At some point in the fourth quarter, someone in your viewing party said "we're definitely winning this" or "it's over." They were estimating win probability. They were probably wrong about the certainty, and they had no idea how wrong. Win Probability is what that instinct looks like when you run it through actual data.

Win Probability (WP) is the probability — 0% to 100% — that the home team wins a game at any given moment. It updates after every play.

// WHAT GOES INTO THE MODEL

The nflverse Win Probability model (which powers this site) is a logistic regression trained on decades of NFL play-by-play outcomes. The inputs:

• Score differential — who's winning and by how much
• Time remaining — seconds left in the game
• Field position — where the ball is on the field
• Down and distance — 1st-and-10 vs. 4th-and-1 dramatically changes the outlook
• Possession — who has the ball
• Spread — pre-game line as a prior (optional, but improves early-game predictions)

What it does not include: team quality, quarterback skill, weather, or injuries. It's a situation model, not a talent model. A drive in a strong situation looks the same whether Patrick Mahomes or a journeyman backup is taking the snaps.

// READING WIN PROBABILITY: SOME REFERENCE POINTS

// WP SWINGS — WHERE THE GAME WAS ACTUALLY WON

WP on its own is informative. WP swings are where it gets interesting. A WP swing is how much a drive (or play) moved the win probability needle.

A 7-play TD drive that swings WP from 32% to 64% is a +32-point swing. That's a drive that essentially won the game. A TD drive late in a 35-7 game might only swing WP by 1%. Both show up as TDs in the box score.

The drives that move WP the most are the ones that decided the outcome:

// WHAT WP TELLS YOU THAT THE SCORE DOESN'T

The score tells you who's winning. WP tells you how likely they are to stay winning. A team up 10-7 at halftime looks good on the scoreboard. If they've had the ball six times and consistently punted from their own territory, their WP curve tells a different story.

WP also exposes the value of conservative vs. aggressive decisions. Going for it on 4th-and-2 when WP says you're at 45% — and converting — moves the needle enormously. Punting "safely" from that same spot locks you into a lower-probability track. The model doesn't care about conventional football wisdom; it cares about historical outcomes.

// WP IN THE DRIVE TRACKER

The Drive Tracker shows WP_SWING per drive — how much each possession moved the needle. The WP chart visualizes the full game arc. You can see exactly when momentum shifted and which drives were genuinely high-leverage versus garbage time.