Integrating grid template patterns and multiple committees of neural networks in forex market

Abstract

We present a hybrid framework for trading in Forex spot market by integrating two different technologies: price patterns based on an array of grid template methods and multiple committees of neural networks. This integration is applied in four currency pairs (EUR/JPY, EUR/USD, GBP/JPY and GBP/USD) using data of a 20 min timeframe. In this research we examine two different fusion approaches for Forex trading: the first one is based on price pattern discovery methods and committees of neural networks as independent entities and works by assigning one entity as basic signal provider and the other as filtering system. The second approach is to take price pattern properties (e.g. forecasting power values) together with technical indicators to feed and train the neural networks. Results show that in both approaches, integration of these independent technologies can improve the trading performance by bringing higher net profits and less risk.