One of the most frustrating things about developing a new study is that you don't have any kind of baseline to which you can compare your initial results. So, when I attempted to quantify Felix's release point a few days ago, I thought the resulting data was interesting, but, lacking numbers for other pitchers, I had no way to know for sure.
So I got to thinking. And, as is usually the case when I'm thinking about something difficult and Mariner-related, I asked Dave what he thought. His suggestion: look at Joel. The Mariners were playing around with his mechanics all year, so if anyone was going to turn up some interesting results, it'd be him.
And so I looked at Joel.
Series of pictures and numerical analysis below. The methodology is the same as it was for Felix, only this time I looked at three starts instead of one - April 15th, May 24th, September 22nd. Remember, the release point dimensions listed represent 95% confidence intervals.
4.6x4.1 in., 20.3 square inches (area)
4/15, BREAKING BALL
7.5x5.4 in., 40.4 square inches (area)
4.0x3.2 in., 12.7 square inches (area)
3.7x3.2 in., 11.8 square inches (area)
5/24, BREAKING BALL
4.3x1.8 in., 7.7 square inches (area)
2.8x1.6 in., 4.5 square inches (area)
3.2x2.5 in., 8.1 square inches (area)
9/22, BREAKING BALL
5.1x4.7 in., 24.2 square inches (area)
3.7x3.4 in., 12.5 square inches (area)
Fastball: 4.9x2.6 in., 13.0 square inches (area)
Breaking Ball: 4.0x2.4 in., 11.0 square inches (area)
Overall: 3.0x1.6 in., 4.8 square inches (area)
What I'm most interested in is that last number - by finding the area of the box formed by the overall x,y 95% confidence intervals, we can get a real good idea of the variability of a guy's release point. So, let's just look at those:
Felix, 10/2: 4.8 square inches
Joel, 4/15: 12.7
Joel, 5/24: 4.5
Joel, 9/22: 12.5
In two of the three games I studied, Joel had nearly three times as much release point variability as Felix - which makes sense, because Felix is awesome, Joel isn't, and release point plays a large role in that. This is valuable information to have, because now I can go forward knowing that I'm not wasting my time with this stuff, that even on the Major League level, there is some significant degree of inconsistency.
But look at the May 24th number. You may recall that, after that game, I criticized Joel's mechanics and Dave criticized his release point. As it turns out, he was a little more consistent than Felix. What does this mean?
Go back and read Dave's post again. I'm going to quote (at some length) the relevant bits:
He then faced rookie Jeff Fiorentino, and the mechanics went to hell again. Early release, fastball up and away. Early release, fastball up. Early release, curveball up. Early release, fastball away. Four pitch walk.
He wasn't "missing his spots". From when he released the ball, it had no chance of being a strike. His fastball was getting out of his hand before his body was in proper position at least 60 percent of the time.
Pineiro had more success with the curveball, allowing his body to rotate before letting the pitch fly, but even still, he was early at least a quarter of the time. Whether it was a clear revelation that his release point was more consistent on the offspeed stuff than with his fastball or not, he threw significantly more of them as the game went on.
Main points: (1) Joel's breaking ball release point was more consistent than that of his fastball, and (2) he repeatedly released his heaters early.
What do the numbers bear out? As far as point #1 is concerned, Dave was right - Joel's release point was more consistent when he threw breaking balls, as you can gather from looking at the confidence interval areas listed above (7.7 square inches for the breaking ball, 11.8 for the fastball).
But point #2 brings up something that bears mentioning - release point consistency is good and everything, but if you're consistently releasing the ball in the wrong place, then you're screwed, because you won't be able to throw strikes. Understand that what I'm doing isn't at all a measure of release point quality. Studying something like that would require very specific knowledge of where a guy's trying to throw the ball and how much his pitches tend to break, which is just impossible. All we can do is point out how often a guy lets go of the ball within a certain area. Joel's fastball release point against Baltimore on May 24th was awful, way early, but it was consistent, which makes it look good in the numbers.
It's also worth mentioning that this is only a 2-D analysis; we can't really be sure about the z-axis location of a ball upon release, because all we're given is an image from behind a pitcher's back. So, again, we can look at how consistently a guy releases the ball within an x,y plane, but the only way we can know if he released too early or held on too long is to watch where the pitch ends up.
One last thing before I finish: look at the September 22nd game against Toronto. Not only was Joel inconsistent with his breaking ball release point, but further analsis of the numbers reveals that his curveballs and fastballs were released at significantly different (95% confidence) points. If the batters were able to pick up on this - and we're talking a matter of inches, here - it would go a long way towards explaining why he got lit up for seven runs that day.