Replay Is Fine, Everyone

We spend a lot of time fretting about baseball. Baseball games take too long; teams use too many pitchers and make too many visits to the mound. There are too many strikeouts; there aren’t enough balls put in play. These complaints are dressed up in anxiety over the game’s future, but I think the real worry is closer to home. It’s about us. I think what’s really at the center of it is a gnawing concern that these slowdowns will make us want to watch something else entirely, that we might come to find baseball boring.

But there are worse things than being bored. Wednesday night, in the fourth inning of the Dodgers-Rangers game, Adrian Beltre scored on a close play. It appeared that Austin Barnes had tagged him out, but home-plate umpire Sean Barber disagreed. A man in the crowd was inspired to make a face.

Dave Roberts challenged and it went to a replay, which began at 2:02:25. The home broadcast showed some slo-mo.

The broadcast was confident the replay would go the Dodgers’ way. Enrique Hernandez, whose throw looked like it had nabbed Beltre, seemed confident. Beltre looks pretty out.

But at 2:05:00, the call on the field was upheld. The crowd booed. They’d spent more than two-and-a-half minutes waiting around — only to lose out. It didn’t end up mattering: the Dodgers won in the 11th inning after Hernandez evaded a tag of his own. But for those few minutes in the fourth, Dodger fans were something worse than bored. They were bored and angry. It’s a terrible combination of things to feel, and one that replay seems to inspire often, which is understandable, though I’ll admit it makes me worry about how passionate we are for justice. And so, in all our fretting about the game, I thought I’d check in on replay and see how it is going so far this season.

Baseball Savant maintains a handy replay database, but it doesn’t include 2018 replays yet, so it’s Retrosheet to the rescue. Retrosheet’s data also includes the duration of each replay — an indispensable data point for those concerned with the dull and enraging. They update their data every two weeks; the replays I’m analyzing are through May 31.

Two quick notes. First, the time listed for each replay is from the beginning of the review until New York’s decision is announced. That might seem like an obvious point, but it may, in some cases, undersell the length of the delay on the field. Last year, I wrote about an 18-minute long replay at Dodger Stadium. It was a rules check and the longest replay of 2017. Retrosheet has it taking 8:34. So there’s a bit of squishiness here.

Second, in case you’ve forgotten (and honestly, why would you remember?), before the 2017 season, MLB released new guidance that, with a few exceptions, the Replay Operations Center in New York has two minutes to render a decision on a play. That’s what they’re driving toward. It’s part of keeping us from feeling bored and angry.

Now, some observations.

You might ask, “When is replay most likely to occur?” Maybe you’re naturally curious about things. As you might imagine, challenges become more common the later into a game a team gets.

Through May 31, managers initiated 436 challenges, a full half of which have came between the sixth and ninth innings; indeed, since replay expanded in 2014, close to 49% of the replays in nine inning games have come between the sixth and ninth innings. The eighth inning saw the greatest number of challenges, but also the lowest success rate, among non-extra innings frames. That makes a certain amount of intuitive sense. Late in games, I would imagine, managers are more inclined to challenge borderline calls, both because the stakes are higher and because why the heck not? You can’t take those challenges with you. More borderline calls also means more calls on the field that stand or are confirmed, but why not try? Maybe that runner in scoring position is actually out on the tag!

And speaking of tags, you might also wonder, “What is getting reviewed, and for how long?” The below table shows all 2018 replays by type, along with the average and median duration of the replay in minutes, and the success rate for challenges of each type.

Fans have, for the most part, stayed out of the way. Despite recent dustups, the slide rule that caused so much controversy in years past hasn’t been much of an issue, or at least has merited little investigation. Umpires mostly know what a catch is. Force plays seem a bit trickier, though they didn’t take long on average to sort out. Stadium boundary replays took the longest, both by average and median length in minutes, though there weren’t many of them. With the exception of the boundary plays, replay officials are, on average, adhering to their two-minute guidance. Of the 490 total calls, 377 have been two minutes or under in length.

But I think the most common category of replay underscores the enterprise’s greatest challenge (no pun intended). Now, I haven’t watched all 197 tag replays, but I would hazard a guess that some portion of them — perhaps a significant portion — involve runners coming off a base ever so slightly for just a teeny tiny touch of time. We’ve seen this sort of replay play out, sometimes in big moments of important games, resulting in a guy who would have been safe for the 100 years prior suddenly being out. We can’t exactly blame managers for asking that tags be reviewed; we’ve told them there might be an out hiding in there. And some portion of these allow us to examine swim moves and close plays, and that isn’t a terrible use of time. But we’ve spent some part of 288 minutes peaking under guys’ fingers and toes. Avengers: Infinity Wars, for sake of comparison, was only 160 minutes and involved a bunch more people. I submit that this is when we are at our most bored, and certainly our most angry.

And of course, fans of some teams should be angrier and perhaps more bored than others.

None of these samples are large enough to tell anything definitive, but as an indication of efficacy so far, we can learn a few things. The Royals, Phillies, and Yankees have fared the best in their challenges. The Braves have challenged more times than any other team, but have a middling success rate. They are still doing better than the Padres, who (in admittedly fewer attempts) have a league-worst success rate. The Orioles fare only marginally better.

Baltimore did initiate the longest challenge of the year, a review of a fair/foul call that lasted 4:32 they ultimately won.

Everyone looked thrilled as they waited.

Just a great day at the office.

We can also see something interesting when we look at the distributions of how long reviews take, grouped by their result.

The graph isolating 2018 is a bit rougher, but retains the same general shape.

From 2014 to -18, a “stands” call took about 40 seconds longer than “confirmed” or “overruled” calls did, which I think shows that replay is generally working how you would want it to when you consider that the standard for overturning calls made on the field is having “clear and convincing evidence” that the call was incorrect. One would hope that if a call were obviously right or obviously wrong, it wouldn’t take very long to reach that conclusion. Absent some bit of striking evidence, best to leave it be.

I think it is worth adding a small bit of perspective to this analysis. We’ve all had the experience of seeing a replay go the “wrong” way. We’ve all felt like our boys have gotten jobbed. We’ve been Wednesday night’s Dodgers’ fans. We begin to question the whole endeavor.

But we might benefit from recalling how frustrating it was, in the era of slo-mo and hi-def, to know that a call on the field was wrong, to be able to see it right there, and then have to watch as a baserunner trudged back to the dugout when he should have been on base, or as a pitcher was left to contend with a runner who should have been erased by a tag. It felt unfair. It felt silly. It felt like an injustice. We’re sometimes bored and angry now, but we were also bored and angry then! And it isn’t costing us that much. The season isn’t done, but so far, 2018 is following a trend of replay times decreasing as the years go on.

These aren’t huge numbers to begin with; I doubt even a close observers could perceive the difference between 2015’s average replay time of 1.85 minutes and 2017’s 1.46. But it isn’t ballooning out the other way, either. Through May 31, MLB was on pace for 1,285 replays, which would be the lowest number since replay was expanded in 2014. That could change, of course, but it hasn’t been so bad so far. It’s an attempt to get things more right more often.

I calculated how much time each team had spent under replay review, including both those reviews they had initiated and those initiated by their opponents or by umpires, and compared that time to their total game minutes in 2018. I’ll spare you another long table, but the team that has spent the most time in replay as a percentage of their total playing time is the Blue Jays at… 0.57%. That’s a little more than 58 minutes across all their games, and theirs is the worst of it.

That isn’t so bad. Replay gets things wrong from time to time; we all have bad days at work, after all, and humans remain fallible, even with slo-mo. But I’m not sure the game is well served by putting too strict a clock on justice. Not even when we’re bored and angry.