Predicting FOREX trend using incremental spiking neural network
Rasmi Ranjan Khansama, Rojalina Priyadarshini, Surendra Kumar Nanda · 2025
FOREX trend prediction is one of the complex and challenging tasks due to its dynamic and non-linear pattern. An accurate prediction can improve the long-term profit of traders and reduce risks. In recent years, many researchers have investigated the prediction of FOREX rate problem. The objective of this study is to forecast the FOREX trend (that is high or low) instead of point forecasting using an incremental spiking neural network (SNN) algorithm. It is a two-layer architecture and uses a one-pass learning mechanism to model the trend of FOREX time-series. This model is tested on two major FOREX such as USD - INR and USD-EUR. The result obtained using this model is quite noticeable. It shows that the model predicts the trend more accurately with 86% with the best value of no. of Gaussian receptive fields. From the experiments, we also infer that the model outperformed widely acknowledged models such as support vector machine (SVM), Logistic Regression (LR), Random Forest (RF) and Neural Network (NN). Moreover, the incremental SNN model is statistically significant than SVM and Neural network as shown by the McNemar test with 5% significance level. To the best of our knowledge, this is the first study of an incremental model for FOREX trend prediction. We are optimistic that this model might be scaled up in financial trend forecasting. The experimental results of the model on FOREX prediction show the higher applicability and accuracy of the model.