Tuesday, August 19, 2008

What's Wrong: Justin Verlander

For most young baseball players, the expectation is that there will be improvement until the age of 27-29, then either a steady or sharp decline in their 30s. But a number of young players this season saw significant failure-- is this a harbinger of doom for these players, or just a fluke season? To answer these questions, we will have to rely on past down-years for similar players, as well as some peripheral statistics for 2008. We're going to start with a pitcher whose mediocre 2008 seasons affected me personally, because I spent a fantasy baseball draft pick on him in the ninth round.

Justin Verlander

2007: 201.7 IP, 3.66 ERA, 183 K, 67 BB, 181 H (ERA+ 125, 8.17 K/9, 8.19 H/9, 3.00 BB/9, 1.23 WHIP)
2008: 164.3 IP, 4.60 ERA, 129 K, 66 BB, 146 H (ERA+ 94, 7.07 K/9, 8.01 H/9, 3.61 BB/9, 1.29 WHIP)

UPDATE: A commenter pointed out that I made an error in Verlander's strikeout rate. In fact, he has reduced his strikeouts by more than one per nine innings, which is a significant difference. As noted in the comments, fewer strikeouts means more balls in play, but the BABIP (See below) suggests that those extra batted balls are not dropping for hits.

How did the Tigers' ace lose a full run/game on his ERA? While his strikeout and hit rates have largely stayed the same, his walk rate is elevated, but not some sharp increase that screams "suddenly bad pitcher!" (See: Willis, Dontrelle). This might suggest that Verlander has simply gotten "unlucky," which in baseball projection terms is talking about BABIP (Batting Average on Balls In Play). Research suggests that a hitter with a high BABIP in one individual season will see a correction in the following season; when the number of hits goes down, so does the ERA. Most interestingly is that pitchers, from season to season, do not seem to have an ability to control where a batted ball is hit (except home runs). So is Verlander allowing a bunch of bloop doubles this year?

2007: 42.0% GB, 30.7% FB, 18.1% LD, .279 BABIP
2008: 42.0% GB, 31.9% FB, 16.3% LD, .271 BABIP

Those are the percentage of all batted balls per trajectory. In addition to having a relatively stable BABIP, his line-drive percentage has actually dropped. Normally one could argue that a pitcher with a higher BABIP will not see a correction because they are allowing a lot of line drives (the trajectory with traditionally the highest associated slugging percentage). That's not true in Verlander's case. Using the new stat site www.statcorner.com, I looked at all of their breakdowns for Verlander. Only one difference between 2007 and 2008 really stands out: Verlander is getting fewer swinging strikes in 2008 (7.8%) than 2007 (8.8%). According to the site, the league average is also 7.8%, so Verlander has regressed in this area.

Swinging strikes immediately suggests that there's something wrong with either Verlander's fastball or curveball. So let's take a look at the pitch f/x data, using Josh Kalk's fascinating database. There's 2007 and 2008. Some things I notice, comparing the two player cards, in no particular order:

1) Release point. In 2007, the variation in Verlander's release point was absurdly tiny; almost exclusively between 5.8' and 7' from the ground, and between 2.25' and 3.5' horizontally. Compare that to the pitches recorded from 2008, which have a similar variance vertically, but are spread from 1.8 to 3.8' horizontally. Could it be that a wider variety of release points explains his higher ERA?

2) Pitch Movement. In 2007, pitch f/x calls three different pitches in three distinct clusters- fastball, moving between 4 and 15 inches vertically; changeup, moving between 1" and 12" vertically; and curveball, moving between 0" and -9" vertically (this means the pitch is falling less than a pitch without spin). In 2008, the system adds the recognition of the cutter, but the fastballs seem to be in the same area as in 2007. The changeup and curve clusters, however, have migrated towards each other and overlap slightly. Using the measures I invented in the A-Rod posts, it appears that curveballs with vertical break greater than 0 cause more missed swings (28.3%) than do curveballs with break below zero (20.2%), like all of the curves in 2007. But, using TB/Swing, the positive curves generate bigger hits (.191 TB/Swing) than negative ones (.151).

3) Fastball velocity: In 2007, his average fastball was recorded at 95.17 mph. This year, it's actually up to 96.07, while the cutter sits at 94.41. So it may be that pitches that last year counted as cutters were lumped in to the 2007 total, but Verlander's velocity has not changed at all.

So, if his peripherals are staying the same, he's not overly lucky this year, and his pitch f/x data is essentially the same, what's going on? The Detroit blog Triple Deke thinks that part of the issue is performance against left-handed batters. But lefties are hitting nearly the same this year (.694 OPS against) than last year (.689), with fewer home runs per plate appearance. Triple Deke also points to Verlander's performance with runners in scoring position, and it's true that he's allowed a .726 OPS against in those situations, compared to .651 last year. But it's unclear whether performance with RISP is really a repeatable ability, or perhaps we looking at one of those datapoints that suggests Verlander is simply unlucky.

Regarding a projection for the future, Verlander doesn't have much of a major league resume to go on just yet. Baseball-Reference listed Clark Griffith the most similar pitcher at age 23, and Elden Auker at age 24, which doesn't help much. Many of the more recent pitchers on the list did not experience a drop-off until later in their career (Mussina, Pettitte, Hudson) or can't stay off the DL (Wood, Harden). Eight of the ten pitchers listed improved their career ERAs in their year-25 season; Verlander dropped his from 121 to 112 this season.

For what it's worth, the Marcel Projection system forecast a step back for Verlander this season: it thought he would go just 179 innings with a 3.85 ERA, a 7.44 K/9, and a 3.02 BB/9. Marcel is the simplest projection system possible, and is based off of the last three years. Since this is Verlander's third full season, the projections for 2009 are not likely to be positive. But for a pitcher who has kept almost everything identical from a good season (2007) to a mediocre season (2008), who can say for sure what happened, much less what's going to happen. Meanwhile, I'm still trying not to giggle from writing (Wood, Harden) last paragraph...


World of Isaac said...

a little geeky but a darn good read man

Anonymous said...

I'm only posting because Verlander is trying to kill my fantasy season. Great summation of everything. However, his strikeout rate has actually dropped drastically this year. The 164.3 IP and 129K's are correct, but this comes to 7.07 k/9, not over 8 as you indicate. I agree with the rest of the analysis, just wanted to point out that the reduced k's means more baserunners allowed in general.

Matt said...

Wow, that was a pretty big error. I was reading the hits column instead of the strikeouts. I will adjust the numbers now.

Thanks for pointing that out.