The trend forecasting the stock index has always been an interesting topic. In the past, there has been several in-depth discussion on the forecast of the stock index. However, these studies developed the prediction models were based on the current quantify index. The purpose of this study was establish a knowledgeable forecast model to expansion of Taiwan stock index futures through the application of artificial neural networks. The input of artificial neural networks included technical analysis and practical experience. This is to enable the inexperienced investors to choose correct investment strategies as the stock index and used to forecast the rise and fall of stock index. The study trained artificial neural networks with Taiwan stock daily lines from August 1997 to September 1998 and forecast the Taiwan stock futures from October 1998 to January 1999. The analysis results show the prediction accuracy was 88%.