4th Down Decision Simulator

Monte Carlo simulation to evaluate 4th down decisions by simulating thousands of game outcomes. Uses team-specific empirical distributions from NFL play-by-play data.

Game Situation

How It Works

The simulator runs thousands of Monte Carlo simulations for each decision (go for it, punt, field goal). Each simulation plays out the rest of the game using team-specific play distributions from real NFL data.

Configure the game situation and run the simulation

Win Probability by Decision

Go for it
-
Field Goal
-
Punt
-

Simulation Trajectories

Each line represents one simulated game. Shows score differential (positive = winning) over remaining game time.

Go for it
Field Goal
Punt

Distribution of final score differentials across all simulations. Positive values indicate wins.

Go for it
Field Goal
Punt

Methodology

This simulator uses Monte Carlo discrete event simulation to evaluate 4th down decisions. For each option (go for it, punt, field goal), we simulate thousands of game outcomes and compute the win probability.

Simulation Approach

  • Play-by-play sampling: Each play outcome is sampled from empirical distributions derived from real NFL data (2023-2024 seasons via nflverse)
  • Team-specific distributions: Run yards, pass yards, sack rates, fumble rates, and interception rates are computed per team
  • Field goal accuracy: Success probability varies by distance bucket (e.g., ~95% at 20 yards, ~55% at 50+ yards)
  • Game state tracking: Score, time remaining, field position, down & distance are updated after each simulated play
  • Automated play-calling: The simulation uses situational heuristics for subsequent plays (pass-heavy when trailing, run-heavy when protecting a lead)

Key Features

  • Win probability: Computed as (# of simulated wins) / (total simulations)
  • Clock management: Each play consumes realistic game time (5-15 seconds)
  • Turnovers: Fumbles and interceptions occur at team-specific rates with appropriate field position consequences
  • End-game logic: Proper handling of game-ending scenarios, overtime, and desperation situations

Web Demo vs. Full Implementation

Preset scenarios (Close Game, Goal Line, etc.) display results from the full Python simulation using real NFL play-by-play data (3,000 simulations each).

Custom scenarios use a simplified JavaScript approximation for instant feedback in the browser.

For the full implementation with team-specific empirical distributions, see the source code on GitHub.