Comparative Analysis Of The Price Prediction Model Of The New Shanghai Composite Index

In the capital market,the stock market occupies a decisive position.Its role is to attract and guide dispersed liquid assets and savings to the optimal path,so that scarce financial resources are fully allocated to the most profitable projects.The fluctuation of stock prices is affected by a variety of external factors and internal factors,such as politics and economy.Price is an effective signal to guide the effective allocation of capital and liquidity.An important issue in the stock market is how to simulate the price of the stock index in order to make a suitable forecast.Stock investors need to determine the actual value of the stock and compare it with the market price to guide their stock trading.In this paper,the ordinary geometric Brownian motion model is first used to simulate the stock price through the discretization method,but the fitting result obtained is not very satisfactory.Observing the fluctuation trend of the stock index curve,we construct a geometric Brownian motion model with threshold based on the original geometric Brownian motion with the control function of the threshold model.Solve the upper and lower thresholds to perform an overall piecewise fitting.This model is more in line with the stock index fluctuations.Based on the characteristics of the data,we choose the autoregressive summation moving average model(ARIMA)model and the autoregressive moving average modelgeneralized autoregressive conditional heteroskedasticity(ARMA-GARCH model)to fit the data.Finally,the four models are analyzed and compared.The conclusion is that the geometric Brownian motion model with threshold is better than the geometric Brownian motion model.Among the four models,the ARIMA model has the smallest prediction error of the new composite index price.