RULE EXTRACTION FROM RANDOM FOREST FOR INTRA-DAY TRADING USING CROBEX DATA

This work tackles the problem of automatic recognition of favourable days for intra-day trading when looking at the data on Croatian stock index CROBEX. Day trading consists in buying and selling financial instruments within the same trading day and favourable for intra-day trading means that the increase between the opening price and the closing price of the same day is large enough for obtaining a profit by buying at the opening price and selling at the closing price. The problem is to discover the relation between the selected indicators at a certain day and the market situation at the following day which would determine the class of day that is observed, according to technical analysis. In this work, a set of indicators is calculated based on opening, closing, high and low values of CROBEX for ten years. The problem is to predict whether a given day is favourable for intra-day trading or not, based on selected indicators. A binary classification problem is formulated for the described problem of intra-day trading and random forest algorithm is used to predict the outcomes. Using machine learning algorithms in stock market prediction problems have proven to be successful in solving such problems. Random forest is one of such alternative algorithms, very promising when looking at its performance in terms of accuracy. However, a possible limitation of random forest is that it generates a forest consisting of many trees and rules, thus it is viewed as a black box model, as most of the machine learning algorithms are after all. This can possibly be overcome by extracting rules from the underlying random forest model which can allow an easy interpretation of such a model. Obtaining a set of rules that are easy to understand in this way should increase scalability and comprehensibility while hopefully preserving satisfactory level of accuracy. In this paper, a set of rules for intra-day treading problem with CROBEX data is extracted from random forest model.