Classifying Forex Price Changes with Induced Naturality Squares

Summary

The article applies category-theory naturality squares to classify price-change data across three linked currency pairs. USDJPY forms the source series, while EURJPY and EURUSD form the paired destination series. The proposed construction connects their price movements through mappings learned with random decision forests, then uses induction across multiple adjacent squares to extend classification beyond a one-bar comparison. The author presents this as a way to simplify repeated mappings and reduce computation relative to separately building many models.

An MQL5 script prints forecast and observed EURUSD changes as a demonstration, but the article does not establish predictive or trading performance. It explicitly describes the results as inconclusive: the implementation is not integrated as a fully testable multi-currency expert-advisor signal, and further testing is needed to assess useful induction lags. The article also suggests that forecasts might inform position sizing, though this remains a rudimentary extension rather than a validated risk method.

Key ideas

• The proposed classification uses USDJPY as a source series and EURJPY and EURUSD as linked destination series.
• Naturality squares express mappings between series, and induction chains multiple squares to represent larger time steps.
• Random decision forests are used to learn the mappings between price-change observations.
• The script compares EURUSD forecasts with observed changes, but the article reports no conclusive evidence of trading value.
• The author proposes further testing and suggests forecast magnitude could inform bounded position sizing.

Tags

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