Patrick Corbin, Reliever

In Game 1 of the World Series, the Nationals found themselves with an interesting decision. They were up three runs on the Astros, but Max Scherzer had labored mightily to hold Houston to two runs. After five innings, he’d thrown 112 pitches. He wouldn’t be heading back out for the sixth.

The Nationals don’t really trust their bullpen. Sure, they could get innings from Sean Doolittle and Daniel Hudson, but they had four innings to cover. Tanner Rainey? Break glass in case of emergency only, and that still leaves an inning. Here is a list of all the relievers the Nationals had used this postseason (as of Game 1) who aren’t Rainey, Doolittle, or Hudson:

Yeesh. Strickland wasn’t on the World Series roster, Suero only made the NLCS roster as a replacement during Daniel Hudson’s paternity leave, and Fernando Rodney — well, we all know the Fernando Rodney Experience. Rainey is no great shakes, either — he was fine this season but uninspiring. Javy Guerra would later pitch an inning in Game 2, but the cupboard was pretty bare.

But Dave Martinez had an answer. Patrick Corbin stepped to the mound to start the sixth. He did his job admirably, striking out two Astros on his way to a scoreless inning, the only blemish a single to Yordan Alvarez. And then he was gone, replaced by Rainey.

Rainey wasn’t good (his four batters: homer, strikeout, walk, walk), but Hudson and Doolittle held on, recording four outs each as the Nationals won 5-4. Martinez squeezed just enough out of the bullpen to make it through the game. Corbin’s inning loomed large: the final margin was one run, and while it’s not automatic that a lesser pitcher would have given up a run in his place, his inning was important.

But the Nationals will pay a cost for Corbin’s appearance. Game 3 is tonight, and Corbin, Washington’s third starter, won’t be starting. Aníbal Sánchez will go instead, and while Sánchez is a nice option, he’s no Patrick Corbin. Corbin’s not gone forever — he’s scheduled to start Game 4 — but delaying his appearance by a day has knock-on effects. Sánchez will be available on regular rest to start a potential Game 7; Corbin would be on short rest that day.

In essence, the team made a trade down from a Corbin start to a Sánchez start. That’s a real cost; our forecasts like Sánchez a lot less than Corbin (1.5 runs of ERA less, to be exact), though after a two-season run with a 3.39 ERA and 4.07 FIP, maybe the projections are low on Sánchez. But he’s still not better than Corbin, whatever you think of the forecasts.

To think about the tradeoff, I mapped out how the Nationals can get to the 63 innings worth of pitching they’d need to compile to win a seven-game series. Before Game 1, that might look like this:

This is an aggressive plan: 14 innings from Scherzer requires a six-inning start, a seven-inning start, and an inning of relief in a potential deciding game. Seven from each of their two top relievers entails appearances of more than an inning and one rest day in DC. This plan requires absolutely everything to go right.

But everything wasn’t going right. Scherzer only went five innings — there was already a shortfall. Being off-plan in the first game of the series isn’t great. So now it was time for a new plan:

The 1.1 innings lost from a shorter Scherzer start are a sunk cost. They had to be replaced; the only question was how. One option would be to extend Rainey and Rodney; two extra outs each would make up for it. That would mean putting Rodney in the game right away, however, and leaning on the rest of the bullpen if he didn’t have it that night.

The other option was to toy with Sánchez and Corbin’s starts to get extra outs from them. The Nationals have used Corbin in relief, but not Sánchez. Let’s take them as the authorities on this and assume Sánchez can’t relieve.

This puts the Nationals in a bind. They clearly believe that an inning pitched by Sánchez is more valuable than one from the stub end of their bullpen, but with only seven games and three great starters, he was only due one start anyway. To get him an extra turn, you’d have to displace one of three better starters.

You can’t displace Scherzer and Strasburg. For one thing, Scherzer had already pitched, and for another — no. That leaves Corbin, and while it might be wild to prefer Sánchez to Corbin to start one game, you’re at least getting a tradeoff here. Swap a Corbin start for a Sánchez start, and you can have Corbin pitch in relief two times — maybe more if you’re using him as a LOOGY.

