Sports Reference Blog
Win Probability Calculator
Posted by Mike on November 6, 2013
By now you've likely seen our win probability tweets or the graphs on box score pages. Today I've pushed out a win probability calculator into which you can put any game situation & get a corresponding win probability. For instance, say it's the 4th quarter and you're trailing by 2 with the ball on your own 20 at 2nd & 10 and 29 seconds remaining, the calculator says you have a 3.8% probability of winning (don't tell the Jets that, though).
Looking at this weekend's Bears-Packers game, we can figure out the risk the Bears took in going for it on 4th and 1 on their own 32 in the 4th quarter. If they fail and the Packers take over at that spot, Green Bay will have a 55.9% win probability (ignoring that our model knows nothing about Aaron Rodgers' injury, of course). However, by succeeding Chicago raised their own win probability to 80.6%.
Thanks! but can you also add a column for WP in your game play-by-plays?
Definitely, good thought. Will add that for this year's games.
Thanks; when can we expect that?
Should be up now.
your week one game with the titans and steelers is not scored correctly.
it gives the opening 2 points to the titans and skews the finals score
can you put in win probability before and after, like its done for expected points. That would be greatly appreciated. putting the line at the top of the game page would be useful as well.
Thanks!
I think that someone might be scraping our pbp accounts. :)
nah :)
I basically only analyze my 49ers, occasionally comparing Kap to Wilson or something.
Was that 17-play, 9-some minute drive the Bears had (of which the above situation extended) the longest of the season? Because holy moly.
Some issues I've noticed with the calculator:
4th-and-1 at the opponent's 1-yard line, down by 5 with 8 seconds remaining in the game gives a result of 3.4% WP. This seems pretty darn low; I think gaining one yard on the goal line (which would result in a victory most likely) should be at least be a fifty-fifty shot. Further, if I change the point differential to -4, the WP shoots up to 29.9%.
I understand that since the results are based on a single formula, these types of problems are unavoidable - how can you tell the calculator that a point differential of -4 and -5 in the above situation means EXACTLY the same thing, namely, a score of 6 beats both.
That being said, the calculator is a great tool, and I'm grateful that you guys put it up AND gave us all the formulas that went in to it.
My question is, are you guys planning on making any tweaks to the formula to take into account the situations I described above (goal line situations, etc.)?
Thanks again, love this site.
Well I just had some fun with this, I plugged in every down situation on the Patriots 59 second drive down the field tonight to try to win in Brady's signature way, and that last play at the 18 yard line with 3 seconds left they still had a good 38% chance to win. But alas the Panthers intercept the last pass and improve to 7-3.