Phil Hughes’ 2010 Curveball through the lens of PitchFX

This post is the second in a four-part series taking a deep, pitchFX-fueled look at each of the five pitches threw during the 2010 season. Please be sure to reference the previous entry as well before, during or after you read this latest post:

Phil Hughes’ 2010 Four-Seam Fastball through the lens of PitchFX [March 2, 2011]

Everyone knows Hughes was primarily a three-pitch pitcher in 2010 — 57% four-seamers, 18% cutters and 17% curveballs. As Hughes threw each of these pitches in all 32 starts (including three postseason starts) he made in 2010, these three pitches will each merit their own post. The fourth and final post of this series will look at Hughes’ two-seamer and elusive changeup (each thrown roughly 3% of the time).

Today we’ll review Hughes’ tertiary weapon, the curveball.  Earlier in his career it looked like the curve was going to be the pitch that catapulted Hughes into #1 starter territory; however, it’s only gotten less effective with each passing season, peaking at 2.2 runs above average in 2007, and declining in each successive season before bottoming out in 2010 at -5.9 runs above average.  If Hughes can fix his curve so that it becomes a plus pitch again, he may not even have to worry about the long-awaited development of that changeup, although four quality pitches are of course always better than three.

The below chart maps the horizontal and vertical breaks of Hughes’ curveball by start throughout the season, and compares them to both Hughes’ seasonal average break for the pitch as well as the MLB seasonal average for the pitch.

All of the data I collected, including a comprehensive game-by-game breakdown of the results of the average breaks for each of the pitches in all of Hughes’ starts, is available for download .

While you didn’t need a chart for me to tell you this, Hughes’ curveball was all over the place in 2010, at least horizontally. He appears to have a tad more consistency with its vertical break. Hughes averaged just under six inches of H-break compared to the MLB average of 5.3, so he wasn’t all that far off the average MLB curveball horizontally; however the real difference in Hughes’ hook is in its vertical break — -8.7 inches compared to the MLB average of -5.5. Whether falling more than three inches compared to the average MLB curveball is a good thing or not remains to be seen.

Hughes’ curve, true to 2010 form, was also all over the place in the playoffs. In his excellent start against Minnesota, he threw it 18 times and averaged 4.79 inches of H-break (a full inch less than his regular season average) and -10.6 inches of V-break (two inches further down). In ALCS Game 2 in Texas he doubled his H-break to 8.33 and had three fewer inches of drop (-7.83). The H- and V-breaks were essentially the same for his Game 6 start. Without linear weights (f0r some reason Brooks’ database isn’t pulling playoff games) I can’t say for sure how the curveball fared in Texas, but given that it was nearly three inches further away from the plate than it was during the season, along with the fact that Hughes didn’t pitch all that well in either game, I’m going to hazard a guess that these were not Hughes’ most successful curveballs.

For this following set of data, I binned each of Hughes’ 29 regular season starts into “good” and “bad” pools, based on the linear weight results he recorded on the pitch for that game. The “good” pool consists of all of the games in which Phil recorded negative linear weights on his curveball (per Brooks Baseball), which occurred 13 times (compared to 20 for his four-seamer). In case you were wondering, the four-seamer and curve both registered in the negatives on April 21 at Oakland, April 27 at Baltimore, May 2 vs. Chicago, May 17 vs. Boston, May 28 vs. Cleveland, June 13 vs. Houston, July 9 at Seattle, July 30 at Tampa Bay, September 5 vs. Toronto, September 15 at Tampa Bay and September 26 vs. Boston, or 11 times. So chances are when his curve was going good he also had his good fastball, which makes sense. The only thing that doesn’t make sense is that most of those 11 outings were strong, except for the September 5 start against the Jays. With great results on both the four-seamer (-1.1306) and curve (-0.7427) that game, his cutter must’ve gotten shellacked that outing, and indeed it did — at 2.3525 linear weights, it was his cutter’s third-worst performance of the season.

The “bad” pool consists of all of the games in which Phil recorded positive linear weights, which happened 16 times. For whatever reason, Brooks does not have PitchFX data for Hughes’ start May 28 start against Cleveland, and so I can’t say with 100% certainty what the linear weights were on any of Hughes’ pitches for that game, but given that it was a gem (7IP, 5H, 2ER, 8K, 1BB, 109P) I feel comfortable including it in the “good” pile.

“Best” is of course the game in which he recorded the best linear weights on the curveball (-0.8177), which was on August 9 at home against Boston (6IP, 6H, 2ER, 3K, 1BB, 114P). His “worst” linear weights for the curve (1.7792) came against Detroit on August 19, which actually wasn’t a terrible outing for Phil (6IP, 4H, 2ER, 6K, 0BB, 84P), primarily due to the fact that he had his second-best four-seamer of the year (-3.3941). Interestingly, two of his top three four-seamer performances came against the Tigers last season. Also interesting was that his “best” and “worst” curveball performances came within 10 days of each other, which goes to show you just how erratic this pitch has been for ol’ Phil.

Hughes’ “good” curveball averages about 0.60 more horizontal movement than usual (average), and half an inch more drop. In his “bad” starts, Phil typically threw his curve about half an inch closer to the plate, while losing about 0.40 inches of vertical drop. These are really small changes, and unfortunately don’t really tell us all that much, especially since in his “best” start he threw his curve about 1.5 inches closer to the plate than average and had nearly an inch less vertical drop, while in his “worst” start he had nearly identical H-break as his “best” curve, though his V-break was essentially the same as his season average, so again, it’s best to take this information with a grain of salt.

As I mentioned in the four-seamer post, we must look at Phil’s average breaks in tandem. Similar to his four-seamer, based on this table it looks like somewhere in the vicinity of 6 inches of H-break is where we want his curve to be; while on the V-break end of things, ideally we’d want the curve to drop closer to nine inches. Interestingly, the 2008 version of , who had the 4th-best curveball in baseball, averaged 6 inches of H-break on the curve (so Hughes at least has that covered), and about -6.75 inches of V-break. Not that Hughes needs to try to mimic what did in order to have a successful Major League curveball, but knowing this makes me wonder if Hughes, with two inches more V-break than Burnett’s excellent curve in 2008, might be better served if he can throw the pitch with less vertical break.

Hughes clearly still has a lot of work to do on his curve if it’s ever going to return to being a plus pitch for him, but it’s not as if all hope is lost — he did have 13 (out of 29) outings where the curve ranked as an asset, though he’ll need to improve that ratio to better than 50% going forward. If he can do that — and/or add an effective changeup — then watch out, American League.