Stock Market Trend Prediction in Sub-Saharan Africa Using Generalized Additive Models (GAMs)

Abstract

Pattern discovery emerges as a significant factor to identify the direction of the market. This study sought to test the usefulness of GAMs in predicting the frontier and emerging stock markets in Africa for pattern discovery by comparing its prediction capability to deep neural models namely Long Short Term Memory (LSTM), Recurrent Neural Networks (RNNs), Gated Recurrent Units (GRUs), Bidirectional LSTM, Bidirectional RNN and Bidirectional GRU. Using daily stock market index, the data from Bloomberg database for the period 2012 to 2018, and this study aims to predict daily closing prices for the next 365 days as well as determining the direction of the stock markets. Prediction accuracies were 99.76%, 97.55%, 100%, 99.21%, 99.50%, 99.32%, 99.58%, 99.88%, 99.59% and 99.52% for Botswana, Egypt, Kenya, Mauritius, Morocco, Nigeria, South Africa, Tunisia, Zambia and Zimbabwe stock markets respectively. The GAM model outperformed the deep neural models and it can be used for enhancing investment decision making in Africa.