Brute-Force Market Pattern Search with Fourier Features and Spread Controls

Summary

This article presents an iterative pattern-search tool for trading, including a finite, modified Fourier-series polynomial as an alternative family of functions for describing price behavior. Its coefficients are generated through several random schemes, and candidate formulas are evaluated on historical data. The author also discusses software changes for configuring the tool and mechanisms intended to account for spread and reduce spread noise, which can make apparent patterns untradeable.

The evidence is exploratory: the article mentions a quick search on a year of USDJPY M15 history and notes that the Fourier formulation found variants, but the author says no truly efficient solution was achieved and describes the results as unstable under spread-noise suppression. A reported improvement for one NZDUSD timeframe under a particular coefficient scheme is presented as an observation, not validation. The author cautions against treating the tool as a guaranteed strategy. Historical brute-force selection can overfit, and the excerpt does not provide rigorous out-of-sample evidence establishing durable profitability.

Key ideas

• The method searches historical prices by testing formulas with randomly generated coefficients.
• A modified finite Fourier series adds a function family alongside earlier polynomial approaches.
• Spread costs and spread-related noise can obscure whether a discovered pattern is tradable.
• The author reports exploratory results but also describes the Fourier approach as unstable and not yet efficient.
• The tool is presented as a research aid whose results require further validation.

Tags

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