Wait, the Tigers Pinch-Hit for Riley Greene?

On Wednesday afternoon, I briefly thought that A.J. Hinch had lost his mind. I really don’t know how else to explain it. With runners on the corners and one out in the top of the seventh inning of Game 2 of the Wild Card Series between the Tigers and the Guardians — which Cleveland won, 6-1 — Detroit had Riley Greene, its best hitter, at the plate with a chance to break a 1-1 tie. The Guardians went to the bullpen, bringing in lefty Tim Herrin. Herrin, a 6-foot-6 curveball specialist, figured to be a tough matchup for Greene; he’s been lights out against same-handed batters throughout his career. But then Hinch made a surprising call to the bench. He pulled Greene back and pinch-hit with Jahmai Jones – and now here I am writing this article.

Jones had one key thing going for him here: Like Inigo Montoya, he is not left-handed. He’s also hit lefties much better than righties in his brief major league career, and in his minor league career, too. Greene, on the other hand, is a poor left-on-left hitter. So you can at least see where Hinch’s decision was coming from. I want to give this kind of shocking decision the full consideration it deserves before just laughing it out of the building – after all, what if it was the right call? So let’s do all the math to get an idea of what Hinch was giving up, and what he was getting.

To model pitcher-against-batter outcomes, I first took projections for both players, the granular ones that consider specific outcomes. I also calculated platoon splits for each player by taking their observed career splits and regressing them toward league average based on sample size. I put those two projections – hitter and pitcher – into a modified log5 formula and used it to predict the likelihood of each possible outcome of a plate appearance. Then I applied those outcomes to the game state when Greene’s spot came up in the lineup.

That’s a lot of explanation jammed into one paragraph, so I think an example is in order. Let’s say that the Jones-Herrin confrontation results in a single 25% of the time, a deep fly ball 25% of the time, a strikeout 25% of the time, and a walk 25% of the time. Those are nowhere near reasonable, of course, but just an example. A single would mean runners on first and second (at least) and a 2-1 lead, for a win probability of 73.4%. A deep sacrifice fly? That would get the Tigers to 66.6%. A strikeout? 50.1%. Walk? 65.7%. Average those four probabilities, and the Tigers come out with a 64% chance of winning the game. There are more than four possible outcomes, of course, but this process is how I turn outcomes into win probabilities.

Let’s start with the Greene-Herrin confrontation. Greene projects as a more or less 100 wRC+ hitter against lefties, with a bit of a strikeout problem but plenty of power. But few lefties project for a 100 wRC+ (that’s overall league average, all matchups) when they’re facing left-handed pitching. That’s because Greene is a very good baseball player. On the other hand, Herrin projects meaningfully better against lefties than righties. But he’s slightly worse against left-handed batters than the average left-handed pitcher, because he’s got a slightly worse projection than the average lefty pitcher in the first place. Put them together, and Greene projected for the following outcomes against Herrin (the fly out and groundout categories bucket possibilities for RBI and double plays):

Sum all of that up, and the Tigers projected to win 59.4% of the time, at least by my math, with Greene at the plate. That’s some blend of big leads, little leads, and ties, with enough multi-run leads to make up for the strikeouts. In slash line terms, it’s .250/.332/.431, a perfectly respectable line. Not bad for a lefty-lefty matchup for a guy like Greene with a strikeout problem. But Herrin doesn’t strike many batters out and does walk a ton, so you can see how this method comes to its conclusions.

In comparison, switching to Jones meant attacking Herrin with a slightly weaker hitter, but one in a more favorable situation. Jones has been slightly above average overall in his career, and he’s been so good against lefties that even with only a small sample of major league at-bats to his name, he projects as a true lefty killer. And Herrin is no great shakes against righties. My same model thinks that Jones projected to absolutely wallop Herrin, to the tune of a .299/.428/.472 batting line. Herrin is kind of wild, Jones can take a walk, and those tendencies tend to increase when hitters have the platoon advantage. Here’s the grid of Jones-Herrin results:

Add those all up, and the Tigers projected to win 60.9% of the time when Jones batted. That’s a huge advantage over the numbers with Greene. It might even seem a little low of an estimate given that absurd slash line, but many of Jones’ positive results are walks, the least impactful way of reaching base here. But either way, check it out: Hinch’s decision was borne out by the math. He inserted a pinch-hitter because that pinch-hitter projected to give the Tigers a better chance of winning than the guy who was previously batting. Adding 1.5 percentage points of win probability with a pinch-hitter is a meaningful gain, verifiably so.

