Delta-RSI Momentum Signals Using Polynomial Differentiation and Fit Filtering
Summary
The strategy treats Delta-RSI as a smoothed estimate of the time derivative of RSI. It fits a polynomial to a rolling RSI sample using a least-squares procedure, then evaluates the fitted curve’s derivative at the latest point. An exponential moving average of that derivative serves as a signal line. The polynomial order, RSI lookback, fit window, and signal smoothing length are configurable.
Entries, exits, and short signals can each use one of three conditions: crossing the zero line, crossing the signal line, or a direction change while the oscillator is on the corresponding side of zero. An optional filter rejects signals when normalized root-mean-square fitting error exceeds a user threshold. The code plots entry and exit markers and defines alert conditions. It provides a method and implementation, but no market-specific guidance, backtest results, transaction costs, or evidence that the fit filter improves performance; parameter choices and out-of-sample behavior remain unvalidated in the document.
Key ideas
- Delta-RSI estimates the derivative of RSI by fitting a polynomial over a rolling window.
- An EMA of the derivative provides an optional signal line for cross-based entries and exits.
- Zero crossings and direction changes offer alternative signal rules for long, short, and exit conditions.
- A normalized fit-error threshold can filter signals when enabled.
- The document gives implementation details but no evidence of profitability or out-of-sample robustness.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.