The way Kevin Cash managed the fifth inning wasn’t about saving bullpen arms. It wasn’t about “soft punting” once the Dodgers got the lead.
Running Tyler Glasnow though 112 pitches up to Cody Bellinger for the third time, was a legitimate attempt to win the game, and an aggressive play by Kevin Cash. It has to do with the matchups.
Let’s walk through the fifth inning. You can read more about the framework I’m using here.
Entering the Fifth
Through four innings, Glasnow had given up two hits, mostly working around his four walks. He had struck out seven Dodger batters, with his only blemish on the night coming off a two-run homer by the excellent Cody Bellinger.
Glasnow had started the game unable to get his fastball up, but had since managed to raise his sights, and now was tending to let it drift a little bit to high. His curve was biting. He’d thrown a few excellent changeups that umpire Laz Diaz was totally incapable of seeing for the strikes they were. This wasn’t the best version of Glasnow that we’ve seen, but it also wasn’t the worst. And — yes this is an unpopular assertion but it’s also a sound one — prior in-game pitcher performance is not a good predictor of future in-game pitcher performance.
The fifth inning brought up the Dodger’s lineup for the third time. We talk a lot here about what the times through the order penalty is. It looms large in the way the Rays approach bullpen management. Let’s also note what it’s not.
The times through the order penalty is not a brick wall. When Mookie Betts walked to the plate for a third time, Tyler Glasnow did not turn into a pumpkin. He was still 6’8”, with a fastball that touches 100 mph. Yes Mookie had seen Glasnow’s curve, and was less likely to chase it and more likely to hit it, but that curve was still the best 12-6 curve in baseball. Baseball outcomes are quantized. The expectations are incremental.
So to analyze this situation, we need to put some numbers to Glasnow the third time through the order.
Based on this framework, we have Mookie Betts leading off as a slightly below average hitter, followed by Corey Seager as an above average hitter, followed by three Dodgers who are still expected to be solidly below average against Glasnow. Then there’s Bellinger, who you cannot (and Kevin Cash would not have) let Glasnow face for a third time in an important situation.
Why are these projections still so manageable looking for Glasnow? Because he’s very good.
Based on regressed handedness platoon splits combined with Depth Charts projections, I have Glasnow as 20% better than the average righty against righties, and 17% better than the average righty against lefties.
In the bullpen, only Nick Anderson, Pete Fairbanks, and Aaron Loup have a better base projection against a lefty than does Glasnow. Ryan Yarbrough, the lefty starter with a sweeping slider who was warming, actually has a worse projection, first TTO (I think Yarbs should be considered a bit better than this, for reasons, but that’s for another time).
The last thing to note about the situation is that there are five innings remaining. Roughly speaking, I think the “A Bullpen” in this matchup consists of Anderson, Fairbanks, Castillo, Loup, and Yarbrough, and it would be possible to roll each of them for one inning and cover the remaining five. Using any of them for multiple innings potentially impacts their availability for the next day (or in Yarbrough’s case, a game four bulk role). Multi-inning appearances are something that should be on the table, but the situation has to warrant it.
Let’s walk through some of Cash’s options.
A Reliever to Start the Fifth
One of Cash’s options would have been to bring in a reliever right at the top of the fifth inning, and prevent any Dodgers batters from seeing Glasnow for the third time. Below are the average wOBA projections I’ve got for the matchups of each Rays pitcher from Betts through Will Smith (the batters Glasnow ended up facing). Glasnow’s projection is third TTO.
Options to Lead Off the Fifth
Pitcher | Proj. wOBA through Smith |
---|---|
Pitcher | Proj. wOBA through Smith |
Anderson | 0.250 |
Fairbanks | 0.291 |
Loup | 0.302 |
Glasnow | 0.305 |
Castillo | 0.313 |
Yarbrough | 0.337 |
Anderson and Fairbanks are clearly better options. Loup is about the same (Loup’s projection here is a divisive topic among Rays fans; you might have him considerably lower in your estimation). Castillo and Yarbrough are easily worse options — Castillo because of his lower-than-some-might-expect 3.72 ERA projection and the presence of Seager up second in the lineup, and Yarbrough because of the presence of Betts leading off, and Justin Turner waiting in third.
Here’s what the inning looks like for Yarbrough.
The one and the three spots are ugly matchups that Cash should not walk his lefty into willingly.
