This paper mainly studies the important role of the percentage prediction of the number of users and the number of word-filling attempts of Wordle game in the future development strategy. The purpose of this article is to research and develop two mathematical models of Wordle game, one of which is the prediction model of user volume and the other is the percentage prediction model of the number of word-filling attempts. The fitting effect is very good, and the result is R2=0.984. By quantifying specific letters at different positions in a word, it is found that the proportion of difficult modes has a direct impact on specific positions and specific letters. Secondly, for the second model, the proportion of each corresponding variable is obtained through comprehensive evaluation and analysis by using the mean expectation model for the percentage of the predicted number of attempts. The average variance formula is used by calculating the average of the relevant percentage of each characteristic value under the corresponding number of attempts. The relevant percentage distribution results of "EERIE" (1, 2, 3, 4, 5, 6, X) are (1.20%, 16.01%, 28.31%, 27.83%, 18.16%, 5.81%). Through the analysis of variance, the P value is less than 0.05, and the prediction result of the model has high confidence.