Building (or Finding) the Ideal Pitcher

The PITCHf/x ERA is approaching ten years old, but the research spawned by the free public access to the data is impressive. We’re now seeing teams start to act on those findings as they try to use the data to inform best practices. Could we take the research as far it might stretch? Can we build the perfect pitcher?

We aren’t talking rates here, at least not in the ‘the perfect pitcher would strike out 100% of the batters he faced and walk zero,’ sort of nonsensical way. We’re talking about research into injury likelihood, and pitch effectiveness, and platoon splits. Let’s list some of those findings because bullet points are easy.

• Heavy breaking ball use may lead to injuries.
• Pitchers with great control stay healthier, on average.
• Pitchers that throw the changeup heavily have lower attrition rates.
• Changeups have one of the lower batting averages on balls in play.
• Throwing sidearm puts more stress on your elbow.
• Changeups generally have a reverse platoon split.
• A ten-mph-plus difference in velocity is indeed important if you’re throwing your changeup for whiffs.
• The rising fastball has the smallest platoon split among fastballs.
• First-pitch strikes are the best peripheral associated with walk rate.
• A pitcher’s velocity, on average, drops every season.
• More velocity means fewer runs allowed.
• Beyond 96 mph, there’s a jump in swinging strikes on the fastball.
• Vertical movement on the fastball means more grounders.
• Pitchers with long arms and long strides release the ball closer to home plate and make their velocity play up.

Just teasing this out in one sentence, it looks like we want a pitcher with a big rising fastball (96+), a sinker with serious drop, a changeup that’s more than ten mph slower than his fastball, just enough breaking stuff to keep same-handed hitters honest, and great control. Bonus points if he’s tall and has a long stride. Considering that the Pirates focus on teaching all of their pitchers command of two fastballs, and the Athletics’ and Rays’ effort to make all of their pitchers learn the changeup, some of this stuff is already being used in the development process. But here’s your mythical pitcher:

Pitcher McPitcherson (6′ 6″, 220, R or L?)
40% 96 mph rising fastball (-4 PFx_x, 10 PFx_z, 9+% swSTR, 40+% GBs)
20% 95 mph sinker (-9 PFx_x, 1.5 PFx_z, 6+% swSTR, 60+% GBs)
30% 86 mph changeup (-5 PFx_x, 7 PFx_z, 20+% swSTR, 50+% GBs)
10% 88 mph tight curve / slider (3 PFx_x, -4 PFx_z, 12+% swSTR, 50+% GBs)

As you can see, this pitcher has stuff that breaks in all directions. He’s got Chin Music, Bat Breakers and Knee Wobblers. He’s got something to break in on lefties, something to break in on righties, something to get whiffs with, something to get grounders with. He’d have a 12% swinging strike rate (five qualified pitchers did last year), and a 48% ground ball rate (only one pitcher did both last year).

Wait. Only one pitcher did both last year?

Is Matt Harvey the perfect modern pitcher? He satisfies the results end of the spectrum, since he was the only starter to combine 12% whiffs with 48% ground balls. He’s got the velocity (96 on the four-seamer, 94 on the two-seamer), and even if he doesn’t use the sinker as much, his secondary pitches are a changeup, slider, and curve. He’s really not far off, according to BrooksBaseball:

Matt Harvey (6′ 4″, 225, RHP)
57% 96 mph four-seam fastball (-6 PFx_x, 9 PFx_z, 12% swSTR, 38% GBs)
1% 94 mph sinker (-9 PFx_x, 7 PFx_z, 7% swSTR, 75% GBs)
12% 87 mph changeup (-9 PFx_x, 5 PFx_z, 20% swSTR, 60% GBs)
12% 83 mph tight curve (1 PFx_x, -3 PFx_z, 13% swSTR, 58% GBs)
17% 90 mph slider (1 PFx_x, 4 PFx_z, 17% swSTR, 52% GBs)

That’s almost uncanny. he’s got stuff breaking in all directions, hits all the benchmarks save the two-seamer, and gets whiffs and grounders. He’s six-foot-four (and dreamy), too.

And he’s hurt. Of course, injury was the reason we wanted a changeup-first guy, and he doesn’t *quite* satisfy the command component. He had good walk totals last year, but not in the minors, and not in his first go-round. Is there anyone else?

Cole Hamels is close. In 2011, he was even right there with whiffs (11.3%) and ground balls (52%). Here’s his pitching mix from that year:

Cole Hamels (6′ 3″, 195, LHP)
45% 92 mph four-seam fastball (4 PFx_x, 12 PFx_z, 5% swSTR, 41% GBs)
25% 83 mph changeup (-9 PFx_x, 6 PFx_z, 29% swSTR, 67% GBs)
12% 76 mph tight curve (-1 PFx_x, -3 PFx_z, 12% swSTR, 53% GBs)
20% 89 mph cutter (1 PFx_x, 7 PFx_z, 9% swSTR, 60% GBs)

It’s close. He’s changeup-first, and even if he doesn’t have the big velocity, he’s got the command we’re looking for. He’s been healthier, too. And he’s got the thing with the pitches breaking in the different directions and all that. His bonus comes from using his left hand more than his dreaminess.

Clayton Kershaw seems perfect now, but he’s still searching for the changeup. James Shields has the mix, but not the results. Anibal Sanchez is close. Really close. Tim Lincecum never had the control. Young man Michael Wacha really has a chance — a sinker might be all he needs. If Gerrit Cole was changeup-first, he’d be it. Maybe David Price and Max Scherzer deserve more attention. Down a little further on the changeup leaderboard, though, you might find the best candidate:

Felix Hernandez (6′ 3″, 230, RHP)
19% 92 mph rising fastball (-2 PFx_x, 7 PFx_z, 9% swSTR, 32% GBs)
36% 92 mph sinker (-8 PFx_x, 5 PFx_z, 5% swSTR, 54% GBs)
22% 89 mph changeup (-6 PFx_x, 0 PFx_z, 25% swSTR, 67% GBs)
11% 84 mph slider (2 PFx_x, -3 PFx_z, 12% swSTR, 58% GBs)
13% 80 mph roundhouse curve (6 PFx_x, -9 PFx_z, 13% swSTR, 60% GBs)

So maybe the younger Felix Hernandez, the one with more gas, maybe he was the perfect pitcher. Except back then he didn’t throw the changeup as much. And he’s never hit the threshhold for whiffs, with four double-digit swinging strike rate seasons out of eight. And there’s some other benchmarks he didn’t quite hit. Even if he does a look a lot like McPitcherson.

Of course there’s no perfect pitcher, in the end. It was a tight definition. And maybe that’s the point of the exercise. After all, we can define our perfect pitcher, we can ask our pitchers to strive for that ideal, and we can mold our practice sessions and coaching strategies to aid players in that process… but we can never achieve it. And so, if you get a 5-11 Dominican guy and you aren’t sure about his mechanics, or even a 6-10 dude that just throws way too many sliders to make you comfortable with his long-term health… sometimes it just works. After all, Pedro Martinez and Randy Johnson were okay pitchers in the end.