In order to explore the correlation between the international financial markets and the Taiwan stock market, and to effectively organize and summarize information in time series data for better prediction of future market trends, this study constructs an LSTM neural network model to investigate whether investors can predict the trends of the Taiwan Stock Exchange Weighted Index (TAIEX) using variables such as gold prices, US dollar exchange rates, and CBOE VIX. Empirical results demonstrate that the LSTM neural network model constructed using variables such as gold prices, USD/TWD exchange rates, and CBOE VIX can effectively predict the trends of the TAIEX. The results obtained from this study are expected to serve as references for hedging investment strategy.