Analyzing Strategic Decision-Making in “NFL Game Plan”

The board game NFL Game Plan simulates American football matches , allowing players to make strategic offensive and defensive decisions. This paper presents an analysis of the game’s strategic decision-making landscape using dynamic programming techniques. We implemented a solver to calculate the expected number of wins for all possible game states through backward induction. Optimal strategies were derived for each state by formulating the decision process as a series of matrix payoff games. To improve computational tractability, simplifications were made to certain game mechanics. Simulations of games played using the calculated optimal strategies provided key data insights. Analysis of the play usage distributions revealed a significant underutilization of passing plays, diverging from NFL data. Further categorization of plays into specific routes and gaps exposed additional areas of misalignment between simulated and real-world usage rates. Examining win probability added (WPA) uncovered lower magnitude impacts for individual plays in the simulated games versus the NFL data. Interestingly, plays with extremely low usage rates tended to have the highest magnitude WPA. Score differential distributions from the simulations exhibited lower variance and slightly higher means compared to NFL games, likely stemming from the simplifications applied to the model. By contrasting the simulation outputs with contemporary NFL data, this analysis identified areas where adjustments to NFL Game Plan’s mechanics, probabilities, and play outcomes could better model real-world football strategies and decision-making dynamics. This work lays a foundation for further strategic analysis of NFL Game Plan using the collected data. Potential applications extend to informing actual coaching strategies and decision optimization processes in professional football.