"Predicting Movements in the Forex Market Using Machine Learning and Genetic Algorithms to Manage Trades"

11/28/2024

This work aims to develop a trading strategy that is capable of profiting in the Forex market, particularly trading with the EUR/USD pair. Divided into two phases, this project starts by using two machine learning algorithms (SVM and XGBoost) to divide and classify market trends into five categories. To train these algorithms, two methods of classifying price data are tested, both labeling candles taking into consideration the direction as well as the strength of price variation. The second stage uses two genetic algorithms (GAs), one optimized for uptrends and another for downtrends in the market. These GAs are used to establish trading rules using the machine learning predictions, generating two sets of signals: the first set determines the best timing to open trades, while the other determines the best timing to close them, allowing trades to remain open for the most profitable duration. To test which type of chart offers the best performance with this system, five different charts are used. The system using different GAs for each market trend and two sets of signals showed positive results, with the 15-minute chart displaying the highest results, achieving an ROI of 11.18% over approximately six months of trading, with a percentage of trades that generated profit of 57.38%, highlighting the ability of the GAs to generate a successful set of trading rules.

Authors in the community:

• 

  A

  Afonso Vira dos Santos Camarão

  ist192999

Supervisors of this institution:

• 

  R

  Rui Fuentecilla Maia Ferreira Neves

  ist14409

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

eng - English

Instituto Superior Técnico