Newer people in the field of sports analytics may not be familiar with Trey Causey’s work. He worked with a NFL team and helped develop the “4th Down Bot”. I was reminded of one of his posts today where he lamented how Win Percentage was used… mainly to show off when a last second play change the state of the game.
I’ve tried before creating something similar to the NY Time’s needle chart (see below) to reduce those last minute blowups, but a big difference is that elections so rarely come down to the ‘last play of the day’ aka the last 1% of votes. More often, thanks to good statistical modeling, the result is known long before every vote is counted. Not so in most sports games!
Anyway, the main crux of the argument Trey makes is that Win Percentage needs to be more than a descriptive summary of a game. We usually can figure that out from looking at the scoring. What needs to be done is thinking about many different options a team can make and comparing the possible win percentages against the possible results. 4th Down decisions are so important and most teams have finally figured that out!
I’m curious to see what other people get out of Win Percentage plots; is there something maybe more interesting that I’m missing? How do you use them?