Forex Trading Systems based on Artificial Neural Networks with Technical Analysis Indicator Inputs

Abstract

Since the beginning of the first financial markets, the fundamental and technical analysis have been widely used by traders for the price movements prediction.. But in recent times, many investors start to use trading systems based on neural networks in the decision making process. This is due to the computer science and technological progress, and the fact that the use of artificial intelligence as a base prediction technique has been proved to be extremely successful. Building these kinds of transactional systems is an art, that can help to answer the question of where to invest the capital. Not many traders and researches combine technical indicators with trading systems based on neural networks. Why is it so? Maybe technical indicators do not influence these kinds of systems? Or efficient market hypothesis is true and no matter what trading strategy is used it is impossible to systematically outperform the market? This thesis tries to find at least partial answers to the above questions. In the first chapter foreign exchange market its characteristic as well as efficient market hypothesis is discussed. The second chapter describes the structure and function of artificial neural network in predicting financial markets behavior. There is also shown a six-step approach for neural network financial prediction model building. Specifically technical indicators are discussed in third chapter: MACD, Stochastics(K%D), Momentum and ADX. There is technical and fundamental analysis shortly discussed. The last chapter describes a survey of technical indicators influence on the trading system based on neural network. The main focus of this research is to prove the hypothesis, that technical indicators have substantial influence on the trading system based on artificial neural network, and that they can help to give better prediction performance.