During the “ 14th Five-Year Plan ” period, cotton has become a key factor restricting the development of the textile industry, so it is proposed to study the volatility of cotton futures price to reflect the change of textile industry cost. Using the time series plot of the cotton CF999 futures price data from January 2, 2008, to September 6, 2021, the conclusion is that the original time series is unstable, and the original time series is processed to obtain the cotton futures price return, which is tested by J-B statistical test and ADF test. It is concluded that the cotton futures price return is a stable time series and does not obey the normal distribution. It has the statistical characteristics of peak fat tail and high-order heteroscedasticity. Therefore, considering that the cotton futures price return obeys the T distribution and the generalized error distribution (GED), GARCH models are constructed and the characteristics of cotton price fluctuation are studied by using GARCH, GARCH-M, EGARCH and MS-GARCH models. The results show that the model based on T distribution is better than GED distribution to describe the volatility characteristics of cotton futures price return series; the cotton market is not a “high risk and high return” industry; the return has an anti-leverage effect; the MS(3)-GARCH(1,1) model best fits the data, and the return has the phenomenon of system conversion. The duration of various volatility is different, and the shortest high volatility state is 2. 12 days. Finally, relevant suggestions are put forward.