standardization, and weighted least squares; the IC procedure adjusts exposures for size and industry.

Across the historical sample examined, the report finds the strongest overall results for selected three- and six-month exponentially weighted signals, a one-month turnover-weighted signal, and a one-month return signal. It also finds that momentum signals tend to be strongly correlated with one another and can overlap with size and liquidity effects, so combining them requires care. The findings are historical and specific to the study’s universe, period, and monthly rebalancing design; they do not establish that the signals will remain effective.

Key ideas

• The study compares thirteen return-based and turnover-adjusted momentum factors using portfolio sorts, regressions, and IC statistics.
• Monthly portfolio sorts can reveal factor ranking and monotonicity, while regression and IC measures assess different aspects of predictive performance.
• Selected exponentially weighted three- and six-month signals and a turnover-weighted one-month signal performed especially well in the reported historical tests.
• Momentum factors were generally positively correlated with one another, raising multicollinearity concerns when they are combined.
• The results are historical evidence from China’s A-share market and may not persist.

Tags

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