Sequential Autoregressive Stock Market Forecasting Algorithm For Morphological Feature Calculation

Stock market is deeply impacted by many factors,which lead to fluctuations in the stock market with instability and high frequency characteristics.Especially in the W form of the stock market in the bottom stage is facing the choice of direction,this time stock market volatility is greater than usual.Based on the wave theory,the thesis chooses the W form in the stock market as the research object.Through the consideration of different indexes and the relationship between indexes,the thesis completes the definition,quantification and prediction of the W shape of the stock market by the aspects of morphological maturity and morphological average hysteresis.The maturity is the quantification of the integrity of the W-form,and the average hysteresis is a quantification of whether or not the W-shape trend is synchronized with the form.Finally the construction of the forecasting model is completed,which is of great significance to the investors' investment and the financing of the enterprise.The main work is as follows:(1)According to the completeness of the construction of W form,A time series autoregressive stock market prediction algorithm SMA with structural maturity is proposed.First,according to Eliot wave theory leads to the W form,from the W shape of the crest,edge,trough to complete the definition of W form.According to the definition of W form,we extract the relevant indicators from the view of maturity point to complete the quantitative calculation of the W shape node maturity,establish Bayesian Network of node maturity in the basis of W-type fluctuations in the timing and the relevance before and after node,complete quantification of relationship strength between nodes maturity by asymmetric information entropy,combined with the relationship strength and node maturity to complete the calculation of node structure mature.Finally,construct a prediction algorithm by combined with node structure maturity and the increase of the target node,we can forecast the next increase of W shape,the algorithm can improve the ability of risk aversion and reduce the error(2)Although the construction of W form can fully reflect the release of W-form risk and the accumulation of rising opportunities,with the W-shape is built in the process,W-shaped trend is not necessarily able to achieve the transition.In order to solve the above problems,the thesis proposes a time series autoregressive stock market situation prediction algorithm based on moving average.First,extract the relevant indicators to complete the definition and quantify of W form.Then,the 20-day moving average is a technical indicator that completes the definition and quantification of W-shape trends,and calculate the average hysteresis of the node.Then,the Markov blanket is constructed with the W-shape node and the node meaning hysteresis,and the non-symmetric information entropy is used to describe the quantization of the relationship between the node and the hysteresis.Then,the degree of structural hysteresis is calculated by combining the relationship strength and the average hysteresis.Finally,the prediction algorithm is constructed by combining the structural lag and the target node gain,the stability of the algorithm is improved and the prediction accuracy is improved(3)This thesis makes an empirical analysis on the algorithm SMA,the algorithm DSMA and the algorithm PCA-BP in the empirical data.According to the conclusion of the analysis,The accuracy of the algorithm is improved by introducing maturity and averaging hysteresis factors.