Stock Index Prediction based on Relation Learning on Technical Indicators by Neural Networks

Abstract

This report discusses using technical indicators to predict stock index movement. The testing index is Hang Seng Index (HSI). The report begins by describing the concept of neural network and Japanese candlestick. Next, two ways of encoding candlestick pattern are proposed. In addition, a program based on neural network is written to identify some well-known candlestick patterns. The report then discusses using neural network and coded candlestick to predict short term up/down trend of the stock index. Finally, the report describes problems and proposed improvements of using candlestick and neural network to predict stock movement.