Enhancing Forex Forecasting Accuracy: The Impact of Hybrid Variable Sets in Cognitive Algorithmic Trading Systems

TL;DR

The paper addresses forecasting EUR/USD in a high-frequency Forex setting by testing cognitive ATS that fuse fundamental macro data with technical indicators. It uses LSTM networks to learn from a hybrid feature set and introduces a directional target $Y(n)$ based on a forward-looking index $D(n,h)$ with horizon $h=10$. Key findings show that models incorporating fundamental data achieve higher predictive power (AUC around 0.64–0.65) and that carefully tuned architectures with 4 layers, ~20 epochs, and a 20-day look-back yield robust out-of-sample profitability, especially under dynamic position management. The work demonstrates that hybrid data integration can provide a statistical edge in currency markets and outlines pathways for future enhancements like adaptive thresholds and multi-time-frame integration.

Abstract

This paper presents the implementation of an advanced artificial intelligence-based algorithmic trading system specifically designed for the EUR-USD pair within the high-frequency environment of the Forex market. The methodological approach centers on integrating a holistic set of input features: key fundamental macroeconomic variables (for example, Gross Domestic Product and Unemployment Rate) collected from both the Euro Zone and the United States, alongside a comprehensive suite of technical variables (including indicators, oscillators, Fibonacci levels, and price divergences). The performance of the resulting algorithm is evaluated using standard machine learning metrics to quantify predictive accuracy and backtesting simulations across historical data to assess trading profitability and risk. The study concludes with a comparative analysis to determine which class of input features, fundamental or technical, provides greater and more reliable predictive capacity for generating profitable trading signals.