Nadaraya–Watson Regression Envelopes with ATR-Based Trading Signals

Summary

This strategy smooths log prices with a Nadaraya–Watson kernel regression and surrounds the estimate with volatility bands derived from ATR. The bands use near and far multipliers, while configurable regression and ATR settings control their behavior. The written description says to buy when price crosses into the envelope from below and sell when it breaks out above; the supplied code instead opens a long when price crosses above the smoothed low series and closes it when the smoothed high series crosses above price.

Published settings show a BTC/USDT futures test spanning about a year, with no reported returns or other performance evidence. The document identifies lag, parameter sensitivity, and false signals in highly volatile markets as limitations, and recommends further validation and filtering. Its claimed adaptability and reduced overfitting are not demonstrated by results in the material, so performance should be assessed independently.

Key ideas

• Kernel regression on log prices forms a smoothed estimate for the envelope.
• ATR and configurable multipliers determine the volatility bands around that estimate.
• The written entry and exit description differs from the conditions implemented in the source.
• The BTC/USDT futures test configuration includes no reported performance results, and the method may lag or whipsaw.

Tags

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