Postseason Starting Pitching is Back, Baby!

The Astros took a two-games-to-none lead over the Yankees in the American League Championship Series on Thursday night, as Alex Bregman’s three-run homer backed seven strong innings by Framber Valdez, whose two runs allowed counted as unearned due to his two-error play on a Giancarlo Stanton chopper. It was the second night in a row that an Astros starter stifled the Yankees, as Justin Verlander held them to one run over six innings while striking out 11.

The Yankees had Verlander on the ropes, forcing the 39-year-old righty to throw 66 pitches in the first three innings, during which he allowed six baserunners including a solo homer by Harrison Bader. Verlander got the strikeouts he needed to escape those big jams, however, and soon he went on a roll, striking out six straight hitters (tying a postseason record he already shared) and nine of his final 11 — reclaiming the all-time postseason strikeout lead along the way — before yielding to the bullpen, which continued to dominate Yankees hitters in a 4-2 victory in the ALCS opener.

Even given that he got knocked around for six runs and 10 hits in four innings by the Mariners in his Division Series start, Verlander probably had a longer leash than most due to his Hall of Fame resumé and manager Dusty Baker’s trust in him. He also had the wind at his back, so to speak. On the heels of a regular season in which per-game scoring fell 5.5% (from 4.53 runs per game to 4.28), in which the league-wide OPS decreased for the third straight season (from .758 in 2019 to .740 in ’20, .728 in ’21, and .706 this year), and in which starting pitcher usage increased for the second year in a row, starters are working deeper into games in the postseason than at any time since 2015.

I’ve been tracking this trend in recent years, noting the extent to which starting pitcher usage parallels what we’ve seen in the regular season. And after some lean years of what I called “The Incredible Shrinking Postseason Starter,” what we’ve seen so far this October is vaguely encouraging.

That’s true for the regular season as well. From 2015-20, starting pitcher usage declined annually, to the point that during the pandemic-shortened season — which had just a three-week “summer camp” preseason after the COVID layoff — the length of the average start slipped below five innings for the first time:

Since 2020, starters have regained some ground, though teams have been cautious about ramping up their usage since the shortened season; making a year-over-year jump from 60 innings to 180 or 200 isn’t something to be taken lightly even in the best of times. Last year, the innings per start average was back over 5.0, but just barely. This year, it increased to a level that was higher than 2019, and that’s even with another compressed preseason schedule, this time due to the lockout. Because pitchers didn’t have a full spring to build up to 90 or 100 pitches, they generally worked shorter outings at the beginning of the year; only three times did a pitcher reach 100 pitches in the first two weeks after the season began on April 7. Sticking with innings as our yardstick, we can see the buildup reflected in the monthly splits:

I’ve shown 2021 for comparison’s sake, and you can see that last year, starting pitcher usage was more level month to month. This year, it increased from 4.8 innings per turn in April to 5.25 in May, after which it climbed a bit, peaking in August before dropping to 5.07 in September, as workload concerns, late-season call-ups, and preparation for the postseason all played their part in shortening the average start.

One more factor that had an effect on starter workloads during the regular season was the decreased usage of openers and bullpen games. Defining these for the purposes of quantification is kind of tricky, but we can get ourselves in the ballpark without too much trouble. Using Baseball Reference, I tallied the number of starts that lasted one or fewer innings, two or fewer innings, and three or fewer innings — you’re free to take your pick as to which you find the most meaningful — then converted those into percentages; this winds up overestimating the share of such starts because it captures those that went short due to injury or ineffectiveness, but I think it illustrates the trend well enough for our purposes:

However we define it, the trend peaked in 2020 and is right back to ’18 levels, more or less, which is to say that 2% of starts lasted one inning or less, 4.7% lasted two innings or fewer, and 9.6% lasted three innings or fewer.

Having gotten the lay of the land as far as the regular season is concerned, we turn to the postseason, updated through the first two games of each league’s LCS:

Look at that! Starters averaged fewer than five innings in four Octobers out of five from 2017-21, but they’re back up above five this month, and closer to the regular season average than any other season in this range.

