Let’s Crown the 2019 Pinch Hitting Champs and Chumps

On September 4, 2019, Ian Miller made his major league debut. With his Twins down 6-0 with only two innings left to play, he replaced Max Kepler to give Kepler a break in the field. Miller’s spot came up eighth in the batting order; he wasn’t guaranteed his first major league plate appearance, but the odds looked good.

But a funny thing happened on the way to his batting debut. Nelson Cruz led off the top of the eighth with a single. Eddie Rosario followed with a home run. The Red Sox went through two pitchers and frittered away a third of their lead — a 6-2 deficit felt less than insurmountable against a shaky Boston team now deep into its bullpen.

So when Miller’s turn to bat came up, with one out and the bases empty in the top of the ninth, Rocco Baldelli made, by some criteria, the best pinch hitting decision of 2019. Rather than have the left-handed Miller face southpaw Darwinzon Hernandez, he brought in Mitch Garver. In addition to being right-handed, Garver was one of the best hitters in baseball last year, full stop. The decision worked: Garver walked, though it didn’t end up mattering — Brandon Workman eventually induced a double play to escape a bases-loaded jam.

Let’s back up. Did I just say the best pinch hitting decision of the year? I did indeed, and I built a framework to get to that conclusion. Here’s how it works: I took each player in baseball’s 2019 single-season line, as well as estimated platoon splits based on regressing their career line, to create rough talent levels for a batter facing each handedness of pitcher (and vice versa from pitchers). From there, I used a lightly modified odds ratio to predict the outcome of the plate appearance that would have pitted the replaced hitter against a given pitcher.

From there, I calculated the plate appearance that actually happened — the pinch hitter against either the original pitcher or a replacement, because the opposing manager gets a say as well. I included a pinch hitting penalty for the new batter. As an example, my model projects Garver as a .396 wOBA hitter as a pinch hitter against an average lefty pitcher. Hernandez is a little worse than average against righties, and so the estimated wOBA of this particular confrontation is .412; think Anthony Rendon’s 2019.

That was a big improvement on having Miller face Hernandez; the biggest improvement of any single plate appearance last year, in fact. This requires a few caveats. First, I excluded pinch hitters who replaced pitchers from the sample, because that’s a different game altogether; the decision to replace a pitcher hinges far more on tradeoffs on the pitching side later in the game than on whether the new hitter is an upgrade from the departing pitcher.

Second, I used 2019 single-season lines to generate baseline talent levels. If a player’s 2019 line was unsustainably poor or excellent, this metric will be unfairly harsh or enthusiastic toward them. I considered using a more regressed estimate of hitting talent, but at some point, we’re just smashing regressions together, and while that may be worth exploring in the future, I decided that this implementation was the cleanest for now.

With those caveats in mind, it’s still a useful framework. Replacing Matt Beaty with Chris Taylor to face Justin Wilson? It comes out to roughly a wash, after considering Wilson’s near-zero platoon splits and the pinch hitting penalty. Replacing Joey Gallo with Danny Santana to face lefty Brian Flynn? Even Santana’s switch hitting can’t rescue that one; it cost the team an estimated 40 points of wOBA — though in fact, Gallo aggravated an injury while batting, which makes this decision far better than it seems.

With an estimate for the cost or benefit of every pinch hitting decision in hand, what should we do with it? Well, I don’t think it’s appropriate to credit or debit hitters for these decisions. After all, they’re not choosing when to pinch hit or when to get replaced. Instead, we can attribute that to the manager. After all, they’re the one making the decision. What’s more, since wOBA is a measure of context-neutral run expectancy, we can convert every change in wOBA due to pinch hitting into some small fraction of a run.

In all, there were 2,052 pinch hit plate appearances last year where the replaced player wasn’t a pitcher. They weren’t divided evenly — the Nationals made only 31 such changes, while the Dodgers checked in at 111. Stick a value on each one and assign them to teams, and a surprising fact emerges: managers simply didn’t add or subtract much value by pinch hitting.

The Rangers were the best team in baseball by this metric of expected runs added by pinch hitting. The Gallo/Santana gaffe aside, they were generally excellent. The average Rangers pinch hitting appearance added a whopping 73 points of wOBA to their expectation, largely driven by batters replacing Jeff Mathis and his jaw-dropping .158/.209/.224 line. Their top seven good decisions, and 16 of their top 20, involved replacing Mathis!

