A Study on the Correlation of Momentum Based on Binary Logistic Regression Algorithm and Monte Carlo Simulation

This paper explores the correlation between momentum and scoring trends using a combination of binary logistic regression and Monte Carlo simulation. The real-time momentum of the game is computed, and the accuracy of the model's results is verified using t-tests. The innovative fusion of binarization and logistic regression, utilizing the Sigmoid equation to calculate the weights of influencing factors, enables the calculation of game momentum. Additionally, the hypothesis that scoring trends are random is overturned through hypothesis testing. Empirical comprehensive tests demonstrate the method's advantages in determining scoring trends. By leveraging the strengths of polynomial fitting and the binary method, the correlation between momentum and scoring trends is established at 0.80. In a reverse test, this paper conducts a Monte Carlo random simulation and performs correlation tests between 1000 sets of random numbers and scores. Pearson correlation coefficients are computed, revealing that none of the coefficient sets have an r-value greater than 0.5 with a p-value less than 0.05.