April 17, 2009

Fantasy Beat

Dreaming of Kevin Millwood

by Marc Normandin

Printer-friendly

This early in the season, everyone is always searching for that pitcher whose newfound success no one knew was coming. That's the kind of acquisition that can push your team over the top if your guess is correct. The last few days I've been asked about Kevin Millwood, and whether or not I think his performance from his first two starts means anything-is he that guy this year? I hadn't looked at it previously, but I thought that the question would serve as a decent thought experiment, where I can show you some of the ways that I look at pitchers while analyzing them. Let's find the answer together, shall we?

Thus far in 2009, Millwood has thrown 14 innings with a 0.64 ERA, 7.1 K/9, 0.64 BB/9, and is yet to give up a home run. Let's ignore for a second that this is just 14 innings, as the point of the exercise is trying to decide if anything has changed that should make us think differently about Millwood. Last year, he threw 168 2/3 innings, posted an ERA of 5.07, struck out 6.7 per nine, and walked 2.6 per nine, while allowing just under a home run per nine. You may be wondering why his ERA was so poor when his peripherals aren't nearly as unattractive-that would be thanks to his BABIP, which was .366, or roughly 70 points above the league average.

His FIP from 2008 was actually a respectable 4.02, and his QERA was 4.44-were he anywhere but Texas, that may have made him worth something heading into 2009. Between The Ballpark at Arlington (.328 BABIP last year at the park) and the Rangers' atrocious defense (.670 Defensive Efficiency, last in the majors) it's no surprise that Millwood underperformed ERA-wise. In fact, his defense was so poor last year that in games when he induced loads of grounders, his performance was worse. Splitting his starts from last year into games with 10 or more grounders and games with fewer than 10 grounders gives us some interesting numbers:

Ground Balls      IP    H/9  K/9   RA
More than 10 GB  88.0  13.0  5.7  5.93
Less than 10 GB  80.2   9.0  7.7  4.57

Basically, Millwood was Livan Hernandez when he induced grounders, despite the fact that ground balls should make a pitcher's job easier. He wasn't incredible when he induced more fly balls, but a 4.57 ERA is rather nifty for a pitcher from Texas over a sample of around 80 innings. One thing I'm wondering about is whether he was doing something differently in the games where he had more strikeouts. Millwood forced 166 grounders in his 88-innings sample, and 161 fly balls during that same stretch. During the 80 2/3-innings sample, the numbers were 86 and 174; there's a significant change in his ground-ball rate, but just a small bump in his number of fly balls, while his strikeouts end up being the recipient of the larger boost.

One thing that those numbers fail to tell us, which is actually a huge problem, is which balls in play are line drives, and which ones are fly balls. ESPN's G/F ratio doesn't differentiate, and Fangraphs isn't broken up into game-by-game samples like ESPN is, which is a shame given that they do differentiate between liners and fly balls. This means that Millwood could have given up around the same number of total fly balls, as documented by ESPN, but what he in fact did was give up fewer line drives in those games while forcing more hitters to pop up weakly due to having good stuff those days, a conclusion supported by his loftier strikeout rates. However, it's hard to tell one way or the other with the given data.

What we can see is that Millwood struggled in games where he gave up lots of grounders, and part of that, if not all, is due to Texas' terrible infield defense. John Dewan's Plus/Minus rates players on grounders and balls in the air separately, in addition to a total score, and shortstop Michael Young was -6 on ground balls last year. First basemen for the Rangers (Blalock, Broussard, Davis, Shelton) accounted for a combined -3 on grounders last year, with just Chris Shelton (+5) doing anything worthwhile among that group. Ian Kinsler was rated -13 on ground balls at second, and the third-base group (Blalock, Davis, Vazquez) accounted for a whopping -26. Given that the outfield group was all average or above (except for Josh Hamilton, who should be playing a corner, in center), it makes sense that fewer ground balls from Millwood would equal more success, whether he's causing guys to fly out or striking them out on his own.

This season, Millwood has a G/F of 1.6, with 46 percent of his batted balls being the ground-ball variety. He's been known to give up lots of line drives, and this season has been no exception in the early going; is line-drive rate is at 24 percent (last year was 25.3, and his average since 2002 is 22.7). Of course, this is in just 14 innings, so this data may look very different in just a few more weeks. The fact that it resembles his recent work makes me think that we won't see very much in the way of change here, but instead we'll see that his walk and homer rates begin to slide back towards expectations.

We can use QERA to see where his ERA should be now, and where we can expect it to go once those numbers jump back up. The definition for QERA is as follows:

QuikERA (QERA), which estimates what a pitcher's ERA should be based solely on his strikeout rate, walk rate, and GB/FB ratio. These three components-K rate, BB rate, GB/FB-stabilize very quickly, and they have the strongest predictive relationship with a pitcher's ERA going forward. What's more, they are not very dependent on park effects, allowing us to make reasonable comparisons of pitchers across different teams.

For Millwood, we can take into account the fact that his infield defense may still be troublesome (Young, Davis, and Kinsler are all still there, though losing Vazquez and Blalock while adding Elvis Andrus should help things). Let's also consider that he's in a hitter's park and is likely to give up a few home runs-he's allowed just under 1.0 per nine in each of his three seasons with the Rangers. His current QERA is 2.95, which makes sense given his ground-ball rate and more than adequate punchout figures. If we change his walk rate to reflect his three-year average with the Rangers, we get 6.9 percent, which would bring his QERA up to 3.64. The reason that number is nearly a run better than last year's QERA is due to the change in strikeout percentage (16.3 last year, 22.0 so far in '09). Let's look on the bright side of things though, and assume that he keeps those whiffs. We still have to account for homers, his home park, and the defense behind him, which should knock him well over the 4.00 QERA mark.

Remember, those same factors are the reason that he underperformed last year; while he may be an undervalued pitcher in real life, all we care about in fantasy is what he can do for our own team with real-world statistics. Were Millwood to be dealt to someone in a less hitter-friendly park with a solid team defense-say if Oakland suddenly needed a starter-then Millwood would be someone to jump all over, but for now, he's probably not "that guy" that we're all looking for to help us gain an edge.