Prediction of KOSPI Fluctuation Based on S&P500 Index Using a Neural Network Model  
Author Taeseung Kim

 

Co-Author(s) Soowon Lee

 

Abstract Stock price forecasting is an interesting subject that is studied extensively in various fields such as economics, mathematics, statistics, and artificial intelligence. However, it is not easy to predict stock price because stock price movements contain a lot of noise and generally have nonlinear characteristics. In the past, there have been studies to quantify the degree of correlation between various variables and stock prices and to predict future volatility through time series analysis of past stock price patterns. There is also a method based on artificial intelligence that generates a predictive model by learning a machine learning model using data affecting stock price formation. In this paper, we propose a stock price prediction model based on LSTM that predicts the next day 's stock price fluctuation of Korea' s stock index, KOSPI, using S&P500, the US stock index. Comparing the prediction accuracy of the proposed method with the prediction accuracy of the comparison models, the prediction accuracy of the proposed method is about 2% higher than the other models.

 

Keywords Stock price prediction, LSTM, Time series, Neural network, KOSPI, S&P500
   
    Article #:  DSBFI19-36
 
Proceedings of ISSAT International Conference on Data Science in Business, Finance and Industry
July 3-5, 2019 - Da Nang, Vietnam