The plan I presented above squeezes an extra inning out of the Corbin-Sánchez duo, which takes an inning from the bullpen. On the surface, that’s an easy tradeoff to evaluate. Take the difference between Sánchez and and Corbin (let’s be conservative and call it a run of ERA) and apply it over the 5.1 innings Sánchez picks up — that works out to .59 extra runs. Take back an inning of the difference between Corbin and Rainey — call that 1.5 runs of ERA. That only works out to .17 runs. The surface-level math just doesn’t add up. Why not use a reliever and keep Corbin starting regularly?

There’s one thing I haven’t mentioned. Corbin wasn’t throwing a random inning. He was throwing an inning that was one of three the Nationals needed to win a game in the World Series. A three run lead is reasonably safe, but it’s the Astros, and it’s the Nationals bullpen — no lead is safe.

During the regular season, leverage index is a great way to think about how important the situation is. Leverage index measures how important an at-bat is in terms of its ability to swing the outcome of the game, on a scale where one is an average situation. When Corbin took the mound, the leverage index was — well, it was 1.07. Not a very important spot!

Leverage index is a great statistic for a game in, say, August. You use your best reliever in high-leverage spots because there are a lot of innings to soak up over a year of baseball, and you want your most talented pitchers to pitch the most pivotal innings. Use a high-octane reliever to get the first three outs of a game that ends 10-2 in your favor, and you’ve wasted an inning of run prevention that could have been filled by someone off the Triple-A shuttle.

But leverage is deceptive in the playoffs. In Game 7, leverage might as well not exist. The ninth inning and the first inning are equally important, regardless of score, because there are no more games after this, no bullets to save. The whole game is about who can get more runs over 27 outs, and the order in which you line your pitchers up doesn’t matter.

Of course, that’s close to being true for Game 6 too. The team that’s facing elimination has to treat it as their last game. They’re not going to save their good relievers because they’re down — there’s no point.

In the playoffs, most every inning is pivotal. You don’t throw in a mop-up man when you’re down two runs in the sixth, because each game is precious. Important innings? Every inning is life or death! Use your best pitchers and hope to scrape out a win.

Except that’s not always true. The Nationals had three mop-up-esque innings in Wednesday night’s game after scoring six runs in the top of the seventh. Those were largely window dressing, and Washington used Rodney, Rainey, and Guerra to get the nine outs they needed.

But low-leverage games are few and far between. Using leverage index as your guidepost in the playoffs is an insufficient plan. In the Wild Card game, for example, Daniel Hudson was worth more Win Probability Added than any other Washington pitcher. But he threw only one inning, while Stephen Strasburg threw three scoreless. Use WPA (and leverage, a cousin of WPA), and you might think that you should have saved Strasburg, the unquestionably better pitcher, for Hudson’s inning. But Strasburg’s contribution was clearly more important. The stakes weren’t higher in Hudson’s inning — they were high all game from start to finish.

In the end, the Nationals’ plan to use Corbin wasn’t what I would have done. They cut the number of innings he would throw in the series, and they didn’t even get a lot in return — a single inning in a three-run game.

Just because I wouldn’t have done it, however, doesn’t mean I don’t understand the motivation. The actual run cost is small — if you believe my estimate, it cost the Nationals less than half a run of expected value to throw Corbin there. And the potential psychological gains, though unquantifiable, were immense.

Games aren’t preordained, but they can feel like it, and it’s at least somewhat reasonable that the mental toll on the Nationals, contingent on them blowing that three-run lead, would have more impact on the series than the half run of value the Nationals gave up. Martinez surely weighed those in his mind; the uncertain future benefit on one side, putting his foot down and claiming Game 1 on the other.

The difference doesn’t show up in the mathematical odds of winning the game. Despite that, it’s a reasonable thing to be running through Martinez’s mind. For half a run of expected value over a seven game series, he could push the Astros’ odds of winning the game in front of him ever so slightly lower, and he decided that the present value of the equation was worth it.

One postscript: there was another way it could have worked out. Corbin threw 21 pitches. What if he’d thrown 15? What if he’d thrown eight? He could be back for Game 3 with no ill effects. Take this into account, and the tradeoff was even closer. Martinez chose a great spot to get Corbin in the game, one with low and potentially no downside. It was, in my opinion, a defensible managerial move in a postseason where Martinez has been pushing all the right buttons with his pitching staff (those intentional walks being another matter). The cost part of it starts tonight, but it won’t truly be felt, if at all, until Game 7.