On the ultimate day of the 2022 regular season, the Detroit Tigers faced the playoff-bound Seattle Mariners, and the game was tied going into the bottom of the ninth at 4-4.
Gregory Soto was summoned from the bullpen in an attempt by manager AJ Hinch to hold the score static and take the game into extra innings. Instead Soto, who’d been pretty reliable all season, got a quick flyout but then walked the next batter. A comebacker to Soto looked to begin an easy inning-ending double play, but instead Soto threw wild to second and the winning run ended up on third. After another walk, Ty France hit a single to right that brought home the winning run, ended the regular season, and dropped Soto’s won-loss record to 2-11.
Since won-loss records for relief pitchers — especially closers — can be a bit misleading, let’s look at his saves. That’s how to measure a post-Eckersley-closer’s value, right? Soto amassed 30 saves in 33 opportunities, which is around 91% efficient. That looks pretty solid.
But, the final game of the season wasn’t a save situation; he came into a tie game to start a “clean” ninth inning, and the hope was for the Tigers to score in the tenth. Then, someone else would come in to the save the Tiger victory. (Andrew Chafin had three saves to that point, Joe Jiménez and Michael Fulmer each had a pair, and Will Vest had one.)
I felt like I’d seen this sort of thing before, after following so many Tiger games this season: if it’s a save situation, Soto was great. If it wasn’t, batten-down the hatches and hope for the best. This also seemed like something I’d been through as a Tiger fan with previous closers like José “Papa Grande” Valverde and, if you want to go a little farther back, Todd “Roller Coaster” Jones.
Was this the case? Was Soto markedly worse in non-save situations? And, was this unusual? Naturally, Baseball Reference had the data, and it was up to me to figure it out.
As mentioned, Baseball Reference was the source for all the data. I looked through all of the box scores for Soto’s 2022 appearances, combed some data out of it, and left some other things behind.
Data gathered: innings pitched, hits, runs, earned runs, walks, strikeouts, home runs, ground balls, fly balls, line drives, inherited runners, inherited runners scored, decision (win/loss/save/blown save)
In order to answer my question, I had to think about how a post-Eckersley closer is typically used. I came up with a category called a Classic Save Situation (CSS), and an appearance needs to check off these criteria:
- pitcher enters the game in the 9th inning or later
- pitcher enters at the start of an inning
- pitcher’s team must be ahead by 1, 2 or 3 runs when they enter the game
Most of the games I examined easily fit, or didn’t fit, these criteria. For a few of them, as you’ll see later, I had to re-think this definition, but ultimately stuck with it as it’s written here.
I was also curious to see how Soto did in other appearances: tie games, for sure, but also I wanted to know how he did when he was put into a game in which the Tigers were down a couple of runs and his job was just to hold things so the score wouldn’t get out of hand. There’s another broad category of appearances in which the Tigers up by more than four runs, and perhaps Soto hadn’t appeared in a game in a while, and Hinch called upon him to “get a little work in.”
All the counting stats were divided by innings pitched, so everything appears as “x per inning pitched.” It makes runs and earned runs look a little strange, but you get the hang of it pretty quickly, I find.
Part 1: Gregory Soto, 2022
For the 2022 season, Soto appeared in 63 games: 33 of these were Classic Save Situations (CSS), and 31 of them were non-CSS appearances. If we break those non-CSS games apart a bit more, 13 of them were games in which he’d entered with the score tied, 6 had the Tigers down two or more runs, and 7 were games in which the Tigers were up more than three runs. (Some games didn’t fit any of those categories, which is why they don’t add up to 31.)
That’s... not great.
Well, it’s great if Soto enters the game in the ninth inning, with no runners on base, and the Tigers are ahead by one, two or three runs. But as you saw, that only happened 33 times this season, and the schedule’s 162 games long, so we have a problem.
The difference becomes even more stark when the non-CSS games are split into those sub-categories. See if you can spot what jumped out at me, too. (I bet you can.)
I know those “down 2+” and “up 3+” games suffer from a bit of Small Sample Size here, but there were a baker’s dozen games in which Soto entered with the score even, and it does not look favourable here. His WHIP in those games was over 2, and that’s terrifying.
Type of contact (ground ball, fly ball, line drive) isn’t quite as useful here as a lot of the other stats, but I figured it would be good to include because it’s pretty tough to hit a home run on a ground ball. Also, the BABIP on those different types of contact is quite different: it’s lowest on ground balls, highest on line drives, and in the middle on fly balls. Maybe someone else can look into this a bit more. I also didn’t bother looking at home runs in any depth because there really weren’t that many.
