Skip to main content
ABC News
The Rays’ Radical Reliever Experiment

It’s not easy to be the Tampa Bay Rays these days. Long renowned as a sabermetrically-advanced team, the Rays have seen their strategies copied, imitated and stolen by other teams in search of a tactical edge. That’s why I’m watching the Rays closely this season. One of the most intriguing storylines of the 2015 season is how the Rays and other small-market teams cope with a league that’s increasingly analytics-savvy. It used to be that sabermetrics were the way small-market teams gained an edge over the competition. But now that analytics is becoming second nature in nearly every front office, its advantages are less pronounced. How will the Rays maintain their edge?

We saw the hint of an answer last Friday, when the Tampa Bay Rays tried something different. It was only a one-game experiment, but it was the kind of experiment that could hint at the post-post-Moneyball era. It started with a relief pitcher.

Faced with a depleted rotation — four starters were hurt — the Rays decided to have a “bullpen day” and let the relievers have the run of the place, opening the game, closing the game and pitching every inning in between. That’s nothing unusual, other than how early in the year it took place. But instead of deploying a reliever with starting experience — most likely Erasmo Ramirez — the team chose Steve Geltz, who hadn’t started a game since he was in high school.

The Rays’ unconventional choice may have been strategically brilliant. While we can only speculate as to the reasoning behind the decision — the Rays declined my request for comment — there were two potential advantages to be gained by starting Geltz. By deploying Geltz before Ramirez, the Rays gave themselves the chance to essentially gain a DH in the NL and to keep Ramirez from having to face the toughest part of the opposing lineup.

Geltz pitched for only two innings, and then a pinch hitter, Mikie Mahtook, batted for him. Bypassing the starting rotation completely allowed the Rays to get the benefit of Mahtook’s bat without sacrificing any pitcher innings. When the replacement reliever came up, a pinch hitter hit for him too — no pitcher ever batted. The pinch hitters went 3 for 5, but even if they had gone 0 for 5, they surely had a better chance of getting a hit than AL relief pitchers who don’t spend much time in the batting cages.

So that explains why they went straight to a reliever instead of a spot starter, but it still doesn’t tell us why the Rays decided to start Geltz in particular rather than Ramirez. It may have been an attempt to make the most of their threadbare bullpen. It comes down to matchups: Geltz is likely superior to Ramirez, with a projected 3.43 ERA for 2015 compared to Ramirez’s woeful 4.42. While no bullpen ace, Geltz stood a much better chance of surviving the first few innings intact.

That matters because the first inning is when teams score at the highest rate. The first inning sees 0.53 runs cross the plate on average, after which the average drops to 0.40 runs in the second before rebounding to an overall average of 0.46 runs per inning for the rest of the game.

Some sabermetricians believe that more runs are scored in the first because pitchers have to face the top of the order. Competent managers fill the top four or five lineup spots with the best hitters, so pitchers are often running the gauntlet in the first inning.

This was especially true for the Miami Marlins, the Rays’ opposition last Friday. The top four hitters on the Marlins combine for a projected 121 Weighted Runs Created+ (wRC+), 12th best in the league. (wRC+ measures the total offensive output of a hitter, on a scale where 100 is roughly average and 120 is 20 percent better than average.)

And while the Marlins are geared for first-inning run production, the bottom of their lineup offers some easy outs. After the top four sluggers, their offense drops off dramatically: The next five best hitters combine for an 83.2 wRC+, which is almost as far below average as the top of the order is above it. The Marlins have the largest wRC+ gap between the top four and bottom five hitters in the league. Since the Marlins are uniquely unbalanced, the Rays would have risked getting their swingman, Ramirez, knocked out early if they had thrown him to the wolves of the first inning. Better to start with a guy who couldn’t go more than a couple innings anyway.

Now that I’ve told you how artful and clever the Rays’ maneuver was, I suppose it’s worth sharing the outcome of the game: The Rays lost. After Geltz was lifted in the second inning (he threw 35 pitches and gave up a run), Mahtook, the pinch hitter, promptly lined out, negating the pseudo-DH advantage. Later, Ramirez had to face the tough part of the order and gave up seven runs in his delayed “starting” stint. He lasted just two innings, undoing the second benefit the Rays’ gambit might have yielded. The Rays later managed to tie it up, but ultimately lost in extra innings.

But, as the saying goes, you can’t predict baseball (especially one individual game). The Rays’ failure underscores a problem with many modern attempts at in-game tactical optimization: at best, their effect is minor. The best-laid plans of Rays and quants can be undone completely with an untimely lineout or bloop single. Still, it’s intriguing to watch the Rays mine every available source to save or gain a run or two here or there.

This unconventional strategy is emblematic of the cutting edge of modern, in-game sabermetrics. With every team from the small-market Rays to the moneyed Dodgers now employing a host of analysts, all the obviously beneficial tactical techniques are already in use. All that remains are the scraps. So even as the tactics have become unconventional, the margins have become thin and the advantages minute. For clever front offices like the Rays, negotiating the brave new world of saber-equality may mean relying on subtle, trifling tweaks like reliever-first outings.

CLARIFICATION (April 17, 1:00 p.m.): An earlier version of this article said the most runs are scored in the few innings of a baseball game. It’s since been updated to more accurately reflect what the data makes clear: the first inning has the highest scoring rate of all innings.

Rob Arthur is a former baseball columnist for FiveThirtyEight. He also wrote about crime.