Research On Stock Index Prediction From Behavioral Finance Perspective

There are many traditional financial theory fails to explain the phenomena in the actualfinancial market, such as investors response to market information excessively orinsufficiently, herding, under pricing of started IPO, etc. Numerous facts show that themarket theory of rational hypothesis is obvious flaws, and the personal interpretation andvolatility of investors for the market information are affecting the investors to make adifferent decision behavior.For this reason, some experts have begun to focus and study on behavior andpsychological fluctuations of investors in the financial decision-making. Under the premisethat keep some general theory of paradigm, the experts using the theory of psychology tostudy the investor behavior, to explore the process of fluctuations in investor psychology andthe influence factors of investment decision, again through the quantification model toconfirm the accuracy of the empirical research.Based on the theory of behavioral finance, neural network and bionic algorithm, toanalysis the evolution and complexity of the stock index series in China securities market,which can help investors to understand of behavior in China securities market and the internalmechanism better, also can help them to understand and grasp the fluctuation regularity of thestock market, and guides them to make correct investment, and to guarantee the healthysustainable development of the securities market in China, which has certain theory value andpractical significance. Specific work content including:(1) Provides an overview of the current stock index prediction method, neural networkmodel and its optimization, and the current situation of the development of intelligent bionicalgorithm; Then emphasis on introduction of the stock index prediction model based on thestatistical principle and its optimization. To compare the common stock index predictionmodels such as BP neural network model, Elman neural network model and support vectormachine (SVM) model, and puts forward the advantages and disadvantages of differentmodels; Also introduces the two common bionic optimization algorithm, genetic algorithmand particle swarm algorithm.(2) Constructs the index system of behavioral finance. First of all, studies the stock pricebehavior research and investor sentiment index, and puts forward the index affectinginvestment behavior, after the screening to factor analysis, to come up with the stock financialindex influencing the behavior. Uses Granger causality analysis to analysis the behavior offinancial indicators, the empirical defined the behavior of the financial indicators has not only quantitative interpretability, also conforms to the economics meaning. Finally, constructs theindex system of behavioral finance.(3) Constructs the index prediction model based on SVM and bionic optimizationalgorithm. According to the characteristics of the stock index prediction and behavioralfinance indexes, uses SVM as a model, and uses the contrast experiment of bionic algorithmto choose kernel function and the best parameters of the model. Then we construct the indexprediction model based on SVM and bionic optimization algorithm.(4) Applied research based on building the index prediction model. Compared with otherprediction model, such as BP and Elman neural network model, and we find that the bestoptimal regression fitting effect model is SVM based on the bionic optimization algorithm,finally, to prove the effectiveness of the proposed model.