On the other side of the coin, the White Sox didn’t do themselves any favors with replacement batters. On average, they lost 24 points of wOBA every time they chose to pinch hit. Pinch hitting for actual major league hitter Yonder Alonso with minor league catcher Seby Zavala, and losing the platoon advantage while doing so, is an example of the kind of move this system hates. So is replacing Yoán Moncada with Daniel Palka, though that decision was likely more about giving Moncada some rest. The aggregate doesn’t work, though.

Those two teams are the extremes of baseball when it comes to pinch hitting choices. The Rangers added serious firepower (on average) 68 times, while the White Sox probabilistically shot themselves in the foot 72 times. And yet — the gap in run expectancy between those two moves comes out to only 5.8 runs; the Rangers improved their fortunes by 4.3 runs and the White Sox cost themselves 1.5.

Those values seem tiny, and in fact they are. In more than 2,000 plate appearances, according to this metric, the entire majors added 5.8 runs of value. That’s a really small number, an average of only three points of wOBA per plate appearance. Getting extra value from pinch hitting is hard! More than half of the teams in baseball, in fact, lost value by pinch hitting:

It wasn’t much value; the Yankees were basically breakeven, and only two teams lost even a run of value with their pinch hitting. And this isn’t intended as an exact declaration of value, for reasons I’ll outline shortly. But it’s a good first sketch of how important (or unimportant) replacing one major league caliber batter with another to gain small advantages is. In fact, given that I haven’t regressed batting lines at all, it probably overestimates the effect.

Quickly, now: why isn’t this a great measure of pinch hitting value? First, it ignores leverage. Subbing out a star for a scrub in a blowout (Palka replacing Moncada, say) isn’t as bad of a decision as it looks because the sacrificed run value is close to meaningless. I also didn’t consider the base/out state; replacing a strikeout magnet with a contact ace is more valuable than the context-neutral numbers suggest if a groundout scores a run.

Those limitations aside, this is still a nice first cut. Could we fold in leverage and context to slightly improve our answers? Certainly. But even without doing so, it’s clear that pinch hitting for a hitter with another hitter simply doesn’t move the needle much.

It doesn’t move the needle much, that is, as long as you’re not replacing specifically Jeff Mathis. Mathis, the archetypical all-glove, no-bat catcher, is singlehandedly keeping the pinch hitting industry afloat. Remember above when I said that every pinch hitting appearance, all added together, added about six runs of value to offenses in 2019? That’s a little misleading, as aggregates are wont to be.

Mathis was replaced by a pinch hitter 33 times last year. Those 33 pinch hitters added a total of three runs of expected value to the Rangers’ offense, nearly a 10th of a run per time at the plate. That’s right — half of the overall value, across the entire major leagues, of pinch hitting for position players, came from replacing Jeff Mathis.

I’m not really sure if this is a credit to the Rangers. It doesn’t take a rocket scientist to pinch hit for Mathis; he was never out there for his bat, and 2019 was his worst year by far. Indeed, each of his triple-slash stats was the lowest he’s ever recorded in a season of 100 or more PA, and he did that in a hitter’s park in the year of the home run.

But useful or not, what an amazing fact! Jeff Mathis, all by himself, made pinch hitting valuable last year. The Rangers, without the plate appearances in which they replaced Mathis, would have been the third-best team in terms of adding value with pinch hitting. But the Mathis plate appearances by themselves added more value than any other team added in total.

The average pinch hitter replacing Mathis increased the expected value of that plate appearance by 105 points of wOBA. That’s roughly the difference between Juan Soto and Yolmer Sánchez, between one of the best hitters in baseball and one of the worst. And the Rangers accomplished that with whatever they had lying around on the bench!

When baseball returns, the Rangers will have a big edge in the adding-value-by-pinch-hitting department. Mathis is still around, and we projected him for roughly 200 plate appearances when we expected a full baseball season. That’s 50-ish games of pinch hitting opportunities, and there’s almost never a reason to let Mathis take his full complement of at-bats.

For the rest of baseball, pinch hitting for a position player will be a difficult way to add value. As a manager, armchair or otherwise, there’s a lot of temptation to get fancy. Unless you have a handy Jeff Mathis to replace, though, you’re probably rearranging deck chairs when you swap in one batter for another.