基于二次分解和TCN的多变量股价预测研究OA

The analysis and forecast of stock market is an important issue to promote the stability of financial market.Aiming at the non-stationarity and complexity of stock market volatility,especially the problem of false information after decomposing the original sequence and filtering related variables,a multivariate mixed model based on quadratic decomposition and TCN is proposed to predict the price of Shanghai Composite Index.The quadratic decomposition algorithm is used to preprocess the original stock price sequence,so as to eliminate noise and capture nonlinear characteristics.At the same time,relevant variables such as techni-cal indicators are introduced as a supplement to the original stock price information.The TCN prediction model is constructed for the decomposed subsequences,and the final prediction value is obtained by weighting the prediction results of each subsequence.Final-ly,the model is established with the Shanghai Composite Index data,and compared with other single models and mixed models.The results show that the proposed hybrid prediction model has lower prediction error than other models.