Short Term Prediction Framework for Moroccan Stock Market Using Artificial Neural Networks

Abstract

In this paper we present a short term forecasting framework for stock markets based on Artificial Neural Networks. We are interested in predicting future trends of stock markets (Up, Down and Unchanged) in the very short term (10 to 60 minutes ahead). We present a framework to efficiently implement different ANN architectures namely Multi-Layers Perceptron (MLP) and the Long Short- Term Memory (LSTM). This framework involves the use of intraday prices data (tick-by-tick data) and a selection of technical indicators as input variables and fixes the issues related to the imbalanced target classes and the non-regularly spaced input data.We conduct different experimentations within the proposed framework on data from the Moroccan stock market.