Back to the good options, Anderson and Fairbanks are clearly upgrades, but here’s the thing — the leverage index to lead off that inning was below average at .86. Cash shouldn’t be using his two highest leverage arms in below average leverage situations in the fifth inning.
So Glasnow faces Mookie Betts, and he walks him.
Options to Face Seager
With Betts on first base, the leverage jumped to 1.34. Now it was above average, but still not very high. And the funny thing is that a walk to Betts was the single event that could have raised the leverage more than any other. For your LI amusement, once Betts stole second, that actually lowered the leverage, as it made the run more likely to score and the Rays less likely to come back.
If the Rays were to come back in this game, they would be presented with higher leverage later on in the game. If they were not to come back, it would be a moot point. This is why not having Anderson or Fairbanks warm and ready to come into the game if Glasnow didn’t get Betts remains a valid option for a reasonable manager seeking to maximize his chances of winning a single game.
And the remaining pitching options still have Glasnow as a very competitive choice.
Options to face Seager
Pitcher | Proj. wOBA through Smith |
---|---|
Pitcher | Proj. wOBA through Smith |
Anderson | 0.250 |
Fairbanks | 0.287 |
Loup | 0.291 |
Glasnow | 0.302 |
Castillo | 0.311 |
Yarbrough | 0.328 |
If you believe my projection for Loup, especially against righties (I have him here as basically the same as Blake Snell), then you should want Loup warm and ready to come into the game right now instead of Yarbrough. Loup would then run through Bellinger. Here’s what that set of matchups looks like.
But if you don’t believe in Loup and want to put this projection aside, it becomes Yarbrough that we’re talking about, and Yarbrough to face Seager, Turner, Muncy and Smith is simply a worse win-now decision than letting Glasnow give his best parting shot to those same batters.
After Seager walked, the decision to leave Glasnow in to face Turner (who he struck out) is obvious enough that I don’t think we need to talk about it.
Muncy, Smith, and Bellinger
Alert readers will notice a problem here. The projections I’m using have Glasnow better against a lefty 3rd TTO than Yarbrough against a lefty 1st TTO. That means that, assuming Glasnow is physically able to be himself, there is no point at which bringing in Yarbrough improves the Rays chances of winning.
So if your argument is that Yarbrough should come into the game, you need to make one of three contentions:
- Glasnow was tiring and no longer able to be himself. This is possible. He was eventually pulled at 112 pitches, a high for him.
- My 2.5% escalation understates the TTO penalty for Glasnow, who is basically a two pitch pitcher.
- Yarbrough is better against lefties than is represented here, perhaps because this method of distributing a split around an average projection fails for a pitcher who has no pitch type overlap between righties and lefties.
I think all of these are reasonable, although I don’t know if they’re true.
What I do know is that, once the inning got to Muncy, there was one out and runners on second and third. The leverage was a high-but-not-very 1.46, because the Dodgers win expectancy was already at 86.3%.
And the range of outcomes where the Rays wiggle out the current bind involve a strikeout. Of the two, Glasnow and Yarbrough, Glasnow is the one more likely to get that strikeout.
Also maybe Loup. Don’t forget Loup.
Conclusion
Game one was an unusual, awkward situation for Rays fans. All season long they’ve watched Kevin Cash bring in his best relievers early when the Rays are ahead, and watched him try to steal innings with lesser arms when the Rays are behind. Watching yesterday, there’s an easy impulse to interpret the fifth inning as Cash “playing for tomorrow” in the world series.
But that’s not what was going on.
Comparisons to the “quick hook” Cash had for Morton in game seven of the ALCS fail, because that quick hook actually came an inning and two thirds later, when the Rays were in a position to bring in their absolute best relievers and use them the rest of the way through the game.
That’s not possible in the fifth inning, and the definition of “highly competitive” relievers is different when the Rays are facing the Dodgers’ balanced lineup as opposed to when facing the righty-heavy Astros or Yankees. John Curtiss and Ryan Thompson are no longer on the table as “A- Bullpen” options. It’s fair to question where Castillo and Loup sit in the pecking order.
And a Rays comeback would almost by definition create high leverage situations later on in the game where Cash would want to call on Anderson or Fairbanks.
Sticking with Tyler Glasnow in the fifth inning was not a “soft punt,” or “throwing in the towel,” or anything else one might want to call it. Because of Glasnow’s quality, it was a calculated, competitive choice that very possibly gave the Rays their best shot at winning the game.
It just didn’t work.