Prediction of Malaysian Exchange Rate Using Microstructure Fundamental and Commodities Prices: A Machine Learning Method

Abstract

The key objective of this research is to investigate the short run dynamics of the exchange rate using commodity prices and microstructure market variables for developing economies, Malaysia. The analysis of the literature revealed different school of thought where one claims the strong correlation among the variables while other significantly reject the relationship. There is mixed results that support and reject the accurate forecasting of the exchange rate through different determinants. Therefore, in this study the machine learning approach is applied to perform the extensive experiments and investigate the relationship between the commodity prices, bid-ask spread and exchange rate. Three techniques were selected from the machine learning i.e. artificial neural network, RandomForest and support vector machine. The experimental results revealed that randomforest perform better than SVM and ANN, both in the performance and accuracy. Thus, the exchange rate can be predicted will reasonable accuracy using the commodity prices in the combination of the bid-ask spread. Thus, the policy maker can be utilized these results for strategies development, corporate planning and building investment plan.

Malaysian Daily Stock Prediction Analysis Using Supervised Learning Algorithms

“…This suggests that the application of the SVM algorithm is less fitting for the five Malaysian stocks used in the paper. Similarly, the recent paper by [20] stated that RF has higher accuracy than SVM in predicting asset prices on Malaysian commodities prices and microstructure market variables. It is also found that the effectiveness of the algorithm used depends on the size of the dataset of the study.…”