One year ago today, those ever-innovative Tampa Bay Rays changed the game again by reintroducing the concept of "the opener" to baseball. On May 19, they gave Sergio Romo his first career start, after his first 588 games came in relief.
Romo struck out Zack Cozart, Mike Trout, and Justin Upton in a perfect inning in that first start, as the Rays went on to win, 5-3. The next day, he started again, throwing a second scoreless inning, though the Rays lost, 5-2. Romo only made three more starts in 2018, but the Rays ended up using the opener dozens of times and continue to do so in 2019. Several other clubs ended up trying out the strategy, too.
The Rays had a 4.45 ERA as a team headed into that first game, the ninth highest in baseball. They've had a 2.85 ERA, by far baseball's best, since. So the opener was a massive success, right?
Sort of. It's been useful, but "Did the Rays' ERA go down?" isn't exactly the best way to measure it, either. Now that we're a full year into the world of the "opener," can we see how it's affected baseball?
What is an opener, anyway?
We do have to set some ground rules here, believe it or not, because it's actually surprisingly difficult to identify "opener" games. There have been 302 games since the start of 2018 where a starter went six outs or fewer, but those were not all "openers."
A bullpen game, where a team knows it's going to use a steady stream of relievers for four or five outs at a time, is not an opener game. A game where the regular starter gets injured, or ejected, or gets hit hard, or otherwise leaves early isn't really in the spirit of an opener game, either.
The entire point of the opener is to still use a "starter," just in a different way, slightly delayed. It allows a pitching team to line up a good reliever against the top of the opposing lineup, which should generally be three of the best hitters available. That's part of the reason why the first inning has traditionally been the highest scoring -- it's the only inning the hitting team can guarantee who comes to the plate -- and it also helps alleviate the "third time through the order" penalty, by making it so the third time through for the pitcher who follows the opener begins with the middle of the lineup, not the top.
That's the way it's supposed to work, anyway. So, the way MLB.com's Tom Tango has helped to define this is to look for any games where:
• The first pitcher, the opener, goes no more than 2 innings or faces no more than 9 batters, and
• The next pitcher, the headliner, goes at least 4 innings, or faces at least 18 batters, and is the pitcher who immediately follows the opener (unless the opener was relieved mid-inning, and this pitcher starts the next inning).
That cuts it down to just over 100 games, but even that's not good enough. We went through and eliminated some games that qualified yet were obviously not in the spirit of the opener. These include all sorts of things, like Kevin Gausman getting ejected in the second inning on May 3, or Trevor Bauer facing six hitters in a post-injury warmup start last Sept. 21, or Dylan Bundy allowing four home runs without recording an out last May 8, or Taijuan Walker leaving with an injured elbow last April 14, or Nathan Eovaldi having his night ended early by a rain delay of over two hours last Aug. 31. Those count, but they don't really count.
That leaves us with ... 74 games.
Who's been using the opener?
The Rays, obviously. More than half of all "opener" games by our definition have been from Tampa Bay. Including the playoffs -- and that's important, because you may remember what the Brewers did with Wade Miley in Game 5 of the NLCS -- these are the teams that have used the opener.
Blue Jays 1
That includes Jose Leclerc successfully opening for Adrian Sampson for Texas on Friday, and Pittsburgh's first shot at it with Montana DuRapau on Saturday, but it doesn't yet include the Yankees, who may use Chad Green as their first-ever opener on Sunday.
No one has been an opener for multiple teams, though Nick Vincent came close. He threw two innings for Seattle last Aug. 21 ahead of Ross Detwiler's six, though even that only came because Mike Leake got sick, and while Vincent did start for the Giants on Wednesday, Tyler Beede went only 2 1/3 to follow him to make it a "bullpen game." You can see how complicated this gets.
That Detroit game at the bottom of the list, by the way, came on Wednesday, with Gregory Soto going two innings. As manager Ron Gardenhire explained, the Tigers were following the crowd.
But has it? Seventy-four games is actually not that many. It's fewer than 2 percent of all games since Romo's first start, and it's something like one opener every 65 games. Even if you want to loosen your definitions slightly, it just doesn't happen that often, outside of Tampa Bay.
It's also fewer than you'd think for the Rays, because Ryne Stanek, the King of the Openers, has made 42 starts, none lasting more than two innings. But while the Rays using Stanek for 1 2/3 innings ahead of Ryan Yarbrough's 6 1/3 innings last June 28 was what you'd call "the opener," using Stanek for two innings ahead of five pitchers who all faced 12 batters or fewer last June 12 was more of a "bullpen game."
