On Saturday, the Oakland A’s made their fifth attempt at using a relief pitcher as an opener to start a game, with a traditional starter coming in later. Here is a full rundown on the idea, and here is a look at how the first four tries went.
This time, it didn’t go well. Liam Hendriks pitched a scoreless 1st inning as usual, but then lefty reliever Dean Kiekhefer gave up two runs in the 2nd. Chris Bassitt entered in the 3rd as the long-man and made it to the 6th before allowing Tampa Bay to score again, but in the end the A’s never held a lead and the Rays won the game.
The early scoring by Tampa Bay wasn’t the only factor in the final result, but it certainly didn’t help matters and it got the evening off on the wrong foot. Either way, though, I’m not blaming the opener strategy for any part of this loss, because frankly the A’s did it wrong. They didn’t follow the game plan as it’s designed to work, so of course it went awry.
There are two key mistakes that Oakland made in this game, and both are detailed in that first “full rundown” link above. One is that they didn’t bring in their long-man (Bassitt) for the 2nd inning, when he would have faced the bottom of the lineup. The other is that they didn’t take the early innings seriously enough, bringing in too weak of a reliever too early in the contest.
1. Bring in the long man for the 2nd inning
As I understand this strategy, the whole point is to reduce the number of times that the starter faces the top of the lineup, which tends to feature the opponent’s best hitters. It’s the only possible reason I could see for this entire exercise. Hold the starter back until the 2nd inning, when he can begin his day against the middle/bottom of the lineup.
Instead, Bassitt didn’t enter until the 3rd inning. By that point, the first batters he faced were the 3-4-5 heart of the order. He actually ended up having a totally decent performance, throwing three full innings and allowing just one run. But the fact remains that he faced the other team’s best hitters twice and their worst hitters once, which is the exact opposite of the game plan. If you’re going to do that, then why not just have him start the game like normal in the 1st inning?
The reason Bassitt didn’t come in for the 2nd was that Kiekhefer did instead. On its own that makes sense, because the Rays had a bunch of lefties due up and the A’s countered with a LOOGY specialist. The things is, though, the Rays lineup was nearly entirely lefties — seven of nine total. Every inning was going to have lefties due up. There was no rush to capitalize on that platoon advantage at the expense of the entire rest of the day’s optimal game plan.
In the same vein, you simply can’t be afraid to hide the long-man from the bottom of the lineup. Again, the primary strategy here is to hide him from the top of the lineup, so if he also avoids the bottom then who is left to face? If he can’t be trusted for the top nor the bottom then maybe someone else should just pitch that day.
It comes down to this: If you’re not specifically targeting to have the long-man come in for the middle/bottom of the lineup, then just start him outright. Otherwise you’ve thrown the whole pitching arrangement and everyone’s individual roles into confusion, all while flushing away the marginal benefit that was supposed to be the reward for all the madness.
2. Take the early innings seriously
Alright, though, let’s say you want to get fancy. Maybe there’s a reason why it was better to hold Bassitt back until the 3rd inning, and maybe no one cares about which hitters he faces first and/or how many times he sees them. Even if that were the case, then it was still a mistake to turn to Kiekhefer so early.
The A’s have 15 relievers in their bullpen, and none of them have less experience than Kiekhefer. He threw a couple dozen innings in 2016, and Saturday was his second MLB appearance of this season. With all due respect, he’s almost certainly the worst pitcher on the team — someone has to be, after all. It’s possible he’s only second-worst, depending on your thoughts about Aaron Brooks.
When Kiekhefer entered the game, the score was tied against a borderline contending team that is technically chasing the A’s in the standings. It wasn’t traditionally high-leverage because it was so early, but the scoring he allowed was still costly. Tampa Bay’s win probability was about 55% when he took the mound (since, as the home team, they had one extra ups remaining), and after Brandon Lowe’s two-run homer the odds increased to 75% for the Rays. The A’s eventually lost by exactly two runs.
Calling on your worst pitcher to dig a hole in the 2nd inning simply is absolutely not part of the opener strategy. It’s something else entirely. Even if you could make a case that bringing in a lefty was the proper decision in that spot, then at least use Ryan Buchter, the actual MLB-caliber southpaw on the roster.
This was essentially a repeat of the first (also failed) attempt at the opener at the beginning of the month. In that affair, Hendriks got into a jam in the 2nd and was relieved by ... Danny Coulombe, who had spent much of the year in the minors and was DFA’d the next day. Coulombe let two inherited runners score in a game the A’s lost by one.
Perhaps this kind of white-flag move would be justifiable in a situation where the starter had left early due to injury and the eight-man bullpen was going to be stuck covering eight innings. Then you’ve got to pace yourself, and you may as well gamble on a couple of your lesser arms first to see where that leaves you. But that wasn’t the case here, because the pseudo-starter was still going to come in and throw several innings. It’s still a tied, winnable game, so treat it like one regardless of what inning it happens to be.
If you’re bringing in another short reliever for matchups in the 2nd inning, then don’t think of it as being the 2nd. Rather, think of it as being the 6th or 7th, but just a bit ahead of schedule. You’re simply reallocating a late inning to earlier in the evening.
