Mired in injuries that have depleted their offense, the Mets were no-hit by the Giants’ Chris Heston on Tuesday. Adding to the frustration, the 9th inning was punctuated by some notably poor calls from home plate umpire Rob Drake.
Namely, this 0-2 curve to Danny Muno, which had an 11.7% strike probability:
Not that we should be too surprised—Buster Posey is an excellent framer. Baseball Prospectus’ comprehensive new pitch-framing stat, CSAA, rates Posey as being worth 20+ framing runs per 7,000 chances for four years running. So his subtle movements in showing off pitches provide the Giants with an additional couple of wins each season. Still, I wondered whether these particular called strikes went a step beyond a typical Posey game. I took note of his 2015 BP framing stats pre-game and post-game for comparison:
|Period||Extra Strikes||Framing Runs||Framing Runs per 7000|
In this single game, his extra strikes/framing runs totals increased by 18.2%, and his rate of framing runs per 7,000 chances jumped up nearly 3 runs (+15.7%). These seem like pretty big jumps, particularly for CSAA, which regresses within subjects (reducing the effect of players’ outlier performances) and stabilizes very quickly. To quote Jonathan Judge, Harry Pavlidis, and Dan Brooks:
After only 10% of the season (about three weeks) a catcher’s 2014 CSAA sports a .81 correlation to his final number. After 30% of the season (about 2 months), the correlation is over .9.
Tuesday’s game was San Francisco’s 59th, so they had played through 36.4% of the season. Shouldn’t framing numbers be mostly immune to large jumps at this point? What can account for this seemingly large increase?
I have to wonder whether the no-hitter played a role in the bounty of extra strikes snatched by Posey. It would be a fair assumption; just like the strike zone expands in three-ball counts and contracts in two-strike counts, as umpires are hesitant to make PA-defining calls, a no-hitter could invite similar fears. Don’t just hand the close calls to the pitcher/defense, and risk becoming another Jim Joyce—a respected umpire who is remembered for blowing the final call of a perfect game.
I also think another factor is at play here. In the CSAA article, Judge/Pavlidis/Brooks write that there’s a “slight, but persistent home-team bias in strike calls.” Well, consider this note from Marc Carig:
With fans wielding their cell phones to record history, the Mets quietly shuffled off their home field as a crowd of 23,155 erupted as if the Giants never moved out of Harlem.
Giants-Mets games often see a large contingent of opposing fans at Citi Field (which will likely continue to grow now that Joe Panik is a Giants staple). I assume that the home-field advantage in CSAA is principally based on crowd support and noise, which would again evoke a natural response from umpires—it’s uncomfortable to disappoint and anger ~40,000 people who are closely watching and hanging on your decisions. You could argue that the Giants had a large share of the field advantage, especially by the 9th inning—yet, Posey will have gotten additional credit for the difficulty of framing on the road.
Of course, there’s no tidy way to control for the distribution of fans in a crowd. One could develop a proxy for a team’s popularity in an opposing city. Controlling for milestones, though, is much more practical. In the context of BP’s generalized linear mixed model, a no-hitter variable would be a fixed effect—a binary value, indicating whether or not a no-hitter is intact—and certainly it would be utilized far less often than the model’s current fixed effects: PITCHf/x probability of the pitch being a strike, batter/pitcher handedness, and whether the catcher’s team was home or away. But, if no-hitters do play a role in producing large swings in called strikes, then maybe it’s a worthwhile factor to control for. Maybe stolen strikes spike in complete game shutouts, no-hitters, and perfect games, where emotions could play a heightened role in umpires’ calls.