This paper first introduces the use of data mining technology in the development of the traditional culture of martial arts. This paper begins with the optimized FCM algorithm, obtains the fuzzy pattern through the affiliation function, and builds a prediction model using the clustering algorithm of fuzzy time series. The historical development trend of Chinese martial arts traditional culture is predicted using this model. The results show that the process of defuzzification prediction divides the literature on the development of traditional culture of martial arts into five groups, and the five clustering centers are A1: 3055-4693, A2: 5603-6919, A3: 6388-7497, A4: 7984-8150, and A5: 8876-9483. The predicted values of the FCM algorithm model for the literature of the first group to the fifth group are respectively 3735.3, 5374.05, 6351.57, 7048.56, and 9144.31. The average error of prediction is 0.134, 0.062, 0.094, 0.126, and 0.025, respectively. The average prediction accuracies of the predictions are all >85%, and in particular, the accuracy of predicted values for the fifth group reaches 98%. It can be seen that the prediction model proposed in this paper is effective in predicting the historical development trend of the traditional culture of martial arts.