Construction of a prediction model for individual investors' psychology and behaviour based on cognitive neuroscience

Traditional forecasting models cannot extract the trend information of retail investors' multi-scale psychological and behavioural data, and the predictions are not accurate. To solve this problem, a Markov-based individual investor psychology and behaviour prediction model is proposed. Using the wavelet multi-scale analysis method, the multi-scale data of individual investor's psychology and behaviour are extracted. A long-term-memory analysis is performed on multi-scale data of individual investors' psychology and behaviour using the correlation analysis method, and the trend information is extracted. On this basis, a Markov prediction model is established, and a modified investment preference model is introduced to improve the accuracy of the prediction. Using the individual similarity degree, the nearest neighbour set of the target individual is established, and a multi-order predictive Markov fusion model for multiple individuals is formed to achieve accurate prediction. The experimental results show that the proposed model achieves better nonlinear fitting and higher prediction accuracy.