Championship win probability added (cWPA)

Win probability added credits a player with the change in his team’s chance of winning the game. Championship win probability added — cWPA — asks the same question about a much bigger prize: how much did this play change the team’s chance of winning the World Series?

The mechanics are identical. Take the team’s championship probability before the play, subtract it from the probability after, and assign the difference to the players involved. The only thing that changes is the scope of the outcome being tracked.

That change of scope has a dramatic effect on the numbers, because most plays barely move a championship at all.

A typical regular-season play has a cWPA in the range of 0.01%. A whole game rarely shifts a team’s title odds by more than a fraction of a percent, because 162 games and four playoff rounds stand between April and a championship.

Then October arrives and the scale explodes. A single at-bat in a World Series Game 7 can move championship probability by twenty or thirty percentage points — thousands of times more than a comparable play in May. There is no bigger swing available in the sport, because at that point the game outcome is the championship outcome.

This is why cWPA is usually expressed as a percentage with several decimal places in the regular season, and in whole percentage points in the postseason. The same stat spans several orders of magnitude depending on when the play happened.

cWPA is not a player-evaluation stat, and using it as one would be a mistake — it rewards being on a contender in October more than it rewards playing well.

What it is genuinely good at is ranking moments historically. If you want to know which plays in baseball history mattered most to a championship, cWPA is the tool that answers it. It's the number behind claims like "this was the biggest hit of the last fifty years," and it makes that kind of statement checkable rather than rhetorical.

It also makes an argument about attention: it quantifies exactly how much more a September pennant-race game is worth than an April one, and how much more a Game 7 is worth than either. Fans feel that intuitively. cWPA puts a number on it.

cWPA needs a championship-probability model underneath it, and that model works the same way playoff odds do: Monte Carlo simulation.

To get the title odds at a given moment, the model simulates the remainder of the season and the entire postseason many thousands of times from that exact game state, then counts how often the team ends up champion. Do it before the play and after the play, and the difference is the cWPA.

In the regular season the simulation covers the remaining schedule plus all four playoff rounds. In the postseason the remaining tree is much smaller, which is exactly why individual plays move the number so much — there are far fewer games left to dilute them.

TwentySeven tracks the game-level version of this on every pitch: live win probability, the leverage of each moment, and the plays that actually decided the outcome.

It also runs the season-scale model — Monte Carlo playoff odds — which is the same machinery cWPA is built on. Between the two you can watch a single at-bat move both the game and, in September, the season.