The Value of Getting Aroldis Chapman Off the Mound

When there’s only one or two games on the television every night, every decision a manager makes gets blown up from all directions. Already this postseason, we’ve had the Zach Britton Decision, and Andrew Miller in the Seventh, and Kenley Jansen in the Seventh, and the Max Scherzer Decision. This past weekend begat us one more signature event: the Walk Chris Coghlan Decision. The interim has seen rabid takes defending both sides of the issue.

Despite having occurred four days ago now, the choice by Dodgers manager Dave Roberts to intentionally walk Coghlan — and, consequently, pitch to Miguel Montero — remains relevant for tonight’s Game 4 in Los Angeles. The Dodgers are still playing the Cubs. Dave Roberts is still their manager. There are still decisions for him to make. And there are still opportunities to be second-guessed. For the moment, I’ll attempt to decide whether Roberts’ logic was suspect — or, alternatively, if he made the best choice he could given the information available to him.

To return to that moment: with two outs and two on in the eighth inning of a tied NLCS Game 1 in Chicago, Roberts elected to intentionally walk Chris Coghlan to get to Aroldis Chapman’s spot in the lineup. Pinch-hitter Miguel Montero then stepped in and stroked a grand slam off of Joe Blanton to put the game out of reach for the Dodgers. Immediately, the second-guessing began.

Let’s try to run through the decision-making process up to that moment, because it’s actually a little complicated, and not at all clear-cut.

The easiest thing to say is that the intentional walk alone raised the win expectancy for the Cubs by 4.5 percentage points, and so it was a bad idea. But that’s way too simple.

Against righties for his career, Coghlan has been 8% better than league average, and until this year, Blanton has been league-average or worse against lefties , mostly because he allows home runs at an above-average rate. Roberts had already chosen to walk Jason Heyward, a lefty who’s been 23% better than league average against righties over his career, so the intentional pass to Coghlan could have represented an attempt to protect Blanton from another lefty batter while also serving to get Chapman out of the game.

Again, though, that’s too simple. Just because Coghlan has been 8% better than league average in the past against righties, and Blanton has been 36% worse against lefties than righties since he’s converted to relief, that doesn’t mean they’d continue to be the same going forward.

You have to regress past performance against league average in order to project what they’d do in that case. Using Matt Klaassen’s post on the subject, the math is easy. Coghlan is currently projected to be 16% worse than league average against all hands. If you regress his platoon split, you’d expect him to have a +3%/-7% platoon split, which produces about a .308 wOBA against righties. The result? By all accounts, Coghlan is probably 14% worse than league average against a right-handed pitcher.

Blanton is tougher to figure out, but Jared Cross used arm angles and arsenals to try and project pitcher platoon splits, and the Dodger pitcher doesn’t actually seem like the sort to suffer from a terrible split. For one, he throws over the top, and over-the-top pitchers had smaller platoon splits than side-armers. He also throws 20% curves and 10% changeups, so he’s not just a fastball/slider guy.

In Cross’ full model, Blanton was projected to have almost no platoon split after accounting for all the relevant variables. His Steamer-projected wOBA allowed was .293 against lefties and .278 against righties, so he was a good choice for the Dodgers against either hand of batter.

Is it possible that the intentional walk, then, was actually more about getting Chapman out of the game? That’s tougher to say. We do know that Chapman was worth about 4.5% of win expectancy in his appearances this year all by himself, going by win probability added. If Blanton is going to be nearly the same guy against righties and lefties, and walking Coghlan loses you 4.5% of win expectancy on the field but removes the closer who is worth about that much, then it’s all even, right? And if it’s even, why risk the downside of a free pass?

Because there are other effects that push it a bit past even into “good decision” territory, is why. “We had some left-handers coming up” said Dave Roberts of Joc Pederson and Andrew Toles, due up second and third in the next inning. As great as Chapman is in general (.243 wOBA), he’s kept lefties to a .196 wOBA over his career, a figure worse than the worst qualified hitter in baseball this year. His Steamer projections versus righties (.245 wOBA) and lefties (.204 wOBA) suggest that Roberts was doing the right thing to get Chapman out of the game.

The run expectancy after the walk was still only 0.42, meaning there are many more other universes in which Blanton gets the Dodgers out of the inning and allows his teams to face Hector Rondon with the game on the line in a tie game.

It’s Rondon who chips away at the benefit of getting Chapman out of the game, though. He wasn’t as good as Chapman this year — understatement siren — but if you use his career stats, he’s been worth about 2% of win probability added per appearance. Even if it’s not a .204 wOBA allowed (Chapman against lefties), Rondon’s .294 wOBA against lefties (.255 vs righties), as projected by Steamer, is still pretty good.

We can sum up these two situations given the regressed platoon projections!

With the relevant data presented this way, you can see the chance of a run goes up — not only because of Chapman’s departure from the game but also because of Montero’s big advantage.

Miguel Montero! Given that the only other bats on the bench were righties, he was the obvious replacement for Chapman. Over the course of his career, he’s had a fairly drastic platoon split (16.1%). But once you regress it, you’d expect it to be more muted (11.9%). Montero would be projected to be a .335 wOBA hitter when he stepped to the plate.

That’s much better than Coghlan’s projected .308 number when he stepped to the plate against Blanton. It’s the difference between someone like Adrian Gonzalez (.334 wOBA this year) and Adonis Garcia (.308 wOBA this year) in that situation, handedness considerations aside.

We also haven’t yet gotten to the Blanton that was pitching on Saturday. It’s a version of Blanton who was hanging more sliders than usual. If you average the drop of Blanton slider over the entire appearance, you get something that’s only about a third of an inch less than his seasonal average — but that’s largely because he started throwing some better sliders after the grand slam. Going into the Montero at-bat, he was missing almost an inch of drop on his slider, and the typical slider he was throwing looked like this one that got Javier Baez to fly out — a pitch that was missing almost two inches of drop.

We can’t translate an inch or two of missing drop on a pitch into win-probability numbers, but the fact that Blanton was throwing something less than his optimal slider seems relevant. Especially considering the shape of the two sliders he threw Montero in a row.

The first:

We know what happened with the second:

Is it possible Blanton was throwing cutters and not sliders? And that’s why the pitch didn’t feature the slider’s signature drop? Not really. The velocity spread on all of Blanton’s sliders on Saturday was between 85 and 87 mph — and that’s the precise range into which both of these pitches fall, as well. He did settle down and throw some better sliders later. Up to the point at which Montero stepped to the plate, however, he’d thrown five of his seven sliders with significantly less drop than usual.

Take that anecdotal, situational knowledge, and pair it with the fact that Montero was going to be at least 8% better than Coghlan — with more power — and it trumps the benefit of getting Chapman out of the game. And as much as Blanton was just one strike away from getting through the inning and justifying his manager’s decision to walk Coghlan, the strike he threw Montero to get strike two didn’t really inspire much confidence.

There’s a lot of armchair managing that relies on information gathered after the decision was made, but this one decision didn’t look great, even if you only look at the data that was available in the moment.