Neftali Feliz: Messing with a Good Thing

In the 2012 season, few players have disappointed on the Texas Rangers roster: Yu Darvish is a stud (2.98 xFIP in his last six starts), Adrian Beltre apparently lied about his age (he’s 10 years younger than previously believed), and Josh Hamilton is using a Game Genie and is 7 homers away — so, like a week away — from matching his 2011 home run total.

In the Rangers Machine, the only cog slipping from the threads (if we discount Matt Harrison’s BABIP’d ERA) is the converted starter Neftali Feliz. He may have a solid ERA (3.16), but the 24-year-old has danced in and out of trouble all season, striking out a career low 21.1% of batters while walking a career high 13.1%, and on top of it all, he’s headed to the DL with a right elbow strain.

Could the injury have caused his decreasing effectiveness? Possibly, but upon closer inspection, it becomes apparent that Feliz has altered the approach that made him a successful reliever and the change has only hurt him.

This is Feliz’s repertoire in 2011:

2011 Feliz PITCHf/x

And here he is so far in 2012:

2012 Feliz PITCHf/x

If open both images in two tabs and flip back and forth real fast-like, you can make some of these observations:

1) He has lost a little fastball velocity. This is what always happens in a move from bullpen to rotation, so this is certainly no cause for concern. The velocity has gone down, but the effectiveness (1.66 wFB/C) is pretty much at his career levels.

2) His slider and cutter appear to have merged into one massive and wildly ineffective pitch.

3) He has added a two-seam fastball and is using his changeup with much greater frequency.

Point No. 2 is the main interest here. In 2011, Feliz had a four-seam/cutter duo that did well to support a less-than-effective slider. In 2010 — by far his best full season in relief — Feliz threw primarily his four-seamer, and then befuddled hitters with an occasional slider or curve.

His 2010 approach probably could never work in the starting rotation. Much of that success was predicated on his reliance on a near-100 mph fastball which he threw 80% of the time. Since 2007, the only starter to throw his four-seamer more than 72% of the time was Orioles pitcher Daniel Cabrera. If you want to model your approach after a 110 FIP-minus pitcher, by all means throw 80% fastballs.

So Feliz — in 2011 — began to change his repertoire, maybe in anticipation of entering the rotation in 2012, maybe in just a continued effort to improve as a pitcher. Nonetheless, he almost entirely dropped the curve (puns not intended) and added the cutter.

At an average of 87.4 mph and with about half as much movement as the slider, Feliz’s cutter quickly became his most effective pitch with 2.98 wFC/C, but his overall productivity dropped (2.77 SIERA to 3.91 SIERA).

Now, in 2012 as a starter, Feliz has seemingly lost the ability to miss bats. Depending on whether we look at his BIS numbers or PITCHf/x numbers, we might think Feliz is either struggling with location or hitting his locations well. BIS says his Zone% is near a career high; PITCHf/x says his Zone% is at a career low. Either way, it is not lower than 78%, which is still within the range of effectiveness for most starters. And since BB% has a low correlation with Zone% (and R-squared of .16 or .20 depending on the data source, BIS or PITCHf/x) and almost exactly zero correlation with K%, we must then look at his contact and swinging strike rates.

And those a rates that are most definitely trending in a bad direction. His Contact% rates are considerably increased from his 2011 and 2010 contact numbers. BIS says they are a career high 80.3%; PITCHf/x says 83.8%. And BIS says the swinging strike rate has taken a pretty sharp turn:

I contend that Feliz he dropped his slider altogether. The cutter, which had been so effective in a backup role (6.7%) to the slider (12.1%) has essentially taken the role of his slider in 2012. PITCHf/x says he is throwing the slider 20.1% and the cutter 0%, but the PITCHf/x charts show the slider velocity has increased while the movement has decreased. To me, the “slider” seems to be just a slower version of his cutter, suggesting the true slider has disappeared entirely.

His fastball lost 1.7 mph of speed, his changed slowed by 0.9 mph, but his “slider” increased by 3.2 mph.

For many years, Tim Wakefield’s best pitch was his fastball. Rather, his most effective pitch was his fastball. Hitters were so focused on his twisting knuckler that they found themselves swinging out of their shoes at his 72 mph “heat.” I would not be surprised if Feliz’s cutter was receiving that same Sucky Pitch bonus.

Would adding back the slider cause Feliz’s cutter to regain a degree of effectiveness? Possibly. If I am Feliz, I would at least give it a try. There is, of course, a chance that he is indeed throwing his slider, but the two pitches are not separate enough from his cutter to create two PITCHf/x classifications — which is as good, in the hitter’s minds, as being one pitch anyway. If that is the case, then the problem may be more severe.

But I would put a wager on the possibility that Feliz scrapped his slider — which, in 2011, was getting 40% whiffs on swings from righties — and it is time to try it, or maybe even dust off the old curveball, once again. Assuming Roy Oswalt hasn’t pushed him back into relief by the time he returns.