Pitching by the Numbers: Grading curve

This is the second column this week because the first one tripped me up and I misread the category of data that was the foundation of the piece. I thought I was looking at per-100 pitch data and I was really looking at just runs saved on the those pitches regardless of how many of them the pitchers in question threw (so quantity likely would trump quality).

That was exactly what I was trying to avoid. So let's do this again the correct way.

In last week's column, I tossed in an aside about how I did not like pitchers whose out-pitch is a curveball, promising to explain why in a future column. Sam K. wanted to know now. So we'll postpone our first look at some under-appreciated relievers until next week and provide the answer.

First, here's what a scout told me a few years ago when we were casually watching an Arizona Fall League game:

Me: "Most overrated pitch?"

Scout: "The curveball. I'm never impressed when I hear a guy has a plus-plus curveball. How many curveball specialists are there?"

The entire piece is highly informative, thanks to him. And it can be read at SNYWhyGuys.com.

But that wasn't enough for me. I need data. Surely if he was right – and I respected that he was – there must be proof. And there was as soon as Fangraphs came out with their pitch value data. I planned this piece without even checking this year's numbers because they are always about the same.

The following is pitch value per 100 pitches of that pitch in question. (Note: This is the RIGHT way this time; thanks again to the alert and vigilant commenters for catching my mistake.) So it's all normalized. The best pitches, therefore, should all be equally valuable at preventing runs. But clearly, they are not.

Best curveballs through Sunday (runs prevented per 100 curveballs):

Similarly, sliders:

And, finally, changeups:

That's more like it. So if you're best pitch is a curveball, you're not going to save as many runs per 100 pitches as the best slider and changeup pitchers. How much less? Well, judging by this sample, close to one run per 100 pitches even if we seek to minimize Cahill's off-the-chart slider ranking, figuring it's an outlier number that is very likely to regress.

It's funny, but when I posted the version where I imported the wrong pitch-type column, one of the posters correctly pointed out that I should have known things were screwy because there is no way guys could prevent near double or double digit runs per 100 pitches. And there Cahill does it even when we sort the right way.

So what's the fantasy/scouting takeaway? When you see a pitcher and you notice he's getting bad-looking swings from professional hitters, it's important to place extra weight on the pitch he's relying on to do that. In order of expected, run-preventing value they are: changeup followed closely by slider and then a big dropoff to curve. In the first version, the slider scored better and that's not the way I remembered it. This should have clued me in that I was looking at things the wrong way. But, with Cahill, again, the slider (in his case anyway) has by far the best per-100-pitch score.

However, sliders ruin arms with much greater frequency than changeups, so that's why the plus change is the pitch we should most value when eyeballing prospective pickups. Yes, I understand that it's Hanson that's hurt now. Pitching is a very dangerous thing for the arm no matter what you are throwing. The risk is always there. But it's more pronounced with the slider, the most violent pitch from a biomechanics standpoint.

Finally, if we stipulate that the curve doesn't prevent runs in line with other pitches, why is that? This is where theory comes in. In science, theories can never become facts, they can only explain facts. I'm more interested in the fact in this case. But my best theory is that curveballs fool umpires into calling them balls even when they are strikes. Plus, the lowering of the strikezone that's taken place over the last 30 years or so has really killed the curveball pitcher. Hitters can afford not to swing and take their chances with the friendly ump.

Michael Salfino writes and edits the SNYWhyGuys blog that projects player and team performance for New Yorkers. He's also a quantative sports analyst whose writing regularly appears in the Wall Street Journal.