Benchmark-Aware Factor Testing for China Index Enhancement

Summary

This research summary argues that standard single-factor information coefficient tests may not reflect index-enhancement practice. A factor’s performance in a broad shared universe may not transfer to a benchmark’s eligible stocks, and simple tests may omit portfolio constraints that matter in implementation. The proposed alternative evaluates single-factor enhanced portfolios within each target benchmark’s stock universe.

The study tests 164 factors across six categories in the CSI 300, CSI 500, and CSI 1000 universes. It then combines factors using information ratios, maximum drawdowns, and relative performance across factor categories, with a seasonal adjustment based on an observed Spring Festival effect. The summary reports historical excess-return and information-ratio results for each benchmark, alongside tracking-error limits, but gives no test period or detailed implementation assumptions in the available text. The authors caution that the models rely on historical data and may fail as market conditions change; the reported results therefore should not be treated as a guarantee of future performance.

Key ideas

• Evaluate factors within the eligible stock universe of the intended benchmark.
• Single-factor enhanced portfolio tests can represent practical constraints that IC tests may miss.
• The study tests 164 factors across six categories for three Chinese equity benchmarks.
• The multi-factor models use information ratios, maximum drawdowns, and relative category performance.
• The reported historical results are subject to model and market-regime risk.

Tags

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