Last year, the postseason average even fell below four innings per start for a variety of reasons. The Astros made it all the way to Game 6 of the World Series despite a 6.33 ERA from their rotation, with eight of their 16 starts lasting three innings or fewer not because Dusty Baker was going opener-crazy but because some of his starters were downright terrible; five times, a Houston starter allowed five or more runs and pitched three or fewer innings, and all eight times one of their starters worked short, their runs allowed total exceeded their innings total. The Braves used Dylan Lee and Jesse Chavez as openers and additionally had to work around Charlie Morton’s broken leg in Game 1 of the World Series. And due to the late-season injury of Clayton Kershaw and some questionable usage of starters in relief, the Dodgers opened with Joe Kelly and Corey Knebel, with the latter even starting back-to-back games in Game 5 of the NLDS and Game 1 of the NLCS. It wasn’t a great look for anyone.

Thus far this year, only seven starts have gone three innings or fewer, with Noah Syndergaard’s three-inning start in Game 4 of the NLDS for the Phillies the only one that was even vaguely opener-ish (10 batters faced, blemished only by a solo homer) as opposed to the result of either an injury (Morton, again) or a crummy performance worthy of a quick hook.

As for the actual performances of this year’s postseason starters, as a group they’ve shaved almost a full run off their ERAs relative to last year:

Relative to last year, starters are facing 20% more hitters, throwing 23% more pitches… and pitching 31% more innings. Their workloads by any of those three measures are higher not only than in 2020 and ’21, but also ’17 and ’18. Again relative to last year, they’re allowing slightly more home runs (1.29 per nine, up from 1.23) but fewer hits (.220 AVG, down from .246), walking far fewer hitters (6.5%, down from 8.8%) and striking out more (25.4%, up from 24.1%). The end result is starters’ lowest FIP of this eight-season period and their second-lowest ERA. They’re throwing quality starts (at least six innings pitched, with three or fewer earned runs allowed) 38% of the time, which is more than double last year’s rate of 18%, and one point off 2019 for the highest mark since ’15. And as for those short starts, we’re seeing fewer of them than in any season since 2016, a year that predates the rise of the opener as a commonly used strategy. Even lumping openers together with bullpen games, injuries, and times when a starter just gets shellacked, such games are happening at about one-third the frequency of last year and half the frequency of 2020.

As for what’s driving all of this, I think the answer is a combination of fewer injury/workload issues now that we’re no longer adjacent to the shortened 2020 season, and the lower levels of offense. In this year’s postseason, teams are scoring just 3.59 runs per game, the lowest mark in the 2015-22 period under examination here, down from 4.28 last year and the local high of 4.40 in ’20. Managers are still pulling their starters relatively quickly if they’re giving up runs – note that average of 2.21 runs per start in the far right column of the table above, and how little those annual averages vary relative to ERA — but the starters are lasting longer before getting the hook.

As for whose starters have done the best, we more or less know that simply by which teams are left standing, but in case you haven’t looked, here they are, ranked by innings per start (the table is sortable):

All told, while there are reasons to quibble regarding the new postseason format as well as the general trends we’ve observed in recent regular seasons — too many strikeouts and not enough balls in play being the primary ones — the restoration of starting pitching to a position of prominence is a very welcome development. After all, the starters are the marquee players for each game; MLB officially bills NLCS Game 3 as pitting Joe Musgrove against Ranger Suárez, and ALCS Game 3 will feature Gerrit Cole against Cristian Javier (Lance McCullers Jr., who started ALDS Game 2, suffered a weird champagne-related mishap that somehow has nothing to do with hangovers and has been pushed back a day). Not every starter is a household name, and the deGroms, Kershaws, and Scherzers have gone home for the winter, but casual viewers are certainly more familiar with the Verlanders, Musgroves, Nolas, and Coles than they are the Knebels, Lees, and A.J. Minters whose elevation to starter status advertises a forthcoming slog containing at least half a dozen pitching changes “starring” lots of less-familiar hurlers who may nonetheless dominate opposing hitters. The game is simply more engaging thanks to the familiar faces on the mound and the continuity they provide. Welcome back, starters.