David Price has a great fastball. He can easily pump it up over 95 mph; the movement is fantastic and varied between the two and four seem varieties, and he's learned to locate it with precision. His cutter is not as impressive. It sits between 90 mph and 92 mph, and is fairly straight (the perceived cut is in relation to a normal "straight" fastball, which actually moves quite a bit). Moreover, it's a new toy—he only started throwing the pitch this year.
Lately, it seems like Price has gone away from his fastball in key situations. His changeup has been a very good pitch, but when he's tried the newly developed cutter, he's given up runs. And people have noticed.
Fox sits on fastballs until he's told to go back to the bench, I don't think you can crucify Price for the cutter, though why not throw a change or a curve there - Sandy Kazmir
Let him sit on Price's fastball and see what happens - sveet
Still furious how you get beat on a cutter instead of your best pitch - sf1
I myself said something along those lines in a game recap a few weeks ago:
And really, that pitch is my only complaint with Price's performance. There are times when you try to trick a batter with your third best pitch (that you've only started throwing midway through this season), but a 1-2 count with the tying run on third isn't it. That's when you rear back and unleash 98 mph up above the zone.
Problem is, this way of thinking is bad analysis, and very possibly wrong (the one doesn't necessarily guarantee the other). I don't wish to be responsible for angering Rays fans against their hometown ace and making us seem dumb in the process, so here's a full retraction of my previous statement, and an explanation for why I was jumping the gun.
First off, let's dispense with the idea that we know he's using the cutter often in important spots. There are many crucial moments in a baseball game. What we fans have noticed is when he's failed while using the cutter in run scoring spots that ended up affecting the outcome of the game. If his offense had helped him out more, we wouldn't have noticed. If the Rays had already been down by two runs when he threw his fateful cutters, we wouldn't have noticed. If the cutters had been lined straight at B.J. Upton, we wouldn't have noticed.
To know the weightiness of the situations in which David Price is using his cutter, you would need to have pitch f/x data hooked up to measures of leverage index, which is not something I've done. Maybe the fact that I'm not working on doing so right now makes me lazy, but I'd rather be a lazy analyst who properly identifies the boundaries of what he knows than a bad analyst who overstates his knowledge.
But moving beyond that point, say that you think Price should primarily use his fastball in situations where the tying or the go-ahead run is in scoring position. Perhaps you think this is an obvious high leverage situation in which DP should fall back on his higher quality offerings. You'd still be wrong.
There are several good articles about the game theory of pitch selection. I strongly recommend reading Mitchel Lichtman's Fangraphs piece, as well as a follow-up article that JinAZ wrote for Beyond the Box Score. In the BtB piece, JinAZ takes a look at why Tim Wakefield's limp, 72 mph fastball graded out as one of the best pitches in all of baseball. The answer is its context.
Batters can hit a pitch—even a great one like Price's fastball—if they're looking for it. This is why pitchers mix their pitches. Game theory tells us that a pitcher is mixing his pitches correctly when he gets equal value from each individual pitch type. In these ratios (called a Nash Equilibrium), a pitcher will throw more of his better (from a scouting perspective) pitch types, but will throw just enough of his lesser pitches to keep the batter guessing. Here are Price's numbers, using linear weights pitch data from Fangraphs:
Percent Used |
Runs Saved /100 |
|
Fastball |
70.6 |
0.87 |
Changeup |
11.3 |
2.84 |
Cutter |
8.9 |
0.78 |
Curve |
9.3 |
-0.97 |
From this data, we can see that David Price's changeup has been absurdly potent. He's really worked on the pitch, and he could probably throw it more often than he does now. The curveball hasn't worked nearly as well, and he might help himself by throwing it less. The cutter, however, has actually gotten pretty similar results on a per pitch basis than the fastball has. He's mixing those two intelligently and effectively.
But, you argue, "We're not saying Price should use his cutter less overall, just not in the super-important, game-losing way he's done recently."
Yes, losing games is always bad, but that's the way game theory works. Imagine if the opposing pitching coaches could say, "Jake Fox, listen up. This pitcher throws a great fastball, but he mixes his pitches too. He'll keep you off balance, and you won't have a chance. Unless you come up with a man in scoring position and the game close. Then he'll just throw fastballs. So sit on the fastball and hit it out."
The whole purpose of varying your pitch selection according to a Nash equilibrium is that you stay unpredictable, and mostly work randomly. Sometimes you make the "bad" decision in an individual spot just to set up your current and future opponents for an overall better result. And this year it's worked out for Price in a big way, to the tune of eighth best SIERA (a good descriptor of pitching process, and predictor of future results) among qualified starters.
Lastly, because pitch selection, sequencing, and effectiveness is something we armchair analysts have pretty good access to data on, I'd like to encourage anybody who's interested to take a crack at proving me wrong or misguided in my defense of Price's decisions. If you've ever wanted to sink your teeth into some pitch f/x data but haven't quite known how, now's a time. I'm happy to try to answer any questions about nomenclature you encounter. Here's a spreadsheet of every pitch Price has thrown this year: PricePitches. All data is from Joe Lefkowitz. In the first worksheet, I have the pitches displayed chronologically and color-coded by type; in the other sheets, I've broken the types out to be considered separately. Keep in mind that what I'm calling a cutter, these classifications call a slider.
Enjoy (responsibly), and go play around with Joe's wonderful tool for yourself.