Stock price prediction using backpropagation neural network in KOSPI

Abstract

We have applied the Backpropagation Neural Network on time series stock data as a learning rule to improve the prediction rate. We chose data whose stock price showed steadiness and used as the training data and test data. Test was performed while varying the number of hidden layer and hidden nodes each time. As the result shows, the satisfied result was obtained and the use of the backpropagation neural network model improved the performance of the stock price prediction for system trading in KOSPI.