James Shields & The Taboo Topic of Hittability

There have been some excellent pieces on James Shields in the Rays blogosphere this season and this post by ghost of draysbay past Tommy Rancel inspired me to tinker with baseball-reference.com's season finder tool. If you read this blog, you already are well aware James Shields is posting an excellent strikeout-to-walk ratio while allowing an insane amount of home runs this season. Perhaps even stranger is the surprising amount of balls put into the playing field are resulting in hits (Batting Average on Balls in Play). Here are some of Shields' numbers:

The pro-Shields crowd will point to Shields' xFIP of 3.70 and declare that HR/FB rates should normalize beyond single seasons. The anti-Shields mantra revolves around our old friend "hittability."  While BABIP rates also typically normalize, the distance from a typical norm of .290-.300 is pretty loud. This is particularly true when you factor the Rays well-above average team defense which is a variable that has helped produce the following for staff BABIP rates:

The home run rate or BABIP alone would be easy to dismiss as random variation, but it does raise some questions when produced in stereo given the lack of overlap.   tERA which removes batted-ball location luck while still accounting for batted-ball types has Shields at 4.47 for the season.

Going back to Tommy's piece, I took a look at all pitchers in the American League going back to 1920 who produced a season of 175 innings pitched, with a K/9 north of 8 and a BB/9 south of 3. This produced a sample of 58 seasons by 30 different pitchers. The list of pitchers who have accomplished this in multiple seasons is: Curt Schilling, Felix Hernandez, Frank Tanana, Javier Vasquez, Johan Santana, Josh Beckett, Justin Verlander, Luis Tiant, Mickey Lolich, MIke Mussina, Pedro Martinez, Randy Johnson, Roger Clemens, and Zack Greinke. On its own, its a pretty impressive accomplishment, but what if we look at BABIP and HR/9 (hittability) for these 58 seasons?

First Shields versus the average of the subset:

OPS+ is how far high or below league league average a pitcher's OPS against, park-adjusted for their stadium. Shields OPS+ is at 116. How big an outlier is this? No other season out of these 58 had an OPS+ above average. Again when considering that BABIP and HR/9 have no overlap, it is hard to deny a real concern of hittability that separates Shields from this peer group.

Finally let's take a look at the worst five performers in this class in both HR/9 and BABIP to see how much cross-over there may be.

These are your worst season in terms of HR/9: three 28 year-olds and old men Schilling and Johnson. Only Schilling comes close to matching Shields' BABIP. This was Schilling's second-to-last season and the home run rate followed him the following season. Unlike Shields who plays behind a top-notch defense, the 2006 Red Sox defense was third from the bottom according to UZR. All things being equal, it is less surprising for Schilling to post the high BABIP due to the quality of the defense behind him. Now for the BABIP underperformers:

Again, we have the fading Schilling, this time accompanied by three pitchers with below-average home run rates. In summary, xFIP is probably not the best indicator to use in evaluating James Shields given the hittability issues he has displayed relative to his peer group. Instead we should continue to ask the questions regarding his process. There is no denying Shields has a plus changeup which he uses to put away some of the best hitters in the game, and the ability to throw strikes to keep his walk rate down. The process of how to best get James Shields in spots to put away hitters with the change while minimizing the amount of hittable pitches remains the difference between Shields and the true talent level of the the majority of this peer group.  

Final note: In 2000 Pedro Martinez posted an OPS+ of 18. This is the result of a slash line against of .167/.213/.259 with a K/9 of 11.8 and a BB/9 of 1.3. Wow.