Published February 15, 2026 | Version 1.0.0
Technical note Open

Golden Gauss AI: A Predictive Machine Learning System for Probability-Based Trade Execution in Gold Markets

Authors/Creators

  • 1. Independent Researcher

Description

The proliferation of "AI-powered" trading systems in retail markets has created a credibility crisis, with most systems exhibiting catastrophic failure when deployed on unseen data due to severe overfitting. This paper presents Golden Gauss AI, an Expert Advisor designed for the XAUUSD (Gold) market that employs instantaneous probability inference to make autonomous trade execution and management decisions.

Unlike reactive indicator-based systems, Golden Gauss AI processes 239 engineered features through a Gradient Boosting architecture to calculate real-time probability scores, using these probabilities to determine entry timing, position direction, and exit conditions. A novel "predictive labeling" methodology trains the model to recognize setup conditions preceding profitable moves, addressing the fundamental latency problem of traditional technical analysis.

The Expert Advisor incorporates sophisticated trade management logic that minimizes losses even when model predictions are suboptimal, providing robustness across varying market conditions. Deployed via ONNX for real-time inference within MetaTrader 5, the system maintains strict feature calculation parity between the Python training environment and the MQL5 execution engine.

The model was frozen on December 31, 2024, and subsequently validated through a rigorous 13-month historical out-of-sample walk-forward test on previously unseen data from January 2025 through January 2026. This out-of-sample forward test demonstrates approximately 83.67% directional accuracy at an 88% probability threshold, with controlled performance degradation from the training period indicating genuine pattern generalization rather than curve-fitting.

Availability: https://www.mql5.com/en/market/product/164091

Files

Golden_Gauss_AI_Paper.pdf

Files (1.7 MB)

Name Size Download all
md5:1fdcbb4ae8144d85d66c51ac3be954a3
1.7 MB Preview Download

Additional details

Related works

Is supplement to
Software: https://www.mql5.com/en/market/product/164091 (URL)

Dates

Created
2026-02-15

Software

Programming language
MQL5 , Python , C++

References

  • Cao, L. J. and Tay, F. E. H. (2001). "Financial forecasting using support vector machines," Neural Computing & Applications, vol. 10, no. 2, pp. 184–192.
  • Chen, T. and Guestrin, C. (2016). "XGBoost: A scalable tree boosting system," Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 785–794.
  • Chen, Y., Hou, K., and Xue, C. (2020). "Cross-asset momentum spillover," Review of Financial Studies, vol. 33, no. 4, pp. 1581–1623.
  • López de Prado, M. (2018). Advances in Financial Machine Learning. Wiley. Bailey, D. H. and López de Prado, M. (2014). "The probability of backtest overfitting," Journal of Computational Finance, vol. 17, no. 4, pp. 39–69.
  • Pardo, R. (2008). The Evaluation and Optimization of Trading Strategies, 2nd ed. Wiley. Bergmeir, C. and Benítez, J. M. (2012). "On the use of cross-validation for time series predictor evaluation," Information Sciences, vol. 191, pp. 192–213.
  • Baur, D. G. and McDermott, T. K. (2010). "Is gold a safe haven? International evidence," Journal of Banking & Finance, vol. 34, no. 8, pp. 1886–1898.
  • Krauss, C., Do, X. A., and Huck, N. (2017). "Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the S&P 500," European Journal of Operational Research, vol. 259, no. 2, pp. 689–702.
  • Lei, Y. and Xu, J. (2020). "Trailing stop-loss rules and their optimality," Quantitative Finance, vol. 20, no. 8, pp. 1239–1251.
  • ONNX Runtime Developers (2021). "ONNX Runtime: High performance inference and training accelerator."