Thursday, January 24, 2008

Thursday Grab Bag: Pitchf/x, Arms, PAP, Prospects, Baserunning, & Projections

Eating dinner before class, and I thought I'd post a link dump/grab-bag post that I've gradually been accumulating over the past week or so:

Jeremy Affeldt's Pitches

There's been some talk that Affeldt might try to add a change-up to his repertoire as a "third pitch," in hopes of improving his ability to perform in the rotation. He did indicate that he threw it a couple of times last year, but that they weren't a common pitch.

Interestingly, here's what Josh Kalk's player card on Affeldt reports:
TypeBreak in x (in.)Break in z (in.)Initial Speed (MPH)Number ThrownPercentVersus RHBPercentVersus LHBPercent
Fastball5.4712.3394.4719058.2810954.778163.78
Curve-3.88-2.3878.125516.872814.072721.26
Slider-2.45.8879.26519.944723.621814.17
Change3.5511.6888.62164.91157.5410.79
It confirms that he did throw his change-up now and then last year--16 times among those games tracked by pitchf/x. The ~6 mph difference in speed isn't particularly tremendous. If a ball travels ~55 feet from the time it