Well, it would be borne out by the math if we treated this plate appearance in isolation. But this is a Ben Clemens article, a FanGraphs postseason decision-making article. We’re not leaving anything in isolation here. The first consideration: the pinch-hitting penalty. Historically, pinch-hitters have produced a batting line of about 24 points of wOBA lower than those same hitters when batting in the starting lineup. Render that into our format, and that would erase around half of the advantage the Tigers got. Not bad, but not great either.

My next consideration: Could the Tigers have saved Jones for a better spot? After all, he’s an improvement here, but he’s an improvement from an already-high starting point. Greene isn’t a weak hitter, even in a bad matchup, so pinch-hitting for him doesn’t excite me when there are plenty of other lefties in the Detroit lineup to replace. Parker Meadows, Kerry Carpenter, Zach McKinstry: In a close game, the Guardians could have used the three other lefties in their bullpen to attack those weak-ish links. Sure, this was a high-leverage spot, but the game was unlikely to be short on those in the later innings (foreshadowing level: 1,000). This is a mark against using Jones, in my estimation.

An even bigger problem: This spot in the order would bat again fairly often. By either extending the inning with a hit, or perpetuating the tie with an out, this spot in the order was likely to come up again regardless of outcome. And indeed, the Tigers had to first use a defensive substitute for Jones and then pinch-hit for that substitute with Justyn-Henry Malloy. I’d probably rather have Greene bat there, you know? When you pinch-hit for a good hitter, you’re giving up future access to that better hitter, so you better be certain of your advantage.

I do think that this was the highest-leverage plate appearance Hinch was likely to get to pinch-hit with Jones. A runner on third and less than two outs in a tie game? That’s about as good as it gets. But why not wait a batter and then pinch-hit for Wenceel Pérez, standing in the on-deck circle, if Greene didn’t blow the game wide open? I’d rather use my best hitters than get the best platoon matchups, and a Jones-Greene pairing greatly outperforms a Jones-Pérez duo in my estimation. Greene’s propensity to hit into double plays might have played into Hinch’s mind, but Jones hit into them nearly as often on a rate basis, and neither had looked particularly GIDP-prone before this year.

Hinch essentially said that he did the math differently in his postgame press conference. “We knew what was at the back end of the game,” he said, referring to the elite Cleveland bullpen, “which is why we were aggressive.” I take his point here. In essence, he discarded most of the long-game math and just tried to get his best matchup in the game against the worst reliever he was likely to see, and in a big spot to boot. After walking through the math, I can understand why Hinch made his decision. He had one extremely valuable pinch-hitting chip to play, and he decided that he was going to play it the first time he got the chance. Was it a bummer that it happened to be Greene’s spot? Sure. But he obviously would have done the same with the other lefties.

I guess, in the end, I’m surprised to say that I don’t hate the decision to pinch-hit for Greene. I wouldn’t have done it. I’m not willing to discard those long-game numbers, because even if the Guardians bullpen is very good, I think that it can be beaten, and that I’d like to give my best hitters the most possible chances to do so. But the expected gains were certainly there. The move that Hinch made unquestionably increased the Tigers’ chances of getting a good outcome in that plate appearance.

They didn’t, of course. Jones struck out. None of it mattered; the Guardians blew the game wide open with five runs in the bottom of the eighth after chewing through Detroit’s best relievers. It all came to nothing in the end. But I still found the decision fascinating. It looked very wrong on the surface. It turned out to be a very good decision within the framework of a single plate appearance after I ran the numbers. A number of other considerations related to the lost opportunity of future at-bats pulled it back to being within the margin of error, all before the opposing team washed away any evidence of the decision with a deluge of runs. Playoff baseball is delightful.