Hamels’ Choiceby Harry Pavlidis
July 05, 2010
Join me in the land of make-believe. What happens when a three-pitch pitcher has success? He adds a sinker, of course. What if that doesn't work out? Try a cutter. Or not.
Meanwhile, be sure to lose confidence in two of your original three pitches. It will make this much simpler in the end. You know, when you're down to one pitch.
Embellished above, the real world case of Cole Hamels. What you're about to see (click to enlarge) is a game-by-game chart of Hamels' pitch selections. We can quibble on some two-seam fastballs (labeled as "sinkers"), I will grant that. Please note the 2007 data does not cover all games.
The sinker arrived late in 2008, the cutter opened with 2010. Peruse each pitch, by season, to shake out some of the variance you took in above. Or just enjoy some pictures before the words and numbers begin.
Old reliable, Mr. Four-Seamer
Not like it used to be
Also in steady decline
A flash in the pan
Merely a stub
What we have here, is a failure to diversify
There sure has been an ongoing effort to do so—until recently. Why? Probably conventional wisdom's emphasis on having at least four pitches to work with as a starter. Performance-wise, I can only see a reason to have dropped the sinker.
This is a big chart—click to enlarge. It shows Hamels' runs per nine innings as a four-game moving average. The red line shows an "expected" ERA that is derived from batted-ball types while the black line is "actual" (but not) ERA based on the types of hits Hamels allowed. Both are adjusted for the count, and are based on pitch-by-pitch outcomes and linear weights.
The chart is divided into pre-sinker, sinker's peak, fewer sinkers, cutters and, with a white divider, the low-cutter games seen at the end of the first chart above.
When throwing the sinker frequently, Hamels seemed to be getting as good, or better, a batted-ball distribution as he did before, as indicated by the red line (rvERAe). His actual outcomes were worse, though. Either he was getting poor defense, bad luck or, despite the trajectory (liner, fly, grounder) he was getting hit harder.
Things really didn't improve for Hamels, beyond a late-season stretch in 2009, after backing off on throwing the two-seamer. Hamels added the cutter this year, and, interestingly, he shows the same "expected is better than actual" results.
Let's put some of these trends into starker contrast. Since 2007 does not have complete PITCHf/x data, I'm going to focus on the last three seasons, including 2010. Remember, Hamels started working the sinker hard at the end of 2008, moved away from it around the second half of 2009 and added the cutter this year. Even though he has backed off the cutter recently, we'll just stick 2010 in one bucket.
Pre-Sinker: 4/2/2008 - 9/7/2008 (3017 pitches)
Sinker's Peak: 9/13/2008 - 8/15/2009 (2511)
Fewer Sinkers: 8/21/2009 - 10/31/2009 (899)
With Cutters: 2010 (1486)
Hamels' rvERAe has been fairly stable over these periods:
Sinker's Peak: 3.34
Fewer Sinkers: 3.18
With Cutters: 3.66
Meanwhile, his rvERAa has been all over the place:
Sinker's Peak: 4.43
Fewer Sinkers: 2.70
With Cutters: 4.61
Is it just me, or does that look like a pitcher who is only comfortable with his original three pitches?
Let's look at some other numbers, like throwing strikes. We'll go with two numbers, the rate pitches are thrown in a wide strike zone (24 inches wide) between the knees and the letters and the ratio of balls-to-called strikes.
Pre-Sinker: .545 and 1.9
Sinker's Peak: .545 and 2.0
Fewer Sinkers: .533 and 1.7
With Cutters: .539 and 1.8
No problems there. Still throws strikes, umpires still think so, too.
How about missing bats? Here are Hamels' whiff rates (misses per swings)
Sinker's Peak: .235
Fewer Sinkers: .309
With Cutters: .261
Fall of 2009 was some quality stuff for Hamels. It's also the shortest run of the four we're exploring. In either case, since cutting back on sinkers, Hamels is missing more bats. Makes sense.
Often whiffs are due to chasing out of the zone. These are swings rates against Hamels, the first rate is for all pitches and the second just for those out of the wide zone ("Chase" or "OOZ")
Pre-Sinker: .499 and .333
Sinker's Peak: .483 and .297
Fewer Sinkers: .501 and .340
With Cutters: .452 and .276
Again, the "Fewer Sinkers" run in late 2009 was impressive. It starkly contrasts low swing and chase rates of 2010 with the cutter in the quiver.
Last set of numbers: home runs per fly balls+line drives
Sinker's Peak: .106
Fewer Sinkers: .038
With Cutters: .140
This year has seen an unusually high rate of home runs for Hamels. League average is about .075 last I checked, so he's really gone from one extreme to the other.
Back to the old
Whether it is luck or command—or places in between—Hamels has performed his best when using just three pitches or three-plus-one or two others sparingly. Along the way, he's relied less and less on his change-up and curveball—both his original secondary pitches—along with a very recent spike in fastball use. If there's something different about 2010, it's hitters laying off pitches out of the zone and fly balls turning into home runs. That and the new pitch.
It is very tempting to conclude the problems Hamels has had when using a fourth (or fifth) pitch frequently are a result of that very change. Either the pitch isn't very good, or throwing it has somehow shifted his focus and negatively impacted his use of his original three. More so, struggles with new pitches and his original secondary pitches may be pushing him to rely more on the fastball. Yes, the temptations of the "new pitches ruined him and now he's only throwing fastballs" theory are palpable. And testable.
Further research on Hamels is begged by these questions, for both the sinker and the cutter (which is an open book at the moment):
- Was the new pitch effective?
- Were the original three less effective with or without it?
- Did he stop or reduce use of an original three pitch in certain situations (counts)?
- Of the original three, has the fastball been impacted less than the curveball and change-up by the absence or presence of a new pitch?
References and Resources
PITCHf/x data from MLBAM and Sportvision. Batted ball data from MLBAM. Pitch classifications by the author.
Harry Pavlidis admits he has a baseball problem. He is the founder of Pitch Info LLC, His pitch classifications power the player cards at Brooksbaseball.net. Feedback, questions and comments are appreciated - Email firstname.lastname@example.org and Twitter @harrypav