Applying the Support Vector Machine for Testing Pricing Inefficiency on the Stock Exchange of Mauritius

Abstract

A popular Machine Learning Technique called the Support Vector Machine (SVM) is adopted on the Stock Exchange of Mauritius (SEM) to determine if stock market returns are predictable based on information from past prices, allowing arbitrage opportunities for abnormal profit generation. The serial correlation test, used as benchmark, and the SVM technique show evidence that previous information on share prices as well as the indicators constructed are useful in predicting share price movements. The implications of the study are that investors have the prospect of adopting speculative strategies and profits from trading based on information and advanced techniques and models are possible.