Grid-Searching Volume Smoothing and Historical Percentile Thresholds

Summary

This forum discussion outlines a proposed parameter search for a volume-based timing rule on the Consumer Electronics industry index. It tests seven volume-smoothing windows and ten historical-volume percentile cutoffs. The proposed rule exits to cash when smoothed volume exceeds its cutoff, otherwise holding the index, then compares forward returns over roughly one month and the frequency of negative outcomes.

The exchange suggests that a simple loop can run a grid search, while the original poster asks whether the platform’s parameter-optimization tools can perform the same task without manually writing loops or relying on a visual interface. The post is a question and example design, not a completed study: it gives no code, backtest findings, or winning parameter combination. It also leaves key implementation details open, including how positions are entered and evaluated across overlapping forward periods, and how the search would guard against overfitting.

Key ideas

• The proposed experiment varies volume-smoothing windows and historical percentile thresholds.
• The rule would move to cash when smoothed volume exceeds its selected threshold and otherwise hold the index.
• Candidate settings are to be ranked using forward returns and the incidence of negative outcomes.
• A reply describes looping over parameter combinations as a basic grid-search approach.
• The discussion provides an experiment outline rather than results, and it does not resolve overfitting or backtest-design details.

Tags

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