Adaptive Market Following with Spectral Analysis and Digital Filters

Summary

This article describes an adaptive trend-following system that begins by estimating an instrument’s power spectral density. It favors parametric spectral analysis, particularly the maximum entropy method, to guide the design of digital filters for market data that may be nonstationary. The filters produce fast and slow trend lines, momentum measures, a range-bound channel measure, and a price deviation measure. Trading rules use these indicators to identify direction and entry conditions.

The author outlines tests of the method, including a later test using stop orders. The reported results suggest improved profitability and a higher share of winning trades after stop orders were added, while losses in flat markets remained a concern. The article concludes that the method has potential but needs further work before use in live trading. Its evidence is limited to the described historical or strategy tests; it does not establish robustness across instruments, periods, or changing market conditions.

Key ideas

• The method estimates market spectral structure to adapt the parameters of digital filters.
• Fast and slow low-pass filters create trend lines intended to separate trend from noise and longer cycles.
• Momentum and range-bound indicators are derived from filtered prices rather than raw prices.
• The author favors parametric spectral estimation, especially maximum entropy, for short samples.
• The reported tests improved with stop orders, but flat-market losses and real-market validation remain limitations.

Tags

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