Nadaraya-Watson Trend Following with ATR Bands

Summary

This strategy smooths a selected price series with Nadaraya-Watson kernel regression using a rational quadratic kernel. A configurable lookback window, relative weighting, and regression start bar shape the estimate. Upper and lower bands sit an ATR multiple above and below that estimate, so their distance changes with volatility. The source opens long when price crosses below the lower band and short when it crosses above the upper band; the accompanying description also presents those as the entry rules. The estimate’s slope or a crossover mode can color the trend display.

The document gives the method and parameters but no performance results. Its published backtest settings specify BTC/USDT futures on an hourly chart over a one-month period, which is too limited to establish robustness. The stated risks include lag, false signals around reversals, excess trading in ranges, and sensitivity to parameter choices. The source’s example entries act against band breaks, so the signal logic should be checked carefully before interpreting this as conventional trend following. Historical parameter tuning and confirmation filters are suggested, but no validation evidence is supplied.

Key ideas

• Kernel regression smooths price using a rational quadratic weighting scheme.
• ATR multiples around the regression estimate create volatility-adjusted bands.
• The example enters long below the lower band and short above the upper band.
• The document warns that ranging markets, turning points, and parameter choices can produce poor signals.
• The published BTC/USDT hourly backtest settings do not provide performance results.

Tags

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