Building Monthly Stock Factors from Intraday Trading Features
Summary
The document describes a way to turn minute-level price and volume data into monthly stock selection factors. It recommends a two-stage process: first summarize each stock’s intraday trading behavior, then aggregate those daily characteristics over the month. The features are grouped around trading sentiment, participant structure, and the state of interaction among market participants, and are presented as capturing information beyond daily open, high, low, and close prices.
Examples reported as effective in long-short testing include intraday beta, the timing of the session high, closing volume share, a volume variation ratio, and a measure of nonstationary price-volume association. The report says monthly information coefficients had absolute values around 3–4%. Most features are described as more useful for excluding weak stocks, while closing volume share is singled out for long-side enhancement. A resulting index-enhancement backtest also reports performance statistics, but the underlying paper is unavailable here; sample construction, costs, and robustness cannot be assessed from this summary.
Key ideas
- The proposed factor pipeline summarizes intraday behavior first and aggregates those summaries monthly.
- Intraday features are grouped into trading sentiment, participant structure, and market interaction characteristics.
- Reported candidate factors include intraday beta, timing of the session high, and closing volume share.
- Most factors are framed as screens for excluding stocks, while closing volume share is also proposed for long exposure.
- The reported backtest evidence is limited here because the full paper and its testing details are not provided.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.