Weighted Harrell–Davis Quantiles and Absolute-Deviation Fences

Summary

This indicator estimates a selected price quantile with a weighted Harrell–Davis method over a rolling sample. It uses volume as the default observation weight, with an option to use equal weights, and displays the estimated quantile as a central line. The method combines information across ordered observations rather than selecting a single ranked sample value, while trimming low-density portions of the weighting distribution can reduce the influence of extreme observations and computation.

The script also estimates absolute-deviation bands around the quantile and plots upper and lower fences. It reports a quantile standard error using a Maritz–Jarrett approach. The document explains the statistical motivation and cites prior methodological work, but supplies no market test, trading rules, or performance evidence. The estimator and fences are presented as a draft indicator implementation; users should assess parameter choices, weighting, and behavior across instruments and regimes before relying on them.

Key ideas

• The weighted Harrell–Davis estimator combines sample observations to estimate a chosen quantile.
• Volume provides the default observation weights, and the indicator can instead use equal weights.
• Trimming reduces contributions from low-density parts of the weighting distribution to limit outlier influence and computation.
• Absolute-deviation fences show dispersion around the estimated quantile.
• The script provides a statistical indicator framework but no trading or backtest results.

Tags

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