Trading Price Crosses of a Rolling Linear Regression Intercept

Summary

This simple directional strategy calculates the intercept of a linear regression fitted to a selected price series over a rolling window. The intercept is the fitted value at the time coordinate zero. The strategy compares the current close with that estimate: a close above it signals a long position, while a close below it signals a short position. A reverse-trading option can invert those directions, and the lookback length and input price source are configurable.

The document explains the calculation through sums of prices and time-weighted prices, then provides source code and published test settings for BTC/USDT futures from late November to late December 2023. It reports no backtest returns, drawdowns, or comparison against a benchmark, so the settings are not evidence of profitability. The author notes that the fit uses historical observations and may not predict future prices, that changes in company fundamentals can invalidate it, and that lookback selection can overfit. Stop losses, complementary indicators, adaptive lengths, and position controls are proposed as possible extensions rather than tested features.

Key ideas

• The strategy fits a linear regression to a configurable rolling window of price data.
• It goes long above the fitted intercept and short below it, with an option to reverse the directions.
• The lookback length and source price are adjustable.
• Historical regression fit does not guarantee useful forecasts, and parameter choice may overfit.
• The published test configuration provides no performance statistics.

Tags

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