The foreign exchange (forex) market is the largest and most liquid financial market in the world, with an average daily trading volume of over $5 trillion
Accurately forecasting exchange rates in the forex market is a challenging task due to the complex and dynamic nature of the market. Machine learning (ML) techniques have been increasingly used to forecast exchange rates in the forex market due to their ability to learn from historical data and identify patterns that can be used to make predictions.
This article provides a comprehensive and organized review of the literature on forex market forecasting using machine learning techniques. The review is based on a systematic literature review (SLR) conducted by , which analyzed the use of ML techniques for forex market forecasting. The SLR focused on three particular features of how prior research performed experiments on forex forecasting models, namely:
- Which machine learning model was used?
- Which sorts of assessment procedures were used?
- What validation approaches were used?
The SLR analyzed 46 studies published between 2010 and 2021 that used ML techniques for forex market forecasting. The studies used a variety of ML models, including support vector machines (SVM), artificial neural networks (ANN), decision trees, and random forests. The studies also used a variety of assessment procedures, including mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE). The validation approaches used in the studies included k-fold cross-validation, holdout validation, and rolling window validation.
Machine Learning Models for Forex Market Forecasting
The SLR found that SVM was the most commonly used ML model for forex market forecasting, followed by ANN. SVM is a supervised learning algorithm that can be used for classification or regression tasks. SVM works by finding the hyperplane that maximally separates the data points in different classes. In the context of forex market forecasting, SVM can be used to predict the direction and magnitude of exchange rate movements based on historical data.
ANN is a type of artificial neural network that is inspired by the structure and function of the human brain. ANN consists of multiple layers of interconnected nodes that process and transform input data to produce output predictions. ANN can be used for both classification and regression tasks and has been shown to be effective in forecasting exchange rates in the forex market.
Other ML models that have been used for forex market forecasting include decision trees, random forests, and Bayesian networks. Decision trees are a type of supervised learning algorithm that can be used for classification or regression tasks. Decision trees work by recursively partitioning the data into subsets based on the values of the input features. Random forests are an ensemble learning method that combines multiple decision trees to improve prediction accuracy. Bayesian networks are a type of probabilistic graphical model that can be used to model the relationships between variables in a complex system.
Assessment Procedures for Forex Market Forecasting
The SLR found that the most commonly used assessment procedures for forex market forecasting were MAE, MSE, and RMSE. MAE measures the average absolute difference between the predicted and actual values, while MSE measures the average squared difference between the predicted and actual values. RMSE is the square root of MSE and is a commonly used metric for evaluating the accuracy of regression models.
Other assessment procedures that have been used for forex market forecasting include mean absolute percentage error (MAPE), symmetric mean absolute percentage error (SMAPE), and directional accuracy (DA). MAPE measures the average percentage difference between the predicted and actual values, while SMAPE measures the average percentage difference between the predicted and actual values relative to their average. DA measures the percentage of correct directional predictions, i.e., whether the predicted direction of exchange rate movement was up or down.
Validation Approaches for Forex Market Forecasting
The SLR found that the most commonly used validation approaches for forex market forecasting were k-fold cross-validation, holdout validation, and rolling window validation. K-fold cross-validation involves dividing the data into k subsets and using each subset as a validation set while training the model on the remaining subsets. Holdout validation involves dividing the data into training and validation sets and using the validation set to evaluate the model’s performance. Rolling window validation involves using a sliding window of fixed size to train and validate the model on sequential subsets of the data.
Other validation approaches that have been used for forex market forecasting include walk-forward validation, time series cross-validation, and bootstrapping. Walk-forward validation involves using a sliding window of variable size to train and validate the model on sequential subsets of the data. Time series cross-validation involves dividing the data into training and validation sets based on time periods. Bootstrapping involves resampling the data with replacement to generate multiple training and validation sets.
Conclusion
In conclusion, the use of machine learning techniques for forex market forecasting has been the subject of extensive research in recent years. The systematic literature review conducted by provides a comprehensive and organized overview of the different ML models, assessment procedures, and validation approaches that have been used for forex market forecasting. The review highlights the strengths and weaknesses of different approaches and identifies open challenges and opportunities for future research. Overall, the review suggests that machine learning techniques have the potential to improve the accuracy and reliability of forex market forecasting and can be a valuable tool for traders, investors, and policymakers.
