Nadaraya-Watson Smoothing with Moving Average Crossover

Summary

This strategy smooths a selected price series with a Gaussian-kernel Nadaraya-Watson estimate, then compares the smoothed value with its simple moving average. It opens long when the estimate crosses above the average and short when it crosses below, using a position-state variable to avoid repeated entries. The stated defaults include a bandwidth of 8, a 500-point lookback, a 15-period average, and position sizing at 10% of account equity.

The document explains the intended role of smoothing: reduce price noise while retaining trend changes. It provides implementation parameters and published backtest settings for daily BTC/USDT data over roughly one year, but reports no performance results, so it does not establish profitability. Both smoothing and the moving average can lag, crossover signals may whipsaw in sideways markets, and results may depend on parameter selection. The source also makes the estimate from trailing observations, so it does not require future data for the calculation shown. The document recommends validating parameters and testing before live use.

Key ideas

• The strategy applies Gaussian kernel weights to recent prices to create a smoothed series.
• A simple moving average of that estimate provides the crossover reference.
• Crosses above the average trigger long entries, while crosses below trigger short entries.
• The document identifies lag, sideways-market whipsaws, parameter sensitivity, and computation as limitations.
• Published backtest settings identify the market and period but provide no performance evidence.

Tags

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