Neural Network Forecasting of TAIMEX Index Futures

摘要

This paper applies a neural network model to the forecasting of the TAIMEX (Taiwan International Mercantile Exchange) stock index futures. This model employs technical indicators as the input variables to forecast the price movements of TAIMEX index futures. It provides trders with a highly profitable forecasting model by using different forecasting target, trading strategies, and capital allocation strategies. Two types of forecasts are made in this study: (a) 5-day price movement, and (b) price movement from next open to cllse. Capital allocation is also considered by entering different numbers of contracts at different times. The results of experiments show that our model is quite accurate and profitable.