Research on Implicit Force Momentum Based on Granger Causal Test and Bi-LSTM

Firstly, this article establishes a momentum lever model based on processed data. Divide the factors that affect momentum into three categories: serving factors, physiological factors, and technical factors. Then, this article uses Phi coefficient and Spearman rank correlation coefficient to calculate the leverage effect of factors at different stages. We have established an exponential weighted moving average method to measure player momentum, which takes into account continuous scoring and serving advantages. Applying it to the final, visualizing the player’s momentum state and depicting the game process, this article analyzes the momentum lever model based on game flow in the other 30 games to demonstrate the importance of momentum, achieving a 97% matching probability between overall momentum and game results. In addition, this article establishes a Granger causality test model to verify the causal relationship between the momentum of both parties and the swing, as well as the successful running of a player, proving the role of momentum in the game. Thirdly, this article scientifically defines the fluctuations and turning points in the competition, transforming the search for these changes and turning points into a classification problem.