CBA is a sports event that allows fans to enjoy themselves and players to give full play, and traditional Chinese cultural values have a profound influence on it. This paper takes the 100 sets of historical rating data of the fourteen teams in CBA league as the basic basis, firstly, we simply deal with the 100 sets of historical rating data and use Excel function formula to find out the mean, extreme deviation and variance of each team, then we carry out SAS normal test, and we find that except for the very few data with large deviation, the historical rating data satisfy the normal distribution. Through the outlier algorithm to screen the values, compare the confidence intervals as well as carry out hypothesis testing, to objectively and scientifically explore the probability of each team winning the championship in the CBA league. Compare the probability of winning the championship of these fourteen teams and predict the top four teams in the CBA league to ensure that the prediction results are as reasonable as possible. With the help of hierarchical analysis to qualitatively analyze the level of each team, and then through cluster analysis to compare these data, and combined with the trend of the development of the world's basketball movement, the use of multiple regression and SPSS to analyze the level of the team's factors, in-depth thinking about the league, a more reasonable to give a more scientific to improve the probability of the team's winning the championship, and to promote better development of the basketball movement.