Declining the Penalty

Sunday’s getaway-day game between the San Francisco Giants and the host Atlanta Braves featured a play we don’t see too often, and a choice we see even less often. Buster Posey was awarded first base on catcher’s interference by Braves backstop Evan Gattis, but manager Bruce Bochy waved off the award. If Posey had gotten a clean hit, such a move would be obvious, but the actual scenario made it a very interesting decision.

Catcher’s interference is that rare thing in baseball, an umpire’s call that acts like a football penalty because it can be declined. Go to a football analytics site, and you can probably find a learned opinion on whether it’s better to take a five-yard penalty and replay the down, or take the seven yards you gained on the play but expend the down. That’s a pretty close call, and what Bochy faced was also quite close.

The incident came early, top of the first inning. The count was 2-2 on Posey, Hunter Pence just having gone second to third on a wild pitch by Braves starter Alex Wood. Posey ticked Gattis’ glove as he bounced a ball to first base. Freddie Freeman made the play at first unassisted as Pence crossed the plate. Home-plate umpire Alan Porter’s call of catcher’s interference, though, sent Pence back to third (unforced runners don’t advance on CI) and Posey to first.

Bochy thought about this—how much, we can speculate—and came out of the dugout. This is something he needed to do: umpires aren’t required to offer a manager the choice, only to enforce the penalty. If he wants to decline, the manager must take the initiative. Bochy did that, taking the early run while giving up the second out of the inning.

(For a look at the play and its aftermath, and some of the Giants TV team’s commentary on the matter, look at the highlight here at MLB.com.)

Did this make sense? There are a few ways we can look at the matter.

First would be to use Expected Runs, the average amount of runs one would expect a team to score given its base-out situation. I will use 2013 numbers, because I have those available and because this season’s numbers would be a much smaller sample, with its attendant dangers. Runs per team per game last year were 4.166 in all baseball, as opposed to 4.013 this year in the National League, but since Turner Field plays around a 104 Park Factor, this almost perfectly balances out.

With runners at the corners and one out—the catcher’s interference accepted state—Expected Runs are 1.1383. With a run in, nobody on, and two outs—the catcher’s interference declined—Expected Runs are 1.0918. These numbers favor taking the CI, and by nearly five-hundredths of a run. For comparison’s sake, in the same run environment sending someone to steal second with a 78 percent success rate would net you the same gain. Not a game breaker, but a good play, so Bochy’s move looks bad in this light.

Let’s shift the perspective to our favorite toy here at THT: the WPA Inquirer. Dave Studeman announced on Friday that it was back in operation at the site, and better than ever. What a fine stroke of luck. Time to put it through its paces.

The Inquirer calculates Win Expectancy rather than Expected Runs, and is thus a more direct gauge of how a play affects a team’s game chances. I would expect almost no deviation from my Expected Runs calculation, though, since the play came so early that the overall state of the game is still close to an ideal situation for Expected Runs to mirror winning chances.

That turns out not to be the case. Or in layman’s terms, I’m wrong.

The WPA Inquirer can measure run environments in only half-run increments: 4.0, 4.5, and so on. But since I just got through figuring that the environment at Turner that day was effectively 4.166 per team, exactly one-third of the way between 4.0 and 4.5, it’s not tough to interpolate between the two and reach accurate numbers.

For interference accepted, Win Expectancy at 4.0 R/G is 57.0 percent; at 4.5 it’s 56.5; for Turner that day, call it 56.83 percent. For interference declined, the 4.0 Win Expectancy is 57.4 percent; for 4.5 it’s 56.4; in the Turner environment it comes to 57.07 percent. In a higher run environment, the Inquirer would have sided with the Expected runs numbers and taken the catcher’s interference. In today’s pitcher’s era, though, it says decline, by a margin of a quarter-percent of a win.

The Expected Runs table says accept; the WPA Inquirer measuring Win Expectancy says decline. I’d call it a split decision, except that Studeman didn’t resurrect the Inquirer from all the traumas of creating the new website format for it to render split decisions. I would be willing to yield to its verdict … were it not for one little detail.

Both these methods assume average situations underlying them: average pitchers on the mound, average batters coming up. But no mathematical average ever threw a ball or swung a bat. There were specific people doing those jobs, and they deserve at least a look before we close the books.

Due at the plate after Posey was the cleanup hitter, Michael Morse. Morse’s nickname is “The Beast,” and he’s been living up to it this young season, posting a .302/.352/.615 slash-line and a 175 OPS+ through Sunday’s game. You want someone performing that way batting in high-leverage moments, so his excellence can have maximum effect on your winning chances. The interference-accepted situation (first and third, one down) comes in at a Leverage Index around 1.9; the declined state (none on, two gone) is about 0.35.

Numbers like that swing the game your way, hard. Taking away Morse’s chance to clear the bases and keep the inning rolling was a mistake—if you believe in those numbers.

He didn’t have numbers like that last season, putting up a mere 84 OPS+ between Seattle and Baltimore. Lifetime, his OPS+ comes out at 121, which is certainly better, but not nearly as beastly. In fact, it almost exactly matches the season, and career, ERA+ of the pitcher he faced, Alex Wood: 123 and 122 respectively. (Wood’s peripherals do vary from that: better last year, worse this year, nothing decisive.)

If you trust this year’s numbers to reflect what Morse would have done, you should have accepted the catcher’s interference. If you think his true talent level is reflected by his career numbers, or last year’s numbers, you decline and take the run now. But come to think of it, if you believed last year’s numbers, why would you have Morse batting cleanup?

Average numbers are a fine tool, but they are not the end of the analysis. When it’s close either way, who you have performing in the specific situation ought to have a strong impact on what you decide. The whole image and cachet of a cleanup hitter is that you want men on base in front of him so he can drive them home. Morse had been doing that job as well as anyone could want, but when a crunch came, Bochy did not trust him with the job.

It may not be as bad as that. Bochy might have thought the odds favored taking the run so much that it wouldn’t really matter who was coming to bat. The incident may not reveal a sneaking lack of confidence in Morse’s hot first five weeks, even a warranted lack of confidence. And since the Giants won 4-1, it really is not a natural locus of second-guessing.

I did not mean to stir up trouble in the Giants’ clubhouse when I first started looking at this incident. I just wanted to look at an interesting move by a manager, and figure out whether he made the right call. The inquiry just went a little past the numbers where I thought it would end up.

(h/t to FanGraphs’ Wendy Thurm for cluing me in to this play.)