GameStateEdge™

The GameStateEdge™ is PriceArb's primary metric for identifying mispricings during live games. It answers a simple question: given the exact current game state, what does history say should happen?

What It Measures

GameStateEdge™ calculates the historical win probability for a team based on three key factors:

1. Score Differential - How many points the home team is ahead or behind (all calculations use the home team's perspective)
2. Time Remaining - Seconds left in the game
3. Quarter - Which period of the game (Q1-Q4, or overtime)

How It Works

When a game is live, we look up all historical plays that match the current game state within a defined bucket:

• Time bucket: ±2 minutes from current game time
• Score bucket: ±3 points from current score differential

For example, if the home team is down 7 with 8:00 left in Q4, we find all historical plays where:

• Time remaining was between 6:00 and 10:00 in Q4
• Score differential was between -10 and -4

The Calculation

The historical probability $P_h$ is calculated by combining two scenarios:

1. With Possession

When the home team has the ball, we query:

$P_{with} = \frac{\text{wins when home team had possession in this state}}{\text{total games in this state with possession}}$

2. Without Possession

When the away team has the ball, we flip the perspective:

$P_{without} = \frac{\text{wins when away team had possession in this state}}{\text{total games in this state without possession}}$

Combined Probability

Since possession alternates roughly equally during a game, we average both scenarios:

$P_h = \frac{P_{with} + P_{without}}{2}$

Why Possession Matters

Consider two identical game states: home team down 3 with 2:00 left:

• Home team has the ball: They can drive for a field goal or touchdown
• Away team has the ball: Home team needs a stop first

These scenarios have significantly different win probabilities (with our historical database since 1999, we find the home team with possession wins ~38% of the time. For the away team with possession, it is only ~27%). By averaging both possession states, we get a more accurate baseline that accounts for the uncertainty of who currently has the ball.

Data Sources

GameStateEdge™ is calculated using NFL play-by-play data from nflverse, covering every play from 1999 to present. This gives us millions of data points across thousands of games.

You can filter the historical data by time period:

• All Seasons (1999-present): Maximum sample size for reliability
• Last Season (2024): Most recent trends
• This Season (2025): Current season data as it accumulates

Interpreting the Values

GameStateEdge™	Market Price	Interpretation
65%	55%	Market is underpricing by ~10% (potential buy)
45%	55%	Market is overpricing by ~10% (potential fade)
53%	50%	Market is fairly priced (but our data shows slight home field advantage)
N/A	50%	Game hasn't started yet

A value of N/A appears before games start because there's no game state to analyze yet.

Note: Even in a tied game, we find that home teams historically win ~53-56% of the time. We attribute this to home field advantage. This is a direct result and part of our edge calculations.

Edge Strength Guide

Edge Size	Signal Strength
±0-5%	Weak - within normal variance
±5-10%	Moderate - potentially actionable
±10%+	Strong - significant historical divergence

Limitations

• Sample size: Unusual game states (e.g., down 28 with 5 minutes left) may have fewer historical matches
• Context-blind: Doesn't account for team strength, injuries, weather, or other factors
• Historical patterns: Past performance doesn't guarantee future results

GameStateEdge™ works best as one input among many, not as a standalone signal. We recommend combining it with RecordEdge™ for a more complete picture.

---

Back to all edges | View RecordEdge™