What Was Matt Williams Thinking? Maybe This

As you’ve probably heard by now, Matt Williams had a bad seventh inning on Tuesday night. Dave Cameron took Williams’ bullpen maneuvers apart nicely in his Just a Bit Outside column at Fox Sports. At the risk of putting words in his mouth, he said that Williams leaving Matt Thornton in the game to face Buster Posey was inexplicable.

It’s not quite that. There is a certain logic to what Williams tried. It may not be good logic, but looked at the right way, you can sorta understand what he was doing.

Now, this post is not meant as a rebuttal of Cameron’s piece. If I got into a saber-war with Dave Cameron, I’d fully expect to lose. I don’t like to lose, so the heck with that. Consider this more of a supplement to his work, an added angle to what he’s already provided.

To quickly recap, southpaw reliever Thornton began the seventh for the Washington Nationals, facing the two left-handed batters at the top of the Giants’ order. Two batters later, it was a man on first and one out, with righty Buster Posey at the plate. Thornton stayed in, Buster singled, Thornton got hooked for a right-handed reliever who was not closer Drew Storen or bullpen ace Tyler Clippard, and the Giants eventually scored the run that would win them the series.

Cameron cited Posey’s outstanding career numbers against lefties as ample cause to get the southpaw out of there, and that’s a compelling argument. Of course, it’s not like Posey’s a slouch facing righties, but with a career wOBA of .356 versus right-handers against a .412 against lefties, he’s obviously been more dangerous with the platoon advantage.

But that’s been less true this season, and there begins the justification, or rationalization, of Williams’ decision. If Williams was going by season numbers, rather than career stats, the problem he wanted Matt Thornton to solve becomes less daunting, if longer than just the matter of Buster Posey.

To start, note that Matt Thornton being a lefty isn’t as much of a problem as it may appear. Left-handed pitchers naturally do worse against righties: the platoon split this season was 32 points of wOBA, for a .322 L-vs-R and a .290 L-vs-L. Thornton’s career split is less than half that, 13 points, and this year it was a smidgen thinner at nine points, .272 vs. .263.

So that’s less bad for Thornton, which isn’t the same thing as good. But if you add in the 2014 platoon wOBA splits for the three batters coming up in the order, you will begin to see what Williams may have seen.

Batter	Career vs. L	Career vs. R	2014 vs. L	2014 vs. R
Buster Posey	.412	.356	.382	.366
Hunter Pence	.367	.346	.340	.341
Pablo Sandoval	.309	.363	.247	.359

Posey’s split is around a quarter of his career mark. Pence’s basically doesn’t exist. And then there’s the 2014 split for the switch-hitting Sandoval, which you probably anticipated from having the numbers jammed in your face during the TV broadcast of every game in the Giants/Nationals series.

Through that lens, leaving in the left-hander for Posey and Pence wasn’t as bad a risk as it once had been, and sticking it out would gain the reward of attacking Pablo Sandoval’s glaring weakness. After Sandoval would come left-hander Brandon Belt, but his platoon splits, both season and career, are mildly reversed.

Add in a further consideration. Williams had already used two relievers before Thornton entered in the seventh. He was three days removed from a game that had gone 18 innings and drained his bullpen. He may have been more than usually attuned to the virtues of having his relievers face more than a couple of batters apiece. Put this all together, and you have the case for Matt Williams let Matt Thornton stay in to face Buster Posey.

Does that make it a good case? Probably not.

First, depending on season splits is pretty iffy. Career splits are considered a better measure of what you can expect in any upcoming at-bat. Possibly the Nationals’ analytics department disagrees, but the sabermetric consensus lies elsewhere. Going by that consensus, there’d be nothing to gain from balancing Posey’s splits against Sandoval’s.

Second, as noted before, just because Thornton would be a better choice than expected in that situation doesn’t mean he’d be the best choice. Tyler Clippard was 81 points of wOBA better against righties this year than Thornton, and just 29 points worse facing lefties. Have them both face Posey-Pence-Sandoval, and you’d expect the better results from Clippard, even with the batters’ splits at their most anti-Clippard single-season slant.

This doesn’t even count Williams’ other relief options, notably Dave Cameron’s own first-guess about using Stephen Strasburg in a long relief role. With three pitchers already burned, this was looking better than it had six innings before. (It’s amusing how cutting-edge analysts are telling managers to be more old-school in their flexible use of starters in relief roles. Next thing you know, they’ll want to bring back the Williams Shift.)

Third, how committed could Matt Williams have been to this putative plan if he bailed out on it so rapidly? It took getting through two dicey AB’s to reach the Sandoval butter zone, and he gave up after one. If he thought he had the right process, shouldn’t he have gutted his way through an adverse small-sample result? He ended up with the worst of both worlds, a lefty pitching to Posey and a righty pitching to Sandoval, made still worse that it wasn’t Clippard or Storen or even Strasburg facing the Panda.

So no, there probably wasn’t a sketchy but plausible plan based on single-season platoon samples that misfired. There was just a manager feeling his way thorough a tight spot in a must-win game, who is now calling for tee times a lot earlier than he had hoped. Would it sting less, for Nationals fans or for himself, if he had had a plan? Probably not, or not by much.

We have the consolation, though, that I avoided that saber-war with Dave Cameron. Talk about your no-win scenario.