Transparent Models for Stock Market Price Forecasting

Abstract

To aid in the prediction of stock prices, forecasting algorithms can be beneficial. Black box modelling techniques can be, and have been, applied to the prediction of stock prices, but do not offer any clarity on the parameters that have impact on the model predictions. Hence, the focus of this work is to utilise a transparent model to enable the understanding and interpretation of the model output. Stock market data between 2009 and 2019 (a 10-year period) were analysed to determine if a Nonlinear AutoRegressive Moving Average model with eXogenous input (NARMAX) model could accurately predict day ahead stock prices. A NARMAX model was initially developed using a single input variable (Open) and then extended using multiple input variables which included Open, High and Low stock prices. Obtained results revealed that the NARMAX model has strong potential for day ahead price prediction and can be compared against existing techniques for stock market price prediction. Performance evaluation, demonstrated across multiple stock market datasets, demonstrate that NARMAX is efficient in predicting stock market closing price.