Model of stock market bubble based on GD-FNN
Sun Bin, Tieke Li · Systems Engineering - Theory & Practice · 2012
For complexity of inside structure and variability of external factors of stock market which make stock market forecasting and identification complex problems,stock market bubble forecasting model was proposed with early-warning strategy.The forecasting index system of bubble model was set which involves price and volume of Shanghai Composite Index and macroeconomic indicators associated with stock market. The long-run equilibrium and cause and effect relationship among the index variables were analyzed.Based on the index system,vector auto regression model(referred to as VAR) was proposed to measure basic value of stock market and impact on stock market of macroeconomic indicators was analyzed.Generalized dynamic fuzzy neural network(referred to as GD-FNN) model was proposed to measure market value of stock market which is based on elliptic basis function and can dynamically adjust the network structure and fuzzy rules of financial system were extracted which can reveal financial nonlinear system operating mode.Finally,according to deviation between market value and basic value,bubble degree of stock market was calculated with corresponding early-warning strategy.