How Productive Were Those Outs?

I’ve long been interested in measuring the value of making an out. Different outs count differently, and yet pretty much every baseball statistic you can imagine ignores that fact. I’m not just talking about advanced ones like wRC+ or wOBA, though those do indeed treat all outs as equal. I’m talking about basic things like batting average, on-base percentage, and slugging percentage. No one says, “Well, he batted .320, but some of those outs were in bad situations, so it was more like batting .313.” That’s not how we think about offensive statistics.

But just because we don’t count outs differently doesn’t mean that they all have the same value. This is obviously true. Striking out with a runner on third and fewer than two outs is a tragedy. Hitting a run-scoring groundout in the same situation gets the batter a long series of fist bumps back in the dugout. But when it comes to wRC+ or batting average, that distinction doesn’t show up.

There are good reasons for existing statistics to work the way that they do. Batters don’t control who’s on base and how many outs there are when they come to the plate. They don’t control whether there are fast runners on base, or whether the outfield has arms so weak that anyone could score from third base on a fly ball. In the same way that a home run is a home run is a home run, statistics that try to measure batter skill treat all outs the same. But still… I wanted to know more.

At the beginning of this year, I tested out a method for measuring the differing value of outs based on situation. The idea was simple: compare each out a hitter made to the average run value of making an out in that situation. Imagine our situation from above, a runner on third with only one out. The average out in this situation costs a team 0.28 runs in expectation. A strikeout, meanwhile, costs nearly 0.6 runs; with a runner on third and one out, teams score about one run the rest of the inning on average, but they only score around 0.4 runs after reaching two outs with a runner on third. On the other hand, a sacrifice fly increases run expectancy by around a tenth of a run. A run scored, after all, and there are still more chances to score in the inning, even if having the bases empty with two outs isn’t that attractive of a spot.

I applied this logic to every single out from the 2025 season. In our example above, a player who hit a sacrifice fly would get credited with +0.38 runs; the average out costs 0.28 runs, and theirs instead added 0.1. A player who struck out in that situation would get debited 0.32 runs; his out was worse than a random out when it comes to run scoring.

I had to make a few decisions about how to handle corner cases, like whether you should credit some portion of a sacrifice fly for every fly ball or look at the specific outcomes, and I made some changes to the methodology with more time to look at it. I mostly tried to stick to exactly what happened on the field. I did have one exception, though: TOOTBLANs. I scraped the database for situations where a baserunner made an out that was unconnected to the batter’s action; think a runner thrown out trying to go first-to-third, or the equivalent. It feels unfair to dock the runner for that one when it’s really the baserunner’s fault, so I specifically excluded action that happened after the fact. I also ended up including plays where a batter reached on error, though I could imagine removing those. My logic was that since we’re excluding those in our calculation of wOBA, hitters need to get credit for being fast enough to reach base somewhere. It’s an out in their batting line, so I think it qualifies for inclusion here.

The whole point of this analysis, more or less, is to account for double plays. Hitting into a double play is so much worse than an average out that it makes every other out look good by comparison. And avoiding hitting into double plays is definitely a skill. If you don’t measure that specific skill, you’re missing something about baserunning, so I was very focused on making sure that the accounting there worked right.

With this methodology in hand, a few things should be true. First, the total OAR (out advancement runs, my clunky attempt at a backronym) should be zero. Second, there should be a ton of variation in this statistic from one year to the next. Third, strikeouts should be really bad. When we’re including the chance of reaching on an error, they’re much worse than other types of out. You could imagine a different accounting of this metric where we put reaching on error into wOBA and then exclude it from this calculation, but I wanted to include it here. All of those things were true, so I declared it ready to put in an article.

With that introduction out of the way, let’s get to the data. I think you’ll be absolutely stunned to see the top of this leaderboard:

We’ll get back to the actual leader in a minute, but the guy ranked second caught my eye first. Ah yes, Juan Soto, master of the productive out. But it’s true! Soto’s outs were pretty dang good for the Mets in 2025. The real key to Soto’s success is what he did with a runner on third and fewer than two outs. In that situation, you know what’s coming: The pitcher is going to try his hardest to record a strikeout. Soto made 20 outs with a runner on third base and fewer than two outs. Exactly three of those outs were strikeouts. That’s roughly half the league average rate. That’s a big tailwind right from the jump. In fact, getting the runner home from third, which Soto did 14 times out of 20 outs, accounted for more than half of his positive OAR value. Soto really was adept at driving in the runner from third this year.

You might expect Soto to give a lot of that value back by hitting into double plays. He’s slow and he doesn’t strike out all that often, so surely he’s a disaster with a runner on first base. He did hit into 17 double plays, it’s true. But Soto also batted in a ton of double play situations, and his double play rate is roughly average; he hit into a lot because he had a lot of chances. Meanwhile, he reached on several errors in those situations, and advanced plenty of runners with fly balls, what with him rarely striking out and all. His out value with a runner on first and fewer than two outs was actually slightly above average after blending all of those together.

With that oddball leaderboard placement out of the way, we can talk about the actual leader, Julio Rodríguez. He does a lot of things that scream productive out to me. When he puts the ball in play, he’s so fast and hits it so hard that he frequently reaches on errors. He struck out a bit with runners on, but not at a ghastly rate. He cashed in plenty of sacrifice flies, which I attribute partly to his power; the average Rodríguez fly ball is deep enough to score a lot of runners.

In fact, we can break down the run value we’ve been looking at into groundouts, fly outs, and strikeouts. There aren’t a ton of players who have positive strikeout value, which makes sense, but the best hitters of 2025 consistently did better than league average when they made outs on the ground:

Meanwhile, the bottom half of the leaderboard made me run my numbers a few times to make sure that I wasn’t putting my thumb on the scale to disadvantage the Rockies:

Goodman was a fun success story this year. He also grounded into about as many double plays as Soto, but in 33% fewer chances. With a runner on third base and fewer than two outs, he made 20 outs, and the runner only scored seven times. One of those was even a run-scoring double play. If you’ll recall, Soto cashed in 14 of his 20 opportunities. In other words, Goodman’s offensive game is weak in two ways that wRC+ doesn’t measure but that definitely correlate with scoring runs. Beck’s outs with a runner on third were even less productive; he made 11 outs in those situations, and the runner only scored twice. One of those was even a double play.

I’m not here to tell you that these leaderboards are perfect encapsulations of skill. In fact, I very much want to caution against that interpretation. There’s a lot of variation from year to year. The r-squared of this metric is about 0.09, meaning 9% of variation in year two can be explained by variation in year one.

On the other hand, it’s pretty clearly measuring something. With my new methodology, Rodríguez has finished first in baseball for two years running (Corbin Carroll was first last year under the old method of calculating OAR, dropping to sixth using the new method). Ryan McMahon made very unproductive outs in Colorado in 2024, made very unproductive outs in Colorado in 2025, and then went to New York and made yet more unproductive outs. There’s a ton of noise here, but there seems to be signal too.

Here, as usual, is a full leaderboard for 2025. I’m curious to hear what you, the reader, make of this statistic. I myself am unsure about how to use it. On the one hand, it definitely measures something that happened in the game. On the other hand, I’m not sure how much of it is actually skill-based and how much of it is just noise. We ignore a lot of things that happen on a baseball field that we all agree don’t really matter when it comes to measuring player skill. I’m unsure whether this is one of those things. It’s undoubtedly interesting, though, and I hope you’re as fascinated by the list as I was.