Hidden Markov Models for Stock Market Prediction

TL;DR

This paper explores the application of Hidden Markov Models to predict stock closing prices, aiming to improve investment decision-making through more accurate forecasts.

Contribution

It introduces a novel evaluation indicator, Directional Prediction Accuracy, for assessing stock prediction models using HMMs.

Findings

HMM achieved competitive prediction accuracy.

DPA indicator effectively measures directional prediction success.

Model demonstrates potential for practical stock trading applications.

Abstract

The stock market presents a challenging environment for accurately predicting future stock prices due to its intricate and ever-changing nature. However, the utilization of advanced methodologies can significantly enhance the precision of stock price predictions. One such method is Hidden Markov Models (HMMs). HMMs are statistical models that can be used to model the behavior of a partially observable system, making them suitable for modeling stock prices based on historical data. Accurate stock price predictions can help traders make better investment decisions, leading to increased profits. In this article, we trained and tested a Hidden Markov Model for the purpose of predicting a stock closing price based on its opening price and the preceding day's prices. The model's performance has been evaluated using two indicators: Mean Average Prediction Error (MAPE), which specifies the average accuracy of our model, and Directional Prediction Accuracy (DPA), a newly introduced indicator that accounts for the number of fractional change predictions that are correct in sign.