The Amazing Phil Hughes

wFB/C wCT/C wCB/C wCH/C
Starter -0.86 -0.17 -1.31 -1.92
Reliever 2.66 5.45 0.78
FB% FBv CT% CTv CB% CBv CH% CHv
Starter 57% 92.3 20.3% 87.7 21.4% 76.5 1.3% 83.5
Reliever 67% 94.7 13.4% 89.3 19.6% 77.5

wFB/C shows how many runs above average Hughes creates per 100 fastballs thrown. wCT/C is for the cutter, CB is the curveball and CH is the Changeup. The numbers there speak for themselves. On the next chart, the FB% is what percent fastballs he’s throwing, and FBv is the average velocity on fastballs. As you can see, the speed on his pitches has ticked up across the board, and he’s throwing more fastballs (which makes sense given he never has to get through the lineup twice).

Let’s take a look at the differences between his fastball as a starter and a reliever. While I’m currently putting in significant work on creating a more complete pitchFX database, with which I’ll be able to do much more complete analysis of this sort, at this moment I’ll use data from Hughes’ September 22 relief appearance, and his start against Texas on May 25th.

Pitch Type Avg Speed Max Speed Avg H-Break Avg V-Break Count
Starter FF (FourSeam Fastball) 90.88 92.8 -5.75 9.69 52
Reliever FF (FourSeam Fastball) 95.42 97 -3.43 11.86 26
Delta 5% 5% -40% 22%
Starter FC (Cutter) 88.21 92.8 -0.33 7.14 14
Reliever FC (Cutter) 89.56 89.8 1.23 6.48 5
Delta 2% -3% -473% -9%

As expected, we can see the significant uptick in speed, but the real gem from this table is the difference in break. Though the delta% is very high, let’s look at the difference in nominal terms–the average horizontal break on the fastball has shifted 2.32 inches back towards the center of the plate, and the cutter now goes from slightly left of center (from the catcher’s perspective) to significantly right of center. Also interestingly enough, the vertical break on the two pitches is opposite in direction–the fastball now finishes more than 2 inches above where it did when Hughes was a reliever (which is measured against where it would have ended up if it followed the expected path given the speed and gravity–in other words he’s got more backspin on the pitch now than he used to which causes the ball to fall less ‘rise’). Conversely, the cutter is actually falling a tick more than it used to.

Listening to announcers and reporters alike, you’re much more likely to hear someone attribute Hughes’ improved performance to the uptick in velocity than to hear someone discuss the difference in movement (other than to simply suggest it’s increased). The thought here, though, is that it’s more due to the divergence of the two fastballs, than it is to the magnitude of the change in movement or the speed. A good example of this is Roy Halladay, whose dominance stems from how good his sinker slider combination is–they look similar out of the hand, while one breaks to the left and one breaks to the right. That leads to a lot of swinging strikes and soft contact. Looks like this is borne out by Hughes’ numbers, as well.

O-Swing% Z-SWing% Swing% O-Contact% Z-Contact% Contact% Zone% F-Strike%
Reliever 24.4% 71.2% 49.9% 59.5% 77.7% 73.7% 54.7% 62.2%
Starter 26.7% 66.7% 45.8% 67.1% 89.8% 82.9% 48.0% 61.4%
Difference 2.3% -4.5% -4.1% 7.6% 12.1% 9.2% -6.7% -0.8%

Little bit of explanation: O-Swing is how often opponents swing at pitches outside of the zone, Z-Swing is how often they swing at pitches inside. Swing% is an amalgamation of the two. O-Contact and Z-Contact are similarly how often batters make contact against pitches inside and outside of the zone (that they offer at). Zone% is how often Hughes’ pitches end up in the strike zone, and F-Strike shows how often he throws a first pitch strike.

So as a reliever, batters are making significantly less contact against Hughes, both outside and inside the zone. Coming in from the pen, Hughes’ has the 8th best Z-Contact% in the league (interestingly, amongst those ahead of him is David Robertson). As a starter, he wasn’t even on the first three pages of names, when sorting by that statistic. And there’s just about nothing better for a pitcher than swinging strikes on pitches inside the zone–a strike whether or not they make contact (assuming the umpire will play along). Notably, his zone% as a starter was below 50% (never a good harbinger of success), which he has improved from the bullpen.

It’s not all good news, though. Hughes the reliever has had a significant chunk of BABIP driven luck. His relieving .250 BABIP is far lower than his 22.9% LD% would suggest. Conversely, as a starter his .317 BABIP is a tick high compared to his 18.7% LD%. Both of these are on small sample sizes (as noted in the first paragraph of this article, so’s the rest of it too!) but at some point Hughes’ luck on line drives will run out.

The question that this brings up is one of weightings. His FIP is astoundingly low, and FIP isn’t affected by hits (unless they travel over the fence). So, while his ERA will almost certainly rise as his line drives start falling in, his FIP shouldn’t change. That said, I have trouble believing that an expected 100 bp increase in his BABIP (and the expected accompanying drop in strand rate) would only lead to an increase of ERA from 1.27 to 1.91.

What do you think? Feel free to talk it out in the comments.