Adaptive Fourier Decomposition based automatic stock movement forecasting system

The stock market is a complex system. Over the years, people have made unremitting studies in predicting the stock price movements. A novel automatic stock movement forecasting system is proposed in this paper, which is based on a newly developed signal decomposition approach - adaptive Fourier decomposition (AFD). AFD can effectively extract the signal primary trend, which is specifically suitable in the Dow theory based automatic technique analysis. The main functions of the system include: (1) data management, which can read and display the data downloaded from the Internet; (2) data process, which can decompose and reconstruct the signal through AFD; (3) forecasting the primary trend, which can automatically identify the bull or bear markets based on the Dow theory; (4) predicting the shock movement, which can predict the detailed movement by using BP neural network. An experiment is conducted based on the stock market data of Shanghai composite index. Effectiveness of the proposed approach is assessed through the comparison with the direct BP approach and manual observation. The result is proved to be promising.