Improving Stock Market Prediction through Reinforcement Learning: A Case Study of Nigerian Telecommunication Companies.

Abstract

Stock prediction a viable tool in building a robust economy for a nation. It provides the impetus of the how, the who and when to invest in a stock primarily used by investors. There is a need for a more effective model of stock prediction which manages the volatile and dynamic nature of the stock market. As such, this research aims at developing a more efficient predictive model; using the Reinforcement Learning RL technique that involves an agent learning to interact with an environment in order to maximize a reward signal. The research focuses on the improvement of the benchmark model, Ordinary Least Square (OLS) model of stock prediction achieved by the value-based RL technique; the reinforced model predict close to actual when compared to the OLS model. Statistical evaluation metric was employed to confirms the efficiency of the reinforced model against the OLS model for real life application.