Yesterday Phil Humber threw Major League Baseball's 21st perfect game. Across 9 innings Humber performed his job flawlessly, not allowing 1 member of the Seattle Mariners' team to ever step on base; 27 batters came to the plate, and 27 batters were quickly sent right back to the dugout. It is a truly amazing accomplishment.
But since Mark Buehrle's perfect game in 2009, we here at TargetPoint were curious: could you actually rank baseball's perfect games from the "most" perfect to the "least" perfect? We are, after all, a firm that specializes in data analysis - surely there had to be some formula that could help us sort through the game's greatest pitching performances. Ultimately we settled on a metric that combined the number of strikeouts with an adjusted pitch count to arrive at a ranking: link.
The principle behind it is as follows: there are two separate paths to perfection, 1) strikeout all 27 batters, or 2) force all 27 batters into put-outs. It's improbable that anyone would ever do either of these two things, so our metric makes a compromise. We take the total pitch count and subtract all thrown strikes (multiplying 3 by the number of K's) to create an adjusted pitch count. The adjusted pitch count and the number of strikeouts are normalized so that each is counted equally in our composite score, then the two are added together for our perfection metric. (We should note here that we cannot measure baseball's earliest games as they did not keep detailed in game stats; only the last 18 of 21 can be ranked).
According to our formula, Humber's perfect game of 97 pitches and 9 strikeouts was the 8th "most" perfect game of all time. The top spot remains with Sandy Koufax who in 1965 threw 113 pitches and 14 K's. Here's the database, along with our calculations: link.
Now we do also have an alternative version of our formula that accounts for the opposing team's on base percentage in order to incorporate the quality of the opponent. After all, a perfect game against a basement dweller is less impressive than one against a future playoff team. Interestingly, accounting for the opponent's OBP drop's Humber's ranking down to 13th, as the Mariner's current OBP is the worst of all the opponents in the perfect game database.
So what do you think? How does Humber's game measure up in your opinion? How can we improve this measure? Let us know!