Gaussian Filtering and Polynomial Fitting for Price Forecast Signals
Summary
This strategy repeatedly smooths a selected price series with a recursive Gaussian-style filter, then combines filtered series in a polynomial-like extrapolation to estimate a future price. It compares that estimate with the current price: a higher estimate indicates a long signal, while a lower estimate indicates a short signal. The published parameters include a filter length and a selectable price source; the accompanying backtest settings specify BTC/USDT futures data and daily bars with a one-hour base period.
The document presents smoothing as a way to reduce high-frequency noise and describes polynomial fitting as a means of modeling short-term movement. It also notes that smoothing can lag sudden moves, fitted forecasts can overfit, and results depend on window and fitting choices. It proposes retraining, more input features, stop losses, and volatility-aware sizing as possible extensions. No forecast accuracy, trading performance, or benchmark results are reported. The source's specific signal logic is simpler than the prose's forecast description, so the implementation should be inspected before treating the stated method as validated.
Key ideas
- Recursive Gaussian-style filtering is used to create progressively smoothed price series.
- A forecast derived from the filtered series is compared with current price to choose a long or short direction.
- Smoothing can reduce noise but may delay a response to abrupt price changes.
- Polynomial fitting and parameter selection create overfitting and stability risks.
- The document gives no quantitative evidence of forecast accuracy or strategy returns.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.