All hands on deck: Starters pitching in relief

If there’s a Big Narrative of this postseason, it is clearly found in the unconventional way that starting pitchers have been used. After all, starters have thrown only 46% of postseason innings, easily a new all-time low. “Openers” are common, and starting pitchers, even accomplished ones, being yanked in the fourth inning is more an expectation than an exception.

That’s not exactly the plan -- several starters have been injured, or ineffective, and surely the effects of a full season’s workload following 2020’s shortened year have something to say about all of this -- but there’s something different going on, too. It’s not just about starters being lifted from games early. It’s about them entering late, as relievers, in situations where they would never appear in the regular season, often on their “throw day.”

This has come with mixed results, and often a price to pay in the next start, because they’re not at full strength -- or so it seems.

“It’s just a challenge on your arm,” said future Hall of Famer Max Scherzer on the TBS broadcast during the National League Championship Series. “We’re so used to going out there and having a routine every five days.”

Scherzer entered in relief to finish out Game 5 of the NL Division Series against the Giants, then lasted only 79 pitches three days later in a 5-4 loss to Atlanta in NLCS Game 2. He admitted afterwards that his “arm was dead … I could tell when I was warming up that it was still tired.” Worse still, Scherzer was later deemed unavailable to make his Game 6 start entirely, forcing Walker Buehler to start on short rest in a 4-2 loss that sent the Braves to the World Series.

“There is a cost,” Dodgers manager Dave Roberts said after Game 2, and he might have learned that lesson in more ways than one. Later in the same game, he brought in starter Julio Urías in the eighth inning. That appearance did not go well; worse, when Urías returned on two days' rest to start Game 4, he clearly didn’t have his best stuff and allowed three home runs, putting the Dodgers in a 3-1 series hole.

That all seems bad. But on the other hand, Boston’s Nathan Eovaldi also started on two days' rest on Friday after having appeared in relief, just like Scherzer did, and he was quite effective, allowing just one run while working into the fifth.

So: Is there a cost? How much of one? What does it mean to (potentially) increase your chances in the late innings of one big game while (possibly) injuring your chances in the early innings of a different big game?

To find out what might be ahead, let’s look backwards. Let’s try to find as many similar instances of a starting pitcher appearing in relief in the playoffs as we can.

1) Is this really happening more often?

You might wonder why we’re bothering with this part. You have eyes, you can see it happening right in front of you. Of course it’s happening more. But is it, really? It depends on how you look at it.

Answering that, in itself, proved tricky. What’s a “starting pitcher,” anyway? We settled on these three rules, excluding 2020 because the schedule was compressed, there were no travel days and just the general 2020-ness of it all.

• The pitcher must have A) started at least five games in a season and B) started at least 70% of the games they appeared in during that season, to try to avoid openers.

• The pitcher had to have started at least one postseason game after appearing in relief, partially to eliminate the back-end regular-season starters who were never going to be in the playoff rotation anyway, but mostly because the entire point of this is what happens in your next start? This also served to eliminate starters who threw the final pitch of their team’s season -- like Madison Bumgarner in 2014, or Charlie Morton in '17, or Chris Sale in '18, because there’s no “next” anything to worry about after that.

• That post-relief start had to have come on three days of rest or less, because things like Don Robinson (who had made 32 regular-season starts) getting to start on 20 days' rest in the earthquake-delayed 1989 World Series hardly seems to count.

That last bit is somewhat imperfect, because it’s not accounting for the added cost of a relief appearance potentially pushing back your start entirely, elevating another pitcher to start instead -- like what just happened to Scherzer -- but that’s a little more than we can accommodate for here. (A good example of this was in 1999, when John Smoltz, relieving on a throw day for the second time that October, retired just one of the five Mets he faced in Game 6 of the 1999 NLCS, then did not start a World Series game until Game 4, a week later.)

Taking all that into account, we were left with 67 pitchers who entered in relief, then started on three days of rest or less immediately after, comprising 80 times. (Rick Porcello somehow did this four times over the years.)

It has happened 31 times since 2010. It happened 50 times over the previous century. So yes, it’s more, kind of, but it’s also a trick question, because there are simply more playoff games now. Before 1969, the league champions went directly to the World Series. The first Division Series came about in 1995, and then the second Wild Card arrived in 2012. There were seven postseason games in 1968, for example. There were 38 in 2017.

