Building Low-Frequency Price and Volume Factors from Rolling Statistics

Summary

This note outlines a basic toolkit for constructing low-frequency factors from price and trading data. For each selected input, it proposes calculating rolling means, sums, standard deviations, and skewness over several lookback windows, ranging from short periods to roughly one trading year. Inputs may include open, high, low, and close prices, as well as volume, traded amount, and turnover.

These transformations provide different summaries of recent market behavior: averages and sums describe levels or activity, standard deviation captures dispersion, and skewness describes asymmetry. The note lists candidate variables and windows but gives no formulas beyond the named statistics, code, factor hypotheses, or guidance on normalization and handling correlated inputs. It presents no tests or results showing predictive value. Researchers would need to define the data conventions, avoid look-ahead bias, and evaluate each candidate out of sample before using it in a strategy.

Key ideas

• Rolling means and sums can summarize price and trading activity over multiple lookback windows.
• Rolling standard deviation and skewness add measures of variability and distributional asymmetry.
• Candidate inputs include OHLC prices, volume, traded amount, and turnover.
• The note proposes factor building blocks but provides no backtest or evidence of predictive performance.
• Any implementation needs careful data handling and out-of-sample evaluation.

Tags

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