Here's the thing, though: Romo wasn't the first opener for the Rays in 2018. By this definition, the Rays had actually used it twice with Andrew Kittredge, who was, in those pre-opener days, referred to as the "beginning pitcher." It didn't get as much attention, apparently, because having the inexperienced Kittredge start wasn't quite as shocking as having a longtime veteran reliever like Romo do it.
Regardless of how you define it, the answer here is something along the lines of "the Rays use it once or twice a week, and other teams have been slow to adopt it."
Has it worked?
Remember above, when we said that the Rays' ERA had gone from baseball's ninth-highest pre-Romo-opening to by far the best? It's powerful evidence in favor of the opener, but it also neglects just how incredible AL Cy Young winner Blake Snell was down the stretch last year, or how good Tyler Glasnow was this year before he got injured, or how valuable free-agent addition Charlie Morton has been. The Rays have good pitching, regardless of usage.
That's what makes it so difficult to look at, on a team level. Even if 42 games was a huge sample -- it's not -- comparing Rays opener games to non-opener games isn't that satisfying. When the opener was used, they allowed a .304 wOBA. Without the opener, it's .287. That's worse, but it also tells you just about nothing, because games where Yarbrough, Kittredge and friends are pitching were always likely to be less productive than games pitched by Snell, Morton and Glasnow -- three traditional starters who never follow an opener.
Instead, let's try to see about the effect on the pitchers involved. After all, most of the point is supposed to be help a non-elite starter like Yarbrough, right? Conversely, has it hurt Stanek to do this? This is imperfect because we're not accounting for which teams he faced, but it should tell us something.
It does, actually. Just maybe not what you'd expect.
As headliner -- .313 wOBA
Other games -- .320 wOBA
Yarbrough was slightly more productive in games where he followed an opener, but by barely any margin at all. He had fantastic outings with it, throwing seven innings of one-run "relief" behind Romo last May 25, and fantastic outings without it, throwing five innings of one-run ball against the Red Sox as a regular starter on Aug. 25. Yarbrough got lit up with it, allowing six runs over five innings to the Angels behind Stanek on July 31, and without it, allowing five runs in 4 2/3 innings to the Phillies on April 15, 2018.
Even if you just look at his 2018-19 in terms of "games started" and "games relieved," ignoring the opener being in play or not, Yarbrough allowed a .317 wOBA as a starter (six games) ... and a .318 wOBA as a reliever/headliner.
It seems like this hasn't made much of a difference at all, other than inflating his 2018 wins total to 16 by letting him pick up some cheap wins, since being not-the-starter waives the requirement of going five innings. It's possible that making it look like "no effect" is the effect, in that it's allowed him to skip some third-time-through-the-order plate appearances that might end poorly for him, it's just difficult to see in these results.
As opener -- .271 wOBA
Other games -- .327 wOBA
For Stanek, we're looking at all of his "starts," because for his purpose, he doesn't really need to care if there's a regular starter behind him or a ton of relievers. That allows us a little more flexibility in terms of finding stats to share, and ... it actually seems like he's been more productive as a starter than as a reliever.
The Rays have also occasionally used pitchers like Diego Castillo, Hunter Wood, Yonny Chirinos and Kittredge as openers, but rarely enough that it's difficult to draw any individual conclusions. (Combined, however, they have had a .271 wOBA in opener games and a .307 in non-opener games in 2018-19.)
That's really the takeaway here, that a year after the opener was reintroduced to the game, it hasn't really taken over. It's been a regular thing for the Rays, but not for most other clubs. We don't have enough information to really say whether it's worked or not, especially for teams who only used it once or twice. It's near impossible to identify games where the headliner would have pitched enough to make it an opener game if he just performed well enough to meet the minimum, which might skew the results, and it's hard to parse out the Rays' success from the fact that they just have good pitchers.
The opener should have never been expected to wildly change outcomes, anyway. It's a gambit to put your weaker pitchers in a slightly better position to succeed. It's a potentially small edge in a sport that thrives on finding those edges. Whether or not it even provides that edge remains up for debate, but the Rays will surely keep trying to find out.
Mike Petriello is an analyst for MLB.com and the host of the Statcast podcast.