A novel method based on FTS with both GA-FCM and multifactor BPNN for stock forecasting

The fuzzy time series (FTS) model has been proposed for many years, and many researches have been conducted to improve or enhance the model. This study proposed a novel method for stock forecasting, which is based on FTS forecasting with genetic algorithm (GA)-fuzzy C-means (FCM) and multifactor back-propagation neural networks (BPNN). The GA algorithm is utilized to alleviate the FCM's issue of falling into local optimum in the process of partitioning the universe of discourse and fuzzifying the time series. The multifactor BPNN considers relatively more information to train the neural networks and then forecast new stock index fluctuations. Finally, the proposed method is compared with other previous research methods by using SSECI and TAIEX data to verify the proposed method's effectiveness and efficiency in forecasting financial time series.