Adaptive wavelet transform model for time series data prediction

With the development of cloud computing and big data, stock prediction has become a hot topic of research. In the stock market, the daily trading activities of stocks are carried out at different frequencies and cycles, resulting in a multi-frequency trading mode of stocks , which provides useful clues for future price trends: short-term stock forecasting relies on high-frequency trading data, while long-term forecasting pays more attention to low-frequency data. In addition, stock series have strong volatility and nonlinearity, so stock forecasting is very challenging. In order to explore the multi-frequency mode of the stock , this paper proposes an adaptive wavelet transform model (AWTM). AWTM integrates the advantages of XGboost algorithm, wavelet transform, LSTM and adaptive layer in feature selection, time–frequency decomposition, data prediction and dynamic weighting. More importantly, AWTM can automatically focus on different frequency components according to the dynamic evolution of the input sequence, solving the difficult problem of stock prediction. This paper verifies the performance of the model using S&P500 stock dataset. Compared with other advanced models, real market data experiments show that AWTM has higher prediction accuracy and less hysteresis.