One of the newest scourges of the stat world is win probability, a statistic that takes into account every aspect of the game (score, possession, time remaining, point spread) and spits out a percentage that a team will win or lose a given game after every single play. It's been in the news Monday because, late in the third quarter of Super Bowl LI, the Atlanta Falcons were given anywhere from a 99.5% to 99.9% chance of defeating the New England Patriots. The numbers have been breathlessly repeated as if they're gospel and Tom Brady managed to not only defy the logic of football but the fundamental tenets of mathematics.
It's all nonsense. Don't believe any of it. Like most statistics, win probability is a fine tool to give fans a general idea of a game's situation, a player's value, a team's strength or whatever the specific number is supposed to tell. But it should never be taken literally because sports aren't played on calculators and win probability doesn't account for the countless human factors involved in a comeback - namely Tom Brady.
The idea that Tom Brady is out of any game with 17 minutes remaining in regulation is nonsense. Throw in the Super Bowl stage, the Patriots' familiarity with it, the Falcons newness on it, the pressure of the moment and the fact that the instant Atlanta started to lose a grasp on the game the momentum shift would be far more pronounced than usual, and the whole thing is patently absurd. This is my biggest problem with stats. The human factor isn't taken into account. Game situations with no relevance to each other are compared. The numbers act as if past performance is the only indication of the future. And don't even get me started on "clutch doesn't exist." If you believe that, I have an excuse about destroying a cell phone before a major investigation I'd like to share with you. In this case, clutch very much existed. Show me a number on that.
According to various models, the Falcons had a 99% chance of winning as early as the middle of the third quarter and as late as 5:53 remaining in regulation. The number hit 99.9% (on one metric, but was nearly uniform in all the ones I looked at) after two plays - one with 2:05 remaining in the third quarter that got the Falcons into field-goal range up 28-3 and another with nine minutes left, right before the strip-sack of Matt Ryan. That's assuming 999 wins in 1,000 games. Think on that. Heck, even after the latter play, the probability barely dipped. How much was Tom Brady getting the ball back down 19 points worth? About one percentage point.
Even when the Patriots got the ball back after Atlanta's crippling sack and holding penalty in the final minutes, the stats gave them a 3% chance of winning. Tom Brady is going to lead his team to a touchdown and two-point conversion three out of 100 times with plenty of time left against a gassed defense with the Super Bowl on the line? Come on. The coup de grace: Even when New England had the ball on the one with 1:25 left, they were only given a 24% chance of winning.
We don't need win probability to tell us the Patriots comeback was unexpected and historic. There's no need to put a number on how amazing it was. Our eyes and minds do a fine job of that on their own. The numbers that matter are 25 (the comeback), 28 (the tied overtime score), 31 (the unanswered points scored by the Patriots), 34 (the number of points that won Super Bowl LI) and 5 (the number of rings Tom Brady and Bill Belichick have won together, statistics be damned).