Filtering Moving Average Crossovers with Price and Statistical Models
Summary
The article examines how to reduce false and delayed signals from a conventional moving average crossover strategy. Its baseline uses fast and slow averages for direction, with ATR-based stop-loss and take-profit levels. A manual refinement adds a price-location filter: long entries require the candle low above the fast average, while short entries require the high below the slow average. The article then compares additional versions that use statistical models to identify patterns missed by hand-designed rules.
The reported backtests show that the baseline lost money, while the wick filter improved net profit and reduced gross loss, though the equity curve remained unstable. Later modeling work reportedly improved out-of-sample forecast RMSE for EURUSD returns and a moving average, but those forecast gains did not improve profitability. The results are tied to the chosen EURUSD sample and test period, and the article cautions against treating lower RMSE as proof of a better trading strategy. The excerpts provide limited numerical detail for the later model variants.
Key ideas
- A basic crossover strategy buys when the fast average is above the slow average and sells when it is below.
- ATR-based stops and targets provide fixed exit levels for comparing entry-rule changes.
- Requiring candle extremes to confirm average alignment can filter some weak crossover entries.
- The manual price filter improved reported profitability, but the resulting equity curve remained unstable.
- Better out-of-sample RMSE did not translate into better trading profitability.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.