Comparative Analysis between Genetic Programming and Machine Learning Algorithms in Forecasting Financial Trends

• Marcos V. R. Pedroza IFMG
• Carlos A. Silva IFMG

DOI: https://doi.org/10.5753/eniac.2025.14096

Abstract

This study presents a comparative analysis of Genetic Programming (GP) and five machine learning (ML) algorithms, namely Support Vector Machines (SVM), AdaBoost, XGBoost, Long Short-Term Memory (LSTM), and Deep Neural Networks (DNN), in the task of financial trend forecasting. We use historical daily data from the NASDAQ, S&P 500, and Nikkei 225 indices, covering the period from January 2015 to January 2025. Model performance is evaluated using Sharpe and Sortino Ratios, capturing both accuracy and risk-adjusted return. Results show that GP exhibits greater stability in Asian markets, while LSTM and XGBoost achieve better performance in North American markets.

References

Atsalakis, G. S., Valavanis, K. P., and Emmanouilides, C. J. (2019). Surveying stock market forecasting techniques – part ii: Soft computing methods. Expert Systems with Applications, 36(3):5932–5941. Disponível na ScienceDirect.

Britannica Money (2025). Financial benchmarks — definition, examples, how to use. [link]. Acessado em 21 de junho de 2025.

Chen, W. et al. (2021). Application of genetic programming in financial trading strategies: A comprehensive study. Journal of Computational Finance, 25(3):121–145.

Fischer, T. and Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European Journal of Operational Research, 270(2):654–669. Disponível na Elsevier.

Krauss, C., Do, X. A., and Huck, N. (2017). Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the s&p 500. The Journal of Financial Data Science, 39(4):116–127. Disponível na SSRN.

Long, X., Kampouridis, M., and Jarchi, D. (2020). An in-depth investigation of genetic programming and nine other machine learning algorithms in a financial forecasting problem. Applied Soft Computing, 90:106188.

Shen, J. et al. (2020). Hybrid models combining genetic programming and machine learning for financial forecasting. Journal of Applied Soft Computing, 92:106280.

Troyanskaya, O., Cantor, M., Sherlock, G., Brown, P., Botstein, D., Altman, R. B., and Tibshirani, R. (2001). Missing value estimation methods for DNA microarray gene expression data. Bioinformatics, 17(6):520–525.