Research on Stock Price Prediction Technology using RNN and Y Text Miner

Authors

  • Sunghyuck Hong
  • Jungsoo Han

Abstract

The stock price prediction system has been developed and, it is being used. Most of them use the time series inference to analyze past stock price analysis data and the recent artificial intelligence based deep learning to learn the pattern of stock price fluctuations and the future. However, there are many factors that affect stock prices, and it is literally impossible to forecast the future with historical data, and the stock market will not be able to repeat past trading patterns, and it will respond to new situations at every moment including international relations, specific events, and celebrity comments. Because of the impact, even with deep learning algorithms, it is nearly impossible to predict exact the price of stocks that change over time. In addition, the 2013 Nobel Prize winner in Economics Eugene Fama mentioned that it can make profits from the market with a temporary strategy, but if other market participants grasp the pattern, and the strategy is not effective. Therefore, it is impossible to continue to win the market. Realistically, Eugene Fama mentioned that it could be able to make a profit if one wants to make a profit, and then it responds to a changing market. Therefore, based on historical data, finding the closest situation to when the stock price has risen will be able to predict the stock price continuously. This research collects all the possible news texts from the Internet by using YTextMiner and analyses time series data from past stock market using RNN(Recurrent Neural Network). This research results can effect on all kinds of prediction system such as earthquake prediction, weather forecast, crime prediction, and election.

Downloads

Published

2020-03-26

Issue

Section

Articles