Hybrid support vector machine rule extraction method for discovering the preferences of stock market investors: Evidence from Montenegro

In this study we developed a support vector machine (SVM) rule extraction method for discovering the effects of the features of investors and stock and corporate performance on stock trading preferences. We used this system to combine strengths of two approaches: SVM as an accurate classifier and a decision tree (DT) as a generator of interpretable models. The method is applied to Montenegro data in order to generate interpretable rules for stock market decision-makers. The results showed that this method, in terms of accuracy and interdependency of factors, outperformed the methods for detecting stock trading preferences from previous studies.