Filtering Shanghai-Shenzhen 300 Constituents with Intraday Short Factors

Summary

This research summary examines whether excluding stocks ranked poorly by high-frequency factors can improve a Shanghai-Shenzhen 300 index-enhancement portfolio. It compares three construction methods: combine standardized factors into one score, use a regression model to estimate returns and remove the lowest-ranked stocks, or combine separate single-factor short portfolios. It recommends orthogonalizing high-frequency factors against style and industry effects and building the composite outside the index constituents to capture effects beyond the benchmark universe.

The summary reports backtest comparisons: with a five-percent short threshold, the standardized composite approach weighted by ICIR raised annualized excess return from 15% to 16.7%; regression and portfolio-combination methods reached 16.3% and 16.0%. It says factor screening can reduce sensitivity to the removal threshold, while regression is particularly sensitive to factor selection and correlation. These are historical backtest findings, not a guarantee of future results; the summary flags model specification, decay of statistical patterns, and liquidity as risks.

Key ideas

• The study compares standardized factor aggregation, return-prediction regression, and combinations of single-factor short portfolios.
• It recommends orthogonalizing high-frequency factors against style and industry effects.
• The summary reports improved historical excess returns after removing stocks ranked in a multi-factor short portfolio.
• Regression results depend on selecting factors carefully and reducing correlation among predictors.
• Model misspecification, changing statistical relationships, and liquidity are identified as risks.

Tags

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