Psst. I've got charts. You want charts? C'mere. Look. Look at this chart.

You like that? Oh, you don't know what that is? I should have put axis labels. Each dot represents a starting pitcher who pitched for just one team in 2012. (The restriction is because I'm bad at doing spreadsheets and databases, not for any research design reason. As you'll see, I'm not doing rigorous work, so this doesn't really matter.) On the x-axis is the overall win percentage of that pitcher's team. On the y-axis is the win percentage of the pitcher's team in his starts. (I would be happy to hear your arguments that I should have flip-flopped these axes. I mean that. I couldn't decide which of the variables I was treating as the dependent one.)

It's kind of a stupid chart, though. Why is it a stupid chart? Mainly because it includes a bunch of pitchers who only started once or twice or thrice, which junks things up. Let's try this again.

That's better! That's pitchers with at least 10 starts on the season. Look at that r squared—I almost tripled it! That's a nice, attractive fit (as baseball fits go) between whether your team wins games and whether your team wins games that you pitch. Here, have one more before I get to the actual point.

This is all pitchers again, not the pitchers with at least 10 starts. The x-axis is FRA, our basic pitcher-value stat. The y-axis is the difference between a pitcher's team's win percentage in his starts and the team's overall win percentage. (Given that high FRA appears to be correlated with a low delta, you can probably tell which direction I subtracted.)

From these three charts we can draw three related conclusions:

  1. If you pitch for a good team, your team will probably win the games that you start; but

  2. If you pitch poorly, you will not win as often as might be expected from your team's quality; and

  3. My charts are the ugliest around.

That's as advanced as I get about the nitty-gritty of analysis and charts and stats and such. If I could say whether the "findings" above are statistically significant at some level, I would, believe me, but I can't, so instead let's get to the real point, which are these guys:

And these guys:

Ready to guess who those seven players are? (I know, there are two players inside one of the stars in the second graph, but the other one isn't supposed to be there. He's an interloper. Just ignore him.)

Kris Medlen
Did you know that Atlanta won all 12 of Medlen's starts despite being just a .580 team overall? ("Just." That's a joke, see.) Did you know that Medlen didn't make a start until July 31st, having spent the year to that point in the bullpen? Can you even imagine how good the Atlantans would have been if Medlen had taken the 10 starts that Jair Jurrjens made? Here is how good: Atlanta would have won exactly 10 more games, going 104-58 for the season. They'd have avoided the Wild Card play-in game and gone straight to the NLDS while letting Washington fall victim to a dubious infield-fly call. They'd have dominated the Cardinals, Giants, and Tigers and won their first World Series since 1995 and their first World Series in which a full slate of games was played since 1957, when they still made their home in Milwaukee.

Is Atlanta to be forgiven for not realizing that Medlen's return from Tommy John surgery would go quite this swimmingly? (How swimmingly? This swimmingly: 3.37 FRA, 120 whiffs in 138 innings, other stats as well.) Here's a PITCHf/x tidbit from the awesome leaderboards: minimum 200 pitches, among starters, no pitcher got more swings on his changeup than Medlen, and just two got more whiffs per swing. Those two were Stephen Strasburg and Cole Hamels.

One notes that Medlen is still just 27.

Jamie Moyer
I don't really want to pick on Jamie Moyer, but the 49-year-old made 10 starts (i.e. just hitting the minimum for inclusion in the chart above) for the Colorado Rockies and saw his team win two of them. Sure, the Mountains won fewer than 40 percent of their games overall, but two out of 10 pretty well blows. Twenty percent.

Moyer had a superficially good April, posting a 3.14 ERA in five starts. He managed to give up five unearned runs, though, so his RA/9 was 4.71. (The earned/unearned distinction deserved to die along with our regard for saves and wins and RBI as measures of value, by the way, a fact I'm likely to shoehorn into every article I write here until we all stop using it.) That RA/9 is not horrendous for the oldest pitcher in memory throwing in Colorado in front of a defense that wound up with the lowest defensive efficiency and the lowest park-adjusted defensive efficiency in baseball by a huge margin—as an illustration, the gap in DE from Colorado to the 29th-best defense is the same as the gap from 24th to 11th.

Moyer gave up just one "unearned" run in May. Unfortunately, he allowed 24 "earned" ones in 25 innings. One suspects that the only thing keeping him from having an even worse June is that Colorado released him. Moyer latched on with the Orioles and Blue Jays, but didn't pitch in the majors with either. He wouldn't have pitched better, mostly likely, but he wouldn't have left his team with losses in 20 percent of his games, either—give Moyer enough starts and he might well have pitched his way right out of outlier status.

Factoid: Moyer averaged under 80 mph with his fastball in 2012. To reiterate: Moyer threw about as hard as a good adult-leaguer and still managed a 90 FRA+ in the major leagues. (FRA+, for the record, is like ERA+ but with FRA as the base stat. As you probably guessed. And no, a 90 FRA+ isn't All-Star territory, but it's perfectly respectable and not "why on earth would they even let him play"-level pitching.)

One wonders whether baseball is not actually as hard as everyone claims it is.

Randall Delgado
I mentioned Kris Medlen taking Jair Jurrjens's starts above, but maybe Randall Delgado, whose Atlanta homeys went 5-12 in his starts, should have been the replacee instead. Well, except for the whole "Delgado was a really good prospect and had an 89 FRA+ and saw his team lose games in which he gave up two runs or fewer four different times" thing. Sadly, none of this was enough to keep him from being demoted.

One would be forgiven for pondering whether Delgado's service time (which currently stands at 165 days—it takes 172 to make a full year) might have something to do with his demotion.

Edwin Jackson
This is my favorite outlier of all, though that's not very empathetic to the plight of Edwin Jackson. He struck out eight men per nine, walked fewer than three, had a BABIP of .278, got groundballs … and saw his .605 winning-percentage team go 12-19 in his starts. How does that happen? How about this: 10 quality starts that his team lost anyway and 3.94 runs-scored-per-nine in his starts despite Washington having the league's fifth-highest-scoring offense.

Jackson is entering his age-29 season and has pitched for seven different major-league teams in his career. By sheer operation of odds, there's at least a decent chance that he won't make it number eight next year. (A seven out of 30 chance, maybe. But not really.) One only hopes that Washington's refusal to support Jackson doesn't cost him any dollars on his free-agent contract.

Julio Teheran
Do you want more ATLers? Here's another ATLer. Teheran's team lost his only big-league start despite his 2.65 FRA, though to be fair, one third of Teheran's innings came in relief, i.e. he threw a start for 4 1/3 frames and a relief appearance for two. None of this is meaningful. It was 6 1/3 innings. Teheran is still a very good prospect.

One should just leave things there because one would not want to force one's schtick when one has nothing else to say.

Shelby Miller
Here's the thing about Delgado, Teheran, and Miller all ending up in this article: I didn't do this on purpose. I swear I drew the hearts on those graphs and then went looking for who the dots represented, not the other way around. It's sheer luck that some of the best pitching prospects around wound up being off by themselves away from the trend-line.

Anyway, Miller, like Teheran, made one start and had an FRA that commences with the numeral two. (Then a dot.) The difference is that the Cardinals won Miller's start, a six-inning, one-hit affair on the last day of the regular season. Miller also doubled Teheran's innings overall, throwing 13 2/3 over six appearances and just once allowing a human to touch home plate in a way that benefited that human's baseball squad. (That means "scored a run." No, I can't just say that.)

One has, to the relief of all and sundry, given up on one's thematic third-person game-playing by this point in the article.

Aneury Rodriguez
Also a one-start wonder, also a youngster, also a National League pitcher, not also a top prospect. Rodriguez was a Rule 5 player in 2011 and spent 2012 putting up a 6.60 ERA for Oklahoma City, which is a minor-league team of class Triple-A in the Pacific Coast League. Rodriguez struck out six in six innings of a May start against Miami, but gave up two homers, and the batted-ball profile was not promising. The reason he makes this list is because both dongers were solo shots, Houston's bullpen did its job, and the offense eked out three runs, so he won a game on a bad team with bad stats. That's a way to stand out. Way to go, Aneury.

One is through. Really.

For what it's worth, and you'll have to make do without a graph on this one, here are a few notable outliers among players with a more substantial number of starts:

  • Drew Hutchison earned a 5.38 FRA, but the Blue Jays won 7 of his eleven starts despite being .451 overall.

  • Scott Diamond's Twins were a game over .500 in his 27 starts despite a .407 overall winning percentage and Diamond's 4.94 FRA.

  • Matt Harvey's late-season callup saw the Mets post a 3-7 record compared to a mediocre-but-not-.300 overall winning percentage (.457) despite Harvey's quite nice 3.88 FRA.

  • The Cardinals won just eight of Jaime Garcia's 20 starts, compared to .543 overall, with Garcia posting a sterling 3.57 FRA.

I'd like to sum up with a grand unifying theme, one that shakes the foundations of sabermetrics to its core: I've found pitchers that just know how to win and pitchers that aren't winners. Right? No, of course that's not what I've found at all. I've found hard-luck guys and small-sample guys and one guy who just happened to be literally unbeatable, which is another way of saying that I've found what you'd find every year if you went through an exercise like this. I'd like to think this is not valueless, though, as the suggestion that team wins and losses are relatively closely but still far from wholly related to individual pitching is something that can stand to be reconfirmed every once in a while in a non-polemical, non-shouty way.

Especially since we may be on the cusp of Jack Morris gaining entry to the Hall of Fame. Cough.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
Terrific article, but I'm not sure how much of an outlier Jaime Garcia really was. His home-road splits are dramatic: 3.01 xFIP at home, 3.40 on the road.

I don't know how that translated to wins and losses for the Cards, but it was something the fans and media talked about every time he lost on the road.
First, this depends on your choice of metric. Garcia actually had identical FRA at home and on the road.

Second, I think dramatic oversells it. The league put up a 3.90 xFIP at home and 4.14 on the road. An extra 0.15 runs (or "runs" I guess since we're in estimator territory) doesn't strike me as a big deal.

In any case, a 3-8 road record vs. 5-4 at home does explain, in a sense, why his team's record was so bad when he pitched, but limiting him to the road still makes him an outlier -- a 3.40 xFIP is excellent! You'd expect a very good team like the Cardinals to go more than 3-8 when their starter gives them a 3.40 xFIP.
Thanks Jason. You're right that I wasn't considering league-wide context. Still, if you were a Cardinals fan who watched some of those games ...
As a minor enhancement, you should look at each pitcher vs. their team's win% only in games that the pitcher did not start. As is, you're double-counting, which reduces the size of the effect you're looking for. (I propose calling this "the 1972 Steve Carlton effect".)

For example, Kris Medlen is more of an outlier than you have him. Atlanta was only a 82-68 (.547) team on days when Medlen didn't start.
That's a good point. I should have subtracted the pitcher's games from the team's record. Darn.