So taken as a percentage of all postseason games, is it happening more often? It’s only happened in 5% of this year’s games, which is not unusual. But it’s happened at least once every year for the last 11 (non-2020) postseasons, which is.

(A special note here for the 1985 Kansas City Royals, who were well ahead of their time. They had five starters make 158 of their starts that season, yet in the playoffs, four of them -- all but ace Bret Saberhagen -- made relief appearances.)

It’s not like starters never pitched in relief in the playoffs before 2010, anyway. In 1988, NL Cy Young Award winner Orel Hershiser got the save in Game 4 of the NLCS. Three days later, Dwight Gooden entered to relieve an ineffective Ron Darling in Game 7 … while Hershiser, on two days' rest, was pitching a shutout. There are similar examples throughout history, it’s just that for the first seven decades or so of modern baseball history, the regular-season champs went directly to the World Series. So if this is happening more, it’s at least partially because there are more opportunities for it to.

2) Did the starters pitching in relief even help?

They’d better have, right? If you bring your starter in, risking his ability to make a next start at full strength, you’d better hope he’s making it worth your effort. For example, it worked out incredibly well in the 2019 NL Wild Card Game, when Stephen Strasburg pitched three scoreless innings of relief to follow Scherzer. It wasn’t quite so tidy in Game 4 of the ALCS this year, when Eovaldi entered a tied ninth inning and along with Ryan Brasier helped turn a 2-2 tie into a 9-2 Astros victory.

For the most part, though, the answer is: The starters relieving were quite good. Our starters-as-relievers posted these numbers throughout the years:

• 436 PA
• .263 OBP
• .564 OPS
• Only 6 home runs -- and 2 of those came back in the 1923 World Series

It’s difficult to know exactly which reliever would have entered instead of that starter on that day, but generally speaking, relievers in the postseason all-time have allowed a .316 OBP and a .685 OPS, so this is better. (As you’d expect; top starters are top starters for a reason.)

More recently, since 2010, (again excluding 2020), it’s looked like this:

• 137 PA
• .272 OBP
• .560 OPS
• 2 home runs

All relievers since 2010 have allowed a .305 OBP and a .677 OPS, so our starters have performed quite well.

The recent misadventures of Urías and Eovaldi aside, the starters-as-reliever gambit has generally paid off, which isn’t terribly unexpected; after all, the types of pitchers who do this are usually elite names, like Smoltz, or Scherzer, or Strasburg, or Sale, or Noah Syndergaard, or Catfish Hunter or so on.

You might remember Jon Lester in 2017, when the Game 2 starter threw 3 2/3 quality relief innings in Game 4, or Smoltz finishing off Game 2 of the 1999 NLCS, or Hunter replacing Rollie Fingers to finish off Game 1 of the 1974 World Series. You probably don’t remember Cubs starter Hank Borowy throwing four innings of scoreless relief in Game 6 of the 1945 World Series in between starting Games 5 and 7, but it happened, in a considerably different time.

“It’s the playoffs,” Scherzer said on the broadcast. “You want to be in those situations. You want to be out there with the ball in your hands. So even though it’s a challenge, you welcome it.”

3) What have the outcomes of the next start looked like?

What you really care about, though, is what it means for the next start. You care about if using, say, Urías over Justin Bruihl was worth what it cost the Dodgers in his poor start in Game 4.

Remember how we said you didn’t remember Borowy? You probably also didn’t remember, then, that after those four scoreless relief innings in 1945’s Game 6, he started Game 7 on one day of rest … and promptly allowed hits to the first three batters he faced before being removed. It was the closest the Cubs would come to a ring for another seven decades.

We don't have velocity or pitch-level data throughout history, but it makes sense enough that an incredibly high-pressure relief appearance is somewhat more stressful than throwing in the bullpen.

“I had no idea I could shut them out on only two days’ rest,” Hershiser said about his start against the Mets in 1988. “My mechanics were very bad for about the first two or three innings. Finally I got into a groove and made some adjustments.”

