Dodgers, Astros redefining starter use -- and winning

While the story of the 2017 Trade Deadline will mainly be about the big deals that sent starters Yu Darvish to Los Angeles and Sonny Gray to the Bronx, that's not really the underlying theme of what happened. Even with Darvish and Gray going, only a half-dozen starters changed places, and few notable bats went anywhere all month.
But with memories of Andrew Miller and Kenley Jansen dominating October fresh in their minds, teams made trades for nearly two dozen relievers in July. For every Jaime Garcia, there was Tommy Kahnle, Tony Watson, Justin Wilson and Addison Reed. For every Jeremy Hellickson, there was Pat Neshek, David Phelps, Player Page for David Robertson and Anthony Swarzak. Even the starter Houston did get, Francisco Liriano, will be deployed as a reliever.
Why? Because as we saw last October, when the games really begin to matter, the traditional roles go out the window. "Starters" and "relievers" don't matter nearly so much as "pitchers" do, because it's simply about getting 27 outs as effectively as possible. More and more, that means starters leaving early and relievers coming in often, rather than having one pitcher turn over a lineup several times -- because such a strategy no longer correlates as well to actually winning games as well as it once did.

If it seems like there's way more interest in RP than SP today, it's partially b/c going deep into games has little correlation w winning. pic.twitter.com/f6ySiNhaZM

— Mike Petriello (@mike_petriello) July 31, 2017

The pushback against expanding the use of postseason bullpen ideas to the regular season has been that while October's days off allowed teams to yank the starter earlier, the grind of the regular season wouldn't. Perhaps. But there's evidence that's changing, with baseball's two best teams, the Dodgers and Astros, at the forefront.
We talked about this nearly two years ago, when we suggested that the next wave in baseball would be that starting pitchers would throw fewer innings. At the time, despite more and more relievers entering games, that trend hadn't actually appeared. In 1995, starters threw 66.3 percent of all innings. In 2014, that number was 66.5 percent. In two decades, little had changed; while there were more relievers, they weren't throwing more innings, collectively.
As it always is, the theory is ahead of the on-field execution. Take a look at the percentage of innings thrown by starters over the last two seasons, and you'll see that now it very much has changed. In 2016, we saw the fewest share of innings going to starters. In 2017, it's lower, down to just 62.6 percent.

Think about it this way: Through Monday morning, just over 52 percent of all Major League starts this year had gone at least six innings. The Dodgers, at 46.7 percent, rank only 22nd by that measure. The Astros, at 44.2 percent, rank just 25th. Or this: Houston recently went nearly a month -- from June 17 to July 15 -- without a single game where a starting pitcher recorded an out in the seventh inning. For a month, Astros starters averaged only 5 1/3 innings per start.
For just about the entire history of baseball, a stretch like that would invariably mean disaster, that the pitchers simply weren't good enough to get further. But here, the Astros won 16 of those 23 games, good for a .696 winning percentage. In their remaining games before and after, as they steamroll the rest of the American League, they've had a .642 winning percentage. Not only did short starts not hurt them, they actually went on a run. Sure, Dallas Keuchel being injured led to some unintentional shorter starts. Granted, the Houston's best-in-baseball offense had something to say about the wins. But then again, neither team is built to have starters go deep. That's not the plan.

Really, innings thrown isn't even the best way to think about it. It's more about "times through the order," since it's long been proven that the more times a pitcher turns a lineup over, the less effective he'll be. There's a variety of reasons for that -- fatigue, certainly, but also hitter familiarity, worsening platoon advantages as pinch-hitters enter, etc. -- and it's why you see starters getting pulled early in the playoffs. Last fall, only 56 percent of innings were thrown by starters, down from 63 percent during the regular season.
Let's show you what we mean. During the 2016 season, as it is every season, the trend towards worsening production as the lineup turns over was clear. Starters get less effective each time through.
MLB, 2016
SP first time through order: .725 OPS
SP second time through order: .753 OPS
SP third time through order: .792 OPS
RP first time through order: .713 OPS
Basically, hitters the first time through the order performed like Kyle Schwarber has this year, a season in which he was demoted to the Minors. By the third time, they're Robinson Cano, a big step up. Meanwhile, no group of pitchers is more effective than relievers seeing hitters for the first time. Why, then, would you want to keep in a starter the third time through, and surrender 80 points of OPS?
Lest you think this is only something that weaker teams face, think again. Dodgers starters this year have had a .611 OPS the first time through, then .699 and .690 the second and third time, while their relievers are at .621 -- that's .210/.269/.352 -- the first time they see a hitter. It's starker for the Astros, who have seen their starters go from .634 the first time through to .753 the second time to an .803 mark the third time through. No wonder they're rarely letting pitchers go that deep.
And they aren't, remember. Having Keuchel on the disabled list surely influenced that entire month without a starter going past six innings, but the Astros (and the Dodgers) are two of the four teams who force their starters deep the least.

It's not a one-to-one relationship, of course. The Nationals have their starters going through the order a third time more than anyone, simply because they've not been able to trust their bullpen, and Washington is cruising to a division title. The Reds have baseball's highest ERA, and they're going to the bullpen early simply because they've been forced to. In order for this to work, you've got to have the depth, as Houston and Los Angeles do, and the new 10-day disabled list has made it far easier to cycle through arms.
That flexibility is important, as are multi-inning arms. For example, the Astros have Chris Devenski and Brad Peacock, each above-average pitchers who can either start or go multiple innings in relief, a role the Dodgers seem to be grooming Thomas Stripling and potentially Brock Stewart for. And each team, even before the season began, was lauded for their enviable depth. This only works if you have 15 or more pitchers you feel comfortable cycling through. Each team does, though the cracks are starting to show in Houston's bullpen, even if Liriano should help.  
"Is [the rotation] where they'd have the most value?" Astros manager A.J. Hinch said to MLB.com's Richard Justice in March about how to deploy a pitcher like Devenski, "or would it be better to have them for multiple innings two or three times a week?"
That's a question that apepars to be well-answered. It's not about starters, it's about pitchers. Put another way, the Dodgers and Astros failing to get deep appearances out of their rotation isn't a bug. It's a feature. It's October baseball, all year 'round.