Does Quality Start Percentage Predict Anything?

Short answer to title question: apparently not. But as onetime academics, we hate nothing more than doing even cursory research that doesn’t produce an article, however uninteresting or intuitive its conclusions. Thus, as part of our tireless quest for fantasy-baseball-pundit tenure, whatever that may consist of, here goes:

The prevailing definition of a “Quality Start” is one in which the starting pitcher goes at least six innings and gives up no more than three runs. Lots of people, us included, think the definition should change. We’d go with five innings and two runs, and we wouldn’t deprive the pitcher of a QS if he goes longer and gives up more runs. We figure it’s the manager’s fault if the bullpen doesn’t take over soon enough. But for the moment, we’ve got to use the readily-available stat, and that’s six and three.

So imagine two starting pitchers, both of whom get 30 starts in a season. Pitcher A pitches exactly six innings every time out. In 15 of those outings, he gives up three runs. In the other 15, he gives up four runs. He’s thrown 180 innings, his ERA for the season is 3.50, and he has 15 Quality Starts, for a relatively low QS% of 50%. Pitcher B has a different kind of season. In five games, he goes 8 innings and gives up 2 runs. In fifteen games, he also goes 8 innings and gives up 3 runs. But in 10 starts, he runs into trouble and gives up five runs in two innings before the manager takes him out of his misery. Like Pitcher A, he will have gone 180 innings with a 3.50 ERA for the year, but Pitcher B will have had 20 Quality Starts, for a relatively high QS% of 66.7%. So—all else being equal, which of course it never is—which of them do you want next season?

We’ve always figured, without thinking very hard about it, that B’s the man. The numbers suggest that his stuff, when he’s got it, is better than A’s, and that A has further to go to have a high-quality or breakout season. But is that the case?

We wondered, so here’s what we did. Starting with 2010, we looked at each season’s ERA qualifiers, and identified (1) pitchers who were in the top third in ERA but the bottom third in QS%; and (2) pitchers who were in the bottom third in ERA but the top third in QS%. We sort of expected that the latter guys—the Pitcher Bs of the world—would do better in the following season than the A cohort.

Nope. But the opposite isn’t true either. Here are the results, with won-lost record, ERA, and WHIP in the following season in parentheses:

2010: Type A: Tommy Hanson (11-7, 3.60, 1.17), Jonathan Sanchez (4-7. 4.26, 1.44)
Type B: None

2011: Type A: Kyle Lohse (16-3, 2.86, 1.09)
Type B: Josh Tomlin (5-8. 6.36, 1.46)

2012: Type A: Ross Detwiler (2-7, 4.04, 1.49)
Type B: Joe Saunders (11-16, 5.26, 1.60)

2013: Type A: Shelby Miller (10-9, 3.74, 1.27), Ubaldo Jimenez (6-9, 4.81, 1.52)
Type B: Matt Cain (2-7, 4.18, 1.25)

2014: Type A: None.
Type B: Scott Feldman (5-5, 3.90, 1.31), John Lackey (13-10, 2.77, 1.21)

2015: Type A: Scott Kazmir (10-6, 4.56, 1.36), Yovani Gallardo (6-8, 5.42, 1.58)
Type B: None

2016: Type A: Kenta Maeda (13-6, 4.22, 1.15)
Type B: Dallas Keuchel (14-5, 2.90, 1.29)

2017: Type A: Drew Pomeranz (2-6, 6.08. 1.75), Jake Arrieta (10-11, 3.96, 1.29)
Type B: Julio Teheran (9-9, 3.94, 1.17), Jeff Samardzija (1-5, 6.25, 1.63)

2018: Type A: Mike Foltynewicz (2-5, 6.37, 1.42 so far), Jon Lester (7-5, 3.83, 1.31)
Type B: James Shields (out of work), Mike Leake (7-6, 4.54, 1.27)

We don’t see a pattern here. If you do, let us know. It looks to us like the collection of delightful surprises, meltdowns, so-so seasons, and pretty-much-as-expecteds that you’d find in any group of pitchers sturdy enough to have thrown 162 innings the year before. We’re glad we investigated, but a little disappointed.

And, as you’re no doubt wondering if you’ve stayed with us this far:

2019: Type A: Sandy Alcantara, Kyle Davies, Jake Odorizzi, Spencer Turnbull, Tanner Roark

Type B: Madison Bumgarner, Lance Lynn, Mike Leake, Joe Musgrove

It will be interesting to see if there are significant differences between the two groups in the second half. But we doubt it.

We hoped you liked reading Does Quality Start Percentage Predict Anything? by The Birchwood Brothers!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




The Birchwood Brothers are two guys with the improbable surname of Smirlock. Michael, the younger brother, brings his skills as a former Professor of Economics to bear on baseball statistics. Dan, the older brother, brings his skills as a former college English professor and recently-retired lawyer to bear on his brother's delphic mutterings. They seek to delight and instruct. They tweet when the spirit moves them @birchwoodbroth2.

newest oldest most voted
ccoville
Member
ccoville

All you need to know about QS is that if you go 8 or 9 innings and give up 4 runs it isn’t considered a QS. Definition needs to be changed.

ChetManley
Member
ChetManley

I’ve often thought if the stat was simply renamed to “Starts of 6+ IP and 3 or fewer ER” people wouldn’t get so dang hung up on trying to objectively define the subjective term “quality” or arguing about the edge cases. It’s a simple stat that measures two thresholds. Of course, So6IPa3oFER isn’t a great stat name, so instead 90% of the conversation around QS is complaining that quality starts don’t match the speaker’s personal definition of quality.

bmt24
Member
bmt24

In many cases you would be right- people get really caught up in names- but not here. The outrage over Quality starts is due to the fact that players with better performances than those meeting the minimum standard get left off due to arbitrary endpoints.

ChetManley
Member
ChetManley

Well if we changed the name to So6IPa3oFER, then the end points wouldn’t be so arbitrary… they’d be right there in the name.

And again… “better performances” is drifting into subjectiveland. Is 5 IP/2 ER better than 6/3? Is 6/3 better than 8/4? I dunno. It depends. And that’s the point.