Reminder of why this is happening
If you’re still wondering why this is happening at all, here are some numbers for you. The sample is too small for serious conclusions, but the early returns show about what you would expect. These are career stats, in 37 games for Daniel Mengden and 30 for Bassitt, measuring how they fare each time through the opposing lineup.
Mengden, 1st time: .713 OPS, 2.9 K/BB
Mengden, 2nd time: .703 OPS, 2.9 K/BB
Mengden, 3rd time: .860 OPS, 1.8 K/BB
Bassitt, 1st time: .710 OPS, 2.4 K/BB
Bassitt, 2nd time: .674 OPS, 2.2 K/BB
Bassitt, 3rd time: .819 OPS, 1.6 K/BB
These splits are only a couple hundred plate appearances each, so frankly I’m more interested in the K/BB ratios than the OPS marks. Both of them tell the same story, though, and the moral is to avoid that third time through. And if you must face someone for the third time, try to at least make it a bad hitter rather than the other team’s leadoff man or 3-4-5 heart.
The career splits aren’t as pronounced for the four veterans (Fiers, Jackson, Cahill, Anderson), which is presumably one reason why they haven’t been subjected to this experiment. However, you can still see the effects clearly in their K/BB rates, which is presumably why they still aren’t allowed to go too deep into games.
Even given all that, though, why is it preferable to turn to a reliever like Hendriks, who was himself DFA’d this summer? First off, Hendriks might honestly be hitting a second wind and finding some of that untapped upside, as he’s currently throwing the best stuff we’ve ever seen from him in Oakland. It might be time to readjust our perception of him from a gascan to a serviceable reliever.
On top of that is the concept of max effort vs. pacing oneself. Starters are generally considered to be better pitchers than relievers, which is one reason they get to carry so much more of the workload. However, the reason why a pitcher moves from starting to relief is often that he never develops a third pitch to diversify his arsenal over a long outing in which he’ll face the same hitters multiple times, or he doesn’t have the endurance to make it deep into a game effectively.
In a short one-inning stint, though, a reliever can be his best self. He can easily be better than a starter might be in one of the six innings he’ll pitch that day. The starter’s strength is that he can stay relatively effective for a long time, because for most of the season there aren’t enough roster spots to have nine relievers throw an inning apiece every day. (Note: Bassitt told me earlier this month that for now, even in these relatively shorter long-relief outings, he still has to pace himself the same way he would in a full start.)
Perhaps if you compared them both in one-inning stints then the starter could play up his stuff and be superior. But it’s not a stretch at all to see how a reliever, going all-out in a much shorter appearance, can end up being more effective than a guy who has to pace himself over the long haul. Heck, that’s exactly where Hendriks The Reliever came from in the first place, maximizing his stuff and becoming a better performer than he had been in longer outings.
It’s really the entire premise behind pulling your starter to bring in a setup man — you wouldn’t have the setup man throw the whole game, but in his one little inning he can be better than the starter who had to pace himself and is getting tired. It’s the reason that the lowest ERAs in the league will always be held by relievers, even though the same pitchers would be crap if they moved up to starting roles. Hendriks has a sub-3.00 FIP since 2015, and we’d lose our minds if we had a starter doing that well (Trevor Cahill is best at 3.51).
And then, of course, factor in that we’re talking about the team’s worst and least reliable Quad-A starters as opposed to established vets much less aces, and it gets even easier to see how a one-inning burst from Hendriks tossing 97 mph could be preferable at an opportune moment against some top hitters.
The A’s lost on Saturday, but I don’t blame the opener strategy at all. The opener threw the scoreless 1st inning that was asked of him, and the long-man was effective even in a bastardized version of his role. Oakland lost because they deviated from the plan in the 2nd inning in favor of something completely illogical, and because their lineup couldn’t find enough big hits, and because later on a member of the elite setup crew melted down and walked his way into a backbreaking rally.
It’s still possible that the opener will turn out to be a bad idea, but it won’t be because of Saturday’s game. That was just Bob Melvin trying some different stuff that didn’t work. The strategy is simple: Opener in the 1st, long-man in the 2nd against the middle of the lineup, period. There is so little room for tinkering that it really should be avoided entirely.
We’ll see it all again tonight, as Hendriks draws his sixth opener start. This time he’ll face the Angels, and Daniel Mengden will follow him. The last two times that duo paired up, they carried shutouts into the 6th inning of each game while allowing one total hit each time. If you want a shining example of the opener working properly, there it is.
Ironically, the Rays only really used the opener once in three games against the A’s. The other two were full-on bullpen games, with no pitcher throwing more than two innings at a time. In their one true opener game, Yonny Chirinos entered for the 2nd inning and threw 5⅓ solid frames. After completing his first two times through the lineup, he stayed in the game to face Matt Olson for a third time. Olson worked a nine-pitch at-bat that ended with him crushing a homer. It’s anecdotal, but that’s why you avoid that third time against the top hitters.