Building and Combining High-Frequency Factors from Chinese Level 2 Data

Summary

This research summary describes high-frequency equity factors constructed from Chinese Level 2 market data, including minute bars, order-book snapshots, order queues, and trade-by-trade records. The proposed signals measure features such as return skewness, downside variation, closing-period volume, price-volume relationships, order-flow changes, large-trade activity, and inferred informed selling. Combining order submissions with executed trades is presented as a broader measure of investor buying intent.

The summary reports positive cross-sectional rank information coefficients for the factor groups and says that adding them to index-enhancement models improved annualized excess returns for the CSI 300, CSI 500, and CSI 800. It describes two uses: incorporating factors into individual-stock return forecasts and excluding weak stocks. These are reported findings from the cited study, not a reproducible evaluation in the supplied text. The underlying sample period, factor definitions, transaction costs, turnover, and out-of-sample procedures are absent, so the claimed performance should be treated cautiously and may not generalize beyond the studied Chinese market.

Key ideas

• Level 2 data includes minute bars, order-book information, order queues, and detailed trades.
• Minute data, order submissions, and trade records can each support distinct high-frequency factors.
• Combining unexecuted orders with executed trades is proposed as a measure of buying intent.
• The summary reports that high-frequency factors improved index-enhancement results for three Chinese equity benchmarks.
• The supplied text omits implementation details and transaction-cost analysis needed to judge portability.

Tags

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