Monday, June 13, 2011

John Lannan: He Is What He Is, But He Wasn't What He Was

I have a confession. I picked up John Lannan in my fantasy baseball league. Now, this is an unusually deep 20-team league, and I had Brett Anderson, Jake Peavy, and Brandon McCarthy all on the DL, Colby Lewis in the tank, and the best other options available were Jason Marquis, Chris Narveson, and Tom Gorzelanny. Still, I feel I owe an explanation.

You see, as some may recall, in 2009, I confidently predicted with a couple months left in the season that there was no way that John Lannan would finish with 200 innings pitched an an ERA under 4.00.

Obviously, I was wrong, and Lannan's continued success has made my skepticism seem even less well founded. But the fact is that I was right to doubt Lannan back in 2009, because he was succeeding then in a spectacularly unsustainable way.

Check it out. That year, Lannan threw 206.1 innings with a 3.88 ERA and coincidentally also a 3.88 K/9 rate. Here's the list of pitchers who have thrown that many innings with an ERA and K-rate that low since 1993, the unofficial start of the juiced ball era (or whatever you call it), year baseball baseball expanded and runs per game jumped from 4.12 to 4.60: Lannan, Mark Gubicza (1995), and Chien Ming Wang (2006).

That's three pitchers. Obviously none of them ever repeated the feat. But that doesn't really show how unusual this was in this era. Consider that there were a total of 356 seasons from 1993 through last year that a pitcher threw that many innings with an ERA of 3.88 or better--this extremely low strikeout group represents just .08% of these seasons.

Also consider this: there were only 3 other seasons in this period in which a pitcher threw 206 innings with a K/9 rate of 3.88 or lower: Jimmy Anderson in 2001 with an ERA of 5.10, Mike Moore in 1993 with an ERA of 5.22, and Jamey Wright in 1998 with an ERA of 5.67.

There's a word statisticians use for an event that occurs this infrequently: it's an outlier. It's also known as an aberration, a fluke. Unless John Lannan had discovered some heretofore completely unknown skill, there was no way he would be able to sustain the favorable outcomes he was getting with the skill set he was demonstrating at the time.

Here's the good news: Lannan's skills have gotten better. Back in 2006, he whiffed just 10.2% of batters faced--an unbelievably low number. The NL average that year was 18.4%. This year he's striking out 12.6%. That's still very low, and may not seem like a big difference, but it's meaningful.

(Pause for a statistical public annoyance announcement. I'm delighted to see how strikeout and walk rates are becoming more prominently noticed by average fans and mainstream baseball media. However, why oh why has the standard rate stat become K/9? I used it here only because Baseball Reference's Play Index only offers K/9 as a search function, and I'm still too lazy to install the raw SQL database on my own computer. When looking at strikeout rate, we want to know how often a pitcher misses bats. That's the skill we're trying to measure. Strikeouts per inning essentially tells us what percentage of a pitcher's outs are strikeouts, but totally ignores all the other outcomes a pitcher gets. Hits, walks, HBPs... these are all tossed out because those events aren't counted as parts of an inning. Only outs count towards the accumulation of innings. In effect, by using K/9, we diminish the difference between the best pitchers and the worst pitchers, making it harder for analysts and observers to notice the gap. So memo to Fangraphs, Play Index, etc. etc.: kill K/9. Just give us SO/TBF.)

Translating back to K/9 so we can search the Baseball Reference Pitcher Season Finder, we find that here were 23 pitchers from 1993 to 2010 who went 206 innings with an ERA of 3.88 and 4.86 K/9, Lannan's rate this year. That's still just 6.5% of these seasons, but we're getting into a range that's at least has some precedent for being repeatable. Carlos Perez did it twice. Tom Glavine did it four times.

And the other key skills these pitchers all possess are low walk rates and high groundball rates, and Lannan's 52.7% groundball rate is 11th among qualified NL starters.

If there's an area to be concerned about, it's Lannan's walks. His rate this season has risen from a good, not great 7.6% to a potentially troublesome 9.4%. Only 87 innings into the season, I think it's reasonably likely that he shaves that walk rate back down to where he's been the last two seasons and can maintain the groundball and strikeout rates where they are.

If that's the case, then he's a guy you can reasonably project for 200 innings with an ERA usually in the 4s, and once in a while, when luck and defense are on his side, an ERA in the 3s.

His current 3.52 ERA still strikes me as better than his skills justify, but it's possible that we've entered a new, post juicing era where this is what guys like Lannan can do. After all, from the start of the expansion era in 1961 to 1992, 118 pitchers went 206 IP with ERAs and K/9s of 3.88 or under. The game was different in many ways back then, but with scoring down two years running, we may be in a new era where guys like John Lannan survive and even thrive.