So, the numbers on Soto in 2022 are pretty alarming. But, if we hop in a time machine and go back eleven years, we arrive in a far-off land called 2011, and the Tigers employed one José Valverde to close things out. Manager Jim Leyland absolutely loved having defined roles in his bullpen, matchups be damned, and Valverde was Mister Ninth Inning that year. How’d he do? Glad you asked.
Part 2: José Valverde, 2011
The same analysis was performed on Valverde’s 2011 season in which he went 49-for-49 in save opportunities. Perfect, right? Not quite.
I had to double-check my formulas to make sure they were right for the calculations of runs and earned runs in CSS’s, and sure enough they were. Get this: in 50 CSS appearances, Valverde gave up a single run four times. That’s it. Four solitary runs. That’s astonishing.
But, of course, that’s not to say he never gave up hits or had baserunners: his WHIP in save situations was 1.160, which is pretty good, but far from automatic. A typical one-inning outing saw him give up a hit, or maybe a hit and a walk, not give up any runs, and get the save.
Alas, when it wasn’t one of those situations, his WHIP was around 1.70, which is pretty terrible. You’re playing with fire with numbers like that.
Let’s break those non-CSS appearances down in the same way here too. (I excluded the “up 2+ runs” games because there were only three of them, and that’s too small a sample.)
It’s exactly the same distribution. Uncannily so. Tie game, Valverde’s brought in? Not gonna end well.
I thought I’d compare Soto in 2022 against Valverde in 2011, in three situations (save, tie, up big) to see how they stack up. It’s pretty remarkable.
If there are two more similar pitchers in Tiger history, I challenge you to find them. (Here the “non-CSS” category includes the tie games, but even if you broke those apart further, you’d get the same numbers.)
Is this distribution true for all closers? As a control, I thought I should bring in one of the all-time great seasons from arguably the best closer of all time. (Apologies to Mike Henneman.)
Part 3: Mariano Rivera, 2005
You could really pick any full season of Rivera’s career after 1996, but I chose 2005 because it was smack-dab in the middle of his career and he had a stupendously good season. Here’s what he looks like in the crudest of breakdowns:
Rivera made 71 appearances in 2005, and he presented a problem to my analysis: namely, there were a few times he was brought in to start a clean eighth inning, in a save situation, and got six outs for the save. Should I alter my definition of a Classic Save Situation to include this? Ultimately I decided against it, because that’s not how closers are used today. Can you imagine Hinch bringing Gregory Soto in to start the eighth in a 4-2 game and expect him to get six outs? Not unless something has gone horribly wrong out in the bullpen and the only people left standing are a catcher, Gregory Soto, and the guy whose job it is to open the gate to the outfield.
Anyway, I digress. Look at those numbers. He was actually better in non-CSS appearances, and he was no slouch when the Yankees were up by a pair and he started a clean ninth. However, we have this vision of Rivera as being perfect and untouchable, getting nothing but flailing strikeouts and broken-bat grounders to second, and the numbers just don’t bear this out. This is because human memories are terrible and fallible.
Let’s look at Rivera’s stat distributions if we break down the non-CSS appearances a little more. Again, we’re leaving out the “Yankees down 2+ runs” category because there’s only two of them.
I was actually pretty surprised at how these appearances were distributed: he only appeared in six tie games, in which he gave up a bit more hits per inning than in other situations, so maybe his manager knew this and gave that job to someone else. Also, he appeared in a lot of games in which the Yankees were up by several runs, but that’s probably because the Yankees were very good that year.
Let’s include Rivera in our big bar graph with Valverde and Soto:
Rivera’s a bit better in CSS appearances, but is head-and-shoulders above the Tigers in those other two categories. That’s probably part of why he was a unanimous Hall of Famer, the first in history.
In short, we’ve seen Gregory Soto’s pattern before: our very own José Valverde, a scant eleven years prior, did almost exactly the same thing. I can’t say whether someone like Mike Henneman, Willie Hernandez or John Hiller would’ve done this as well, but that’s not a fair comparison because those pitchers were on the scene before Tony La Russa revolutionized the closer role in Oakland.
I hope this has helped to either confirm (or disconfirm) your notions about Soto’s season. I hope AJ Hinch has broken these numbers apart or, at the very least, has top men working on it right now. The Tigers will decide whether to retain Soto this offseason as he enters his arbitration years. If they keep him, rather than perhaps trading him or eventually non-tendering a contract, it’s time to find a more suitable role.