This time, we’ll look at it differently, because unlike the uncertainty over which reliever a starter might be replacing, here we have a much easier comparison point to make. We can look at our 67 starters on three-days-or-less rest after a postseason relief appearance, and we can compare them to themselves in all other postseason starts.

Starts on three days of rest after a relief appearance:

Innings per start: 5.8
Runs allowed/9: 3.71

All other postseason starts by same pitchers:

Innings per start: 6.0
Runs allowed/9: 4.08

That’s all-time, though. If we look just since 2010, the gap does become almost imperceptible. The pitchers on short rest? 4.24 RA/9 over 5.1 innings/start. On regular rest? 4.20 RA/9 over 5.6 innings/start. It might not be better, really not as the all-time numbers would indicate. But it certainly doesn’t seem worse, does it?

Pitchers who have started after relieving like this go slightly shorter after that quick turnaround, which is expected, and don’t allow more runs. That is surprising, though maybe it shouldn’t be; there are tons of examples of great short rest starts throughout history. Eovaldi, as we said, was good on Friday.

Even within our group of pitchers, you can cherry-pick some examples. Take Scherzer, who admitted he felt affected by the maneuver earlier this month. But in 2019, he did something similar, throwing a dominant inning of relief in Game 2 of the NLDS (striking out the side); then on two days' rest in Game 4, he threw seven one-run innings. In 1997, Denny Neagle threw three innings of scoreless relief in Game 1 of the NLCS, then threw a shutout against the Marlins on three days' rest in Game 4.

It doesn’t always work out that cleanly, of course, with massive highs and lows. But in the aggregate, and relatively few examples we have to look at it, it doesn’t seem like it’s affected these pitchers as much as you would have thought, and the managers get the benefit of their talent in their relief appearances, too.

Importantly, though, remember what we’re looking at here. We are not looking at “starters on short rest.” We are looking at “starters on short rest after a relief appearance,” which really means we’re looking at only the best starters in the first place.

We can do one more thing, though. Something more telling.

4) When have those starters begun to hit the wall?

Scherzer talked about his arm feeling tired, and it wasn’t hard to see it in the data. You could see it in real time, that his velocity had begun to dip. He's also 37, and only two pitchers have thrown more pitches than he has (more than 43,000) since pitch tracking came online in 2008.

But it’s not always about velocity. In fact, in our group of pitchers, looking inning by inning, you can’t really see much difference. Instead, it’s about something harder to quantify. It’s about command, or control. It’s about feel.

"I felt like the ball was coming out good, so no issues on that part," said then-Arizona starter Robbie Ray in 2017 after walking four Dodgers, hitting a fifth, and uncorking three wild pitches in Game 2 of the NLDS on two days' rest after relieving. "I was just a little bit all over the place."

Kershaw, speaking after a rough outing on short rest in Game 1 of the 2018 World Series, didn’t worry about velocity so much as he did movement. “[The] slider wasn’t very good tonight,” he said. “"Didn't have the depth, kind of flat in the zone, and they made me pay for it. All the way around, wasn't a great night.” He allowed 10 hitters to reach in four innings.

So instead of looking at velocity, we’ll look for effectiveness. Let’s check into inning-by-inning swing-and-miss rate. For this, we’ll stick to only appearances since 2010, which leaves us with 20 pitchers who have started a postseason game on short rest after a relief appearance.

Again, these are mostly top guys, like Scherzer, Sale, Roy Oswalt and so on.

For the first three-plus innings, there’s really no difference between a short start after a relief appearance and a long one. (That each are lower than the regular season is to be expected, since the quality of hitter you’re facing in the playoffs is better.)

But while the regular season swing-and-miss holds steady ... and the regular rest postseason starts decline slightly after the fifth inning ... the short rest starts collapse. This is the thing to watch for.

Since only three of our starters failed to complete four innings, that’s a reflection of performance, not availability, for the next three innings, though no one has managed to get a single out in the eighth inning in one of these games in years.

So that, maybe, is your takeaway. If you have a top-end pitcher, you might do well to get some outs in relief in a big spot, because that starter is likely better than the reliever you might otherwise use. If you’re comfortable with that starter going only four or five innings in their next start, then you might not have lost much. But if you had hopes for going deeper than that, you’ve probably cost yourself that opportunity.

It doesn’t always work out. We just remember those times a lot more than the ones that do.