Football Analytics: Win Probability and Win Probability Added

Two weeks ago, I wrote on this blog about the advanced stat “expected points” and “expected points added”: A stat that shows the points that a team will probably score because of the last play.

Today, let’s look at win probability (WP).

If you look only at the name of the stat, you’ll probably think of betting in Las Vegas. How high is the probability of a team winning? That’s the question bookmakers ask themselves and where they base their calculations on.

Winning probability explained

But WP has nothing to do with it. WP only focusses on the game itself. It’s a live stat that calculates the chance of winning the game based solely on score, time left, down & distance and field position – comparable to expected points, but with the difference that you look at the end of the game, not the end of the series.

If a team is down by seven points at the start of the second quarter with a 2nd & 5 at their own 25, the chance of winning the game based on this calculation is 36%.

By reckoning the course of the game and the time left, interceptions and touchdowns will influence the WP more at the end of the game than at the beginning. Also, leading by 20 points in the fourth quarter gives you a higher chance of winning than leading by 20 points at halftime – the Bears might know that ;).

WPA explained

Winning probability added is essentially the difference that one play makes. An interception in our example above would cost the team 0.08 points (8%), but in the fourth quarter with only two points in front and the opossing team being in field goal range, it would much, much more.

Now, this stat doesn’t include all variables, e.g. a team being really, really strong in a certain time frame of the game or a home field advantage when the game is on the line with a narrow for one team or the other.

So why using it?

Well, there are multiple ways to use WP and WPA.

First, the fans can look at WPA and see at first glance what the most important plays of the game where. Where did your team win or lose the game? Just look at the biggest WPA numbers – the deciding plays will show up, either positive or negative.

For coaches it can be a big part in the decision making progress by weighing the options. How much do we win by going for it? How much do we lose by kicking the field goal? When is it better to take a sack than throwing into traffic and risking the interception?

Where to find the stats

Now, there are probably hundreds of win probability models out there, and every formula has its own twists and tweaks. They’re not shown publicly, but you can see the results of these calculations for example on ESPN, where the famous curve changes over the course of the game. I’ve found this tweet of ESPN Stats & Info about the 28-3 Super Bowl:

The numbers I used above are from, a site that is not longer online, but has a good archive of articles about these stats.

Max Sachs is an international Packers fan from Germany. As a part of the Packers Germany, he tries to convince everyone around him to cheer for the greatest franchise on earth - or to start with, for American football in general. You can find him on Twitter @KaesekopfDE (the german translation for cheesehead).