My battered fantasy team at least is hoping that's the case.


Kevin Rusch said...

I've heard a lot of people say Lannan stinks for several years now. You're right, his stats are pretty awful, but how long does he have to drastically outpace his stats before you start to question the validity of the statistical model? I mean, he's going out there and not giving up runs, isn't he?

I looked at his GB/FB/LD rates, (52.4%/28.6%/18.9%), and they're exceptionally skewed towards ground balls. He's right around the 90th percentile in GB rate, which considering he has 3 guys behind him who were drafted as shortstops plus Zim, they're likely to turn most of those into outs.

Again, this isn't a flame or anything, but everyone's been hating on him for years, but except for last year when he got knocked around so much, you can't say he hasn't produced, and has done so for 4+ years now.

I have a modification to the theory to propose, and I'm too lazy to do the research, but if "effective pitching" means "missing bats" as you suggest, then you'll need lots of Ks to make that show up. However, if "missing the sweet spot on the bat" also makes for effectiveness, you'd expect a much-higher-than-average GB rate, (check) a lower-than-average LD rate (check) and a lower-than-average HR/FB rate (no, he's about average).

So maybe that's why he's effective? Instead of making guys miss the ball entirely, he makes them miss the ball by 1/2", which turns a single into a 6-3?

just a thought.

Steven said...

I'm not sure we're disagreeing really. I'm making the case that Lannan can continue to enjoy a level of success that has real value in MLB and is reasonably close to the outcomes he's getting so far this year. That's the opposite of saying he stinks.

You're on the right track focusing on the groundball rate. That's the #1 reason why he's successful. Groundballs are good because they can't be home runs. Actually, groundballs become hits more often than flyballs, so Lannan's BABIP in the 270s is not likely to continue, but the Nationals have a pretty good infield defense now, and the error bars are pretty big there. In any case, if you're getting a lot of groundballs, you're forcing the other team to string together a bunch of them to get runs, because they aren't hitting many balls in the air that could go out. That's the whole idea.

Walks are a problem because if you're giving extra base runners, the other team doesn't need to string together as many. The balls in play on the ground become RBI singles rather than stranded runners "scattered."

His HR/FB rate isn't something he can consistently suppress--that's way more a function of the power of the hitters he's facing and to a lesser degree park effects. The pitcher probably has some influence, but the pitcher's contribution is swamped by the hitter. Just think about it from a common sense point of view. Adam Dunn hits a flyball off Jason Bergmann, and Wilson Valdez hits a flyball off John Lannan. Which is more likely to become a homer? Switch up the pitchers and I bet your answer is still Dunn.

Kevin Rusch said...

"His HR/FB rate isn't something he can consistently suppress--that's way more a function of the power of the hitters he's facing and to a lesser degree park effects. "

I'm not convinced. Let's assume at least 50% of batters are trying to hit a line drive. If that's true (a big if, I'll admit) then the majority of the flyballs are "failed line drives".

From a different direction, if you mis-hit the ball by 1/2", some of those will be popups, some will be cans of corn, and most will be grounders. But outside of Barry Bonds, almost none of them will be homers.

To put it a bit differently, I'm going on the idea that a "well-struck ball" is either a line-drive or a homer. A small smattering of homers will be ordinary flyballs that carried or were hit somewhere tiny, but in general, a liner is what a batter is trying for, and a good percentage of them reach the seats.

The prevailing assumption seems to be that in general, most flyballs are a problem and the difference between a flyball and a homer is park effects and the batter's power. But it stands to reason that a large percentage of flyballs (most of which are outs anyway) are mis-hits as well.

So, for your example, Bergmann-Valdez v. Lannan-Dunn is a no brainer. But who do you think would do better against Dunn -- Lannan or Bergmann?

Steven said...

Bergmann, because he's a flyball pitcher.

You don't have to take my word for it. Go over to Fangraphs or Baseball Reference and look at pitchers' year over year HR/FB rates. You'll see that it's not a skill that pitchers can repeat year over year. Given enough repetitions, that rate will eventually land at around 11-12%. Major deviations from that are randomness, statistical noise.

Harper said...

you got me thinking so I checked to see if Lannan has seen a spike in K's recently. Just the opposite. His K-rate has gone down significantly since he's been pitching better (from roughly 5.40 to 4.00). That leads me to believe maybe K/BB is more telling (at least for a specific subset of pitchers) If we look at those pitchers that you named outright, lannan, gubiza and wang all have K/BB over 1.30 while the other three don't top 1.07.

(I'd do more checking myself but my play index account needs to be renewed...)

Steven said...

Gotta be careful about sample sizes as small as a few starts.