Can Machine Learning Outperform the Market? Testing the Weak-form Efficiency Hypothesis of the Indian Stock Market Using Support Vector Machines

Abstract

This article examines the validity of the weak form of the efficient market hypothesis (EMH) in the context of the Nifty stock market index by applying a support vector machine (SVM) model. The aim is to forecast future stock prices using historical data and to evaluate the performance of the SVM model based on accuracy, precision, recall and the area under the receiver operating characteristic (ROC) curve (AUC). The findings offer important implications for the efficiency of the Nifty market and its consequences for investors. The EMH posits that stock prices incorporate all available information, making it impossible to consistently beat the market using historical data. This article tests this proposition by using an SVM model to forecast future stock prices using historical data. The methodology consists of applying the SVM algorithm on historical data of the Nifty stock market index. Performance measures, such as accuracy, precision, recall and AUC, are used to assess the effectiveness of the SVM model. The results show an accuracy of 63.25% in forecasting stock prices, indicating a substantial agreement between predicted and actual labels. The precision score of the model is 97.97%, indicating a high proportion of correctly predicted positive instances. However, the recall score is relatively low at 34.36%, suggesting that some actual positive instances were overlooked. The ROC curve visually illustrates the trade-off between true positive rate and false positive rate for different classification thresholds. This article contributes to the literature on market efficiency by applying a novel SVM model to forecast future stock prices of the Nifty index and finding that the model outperforms random chance, thus challenging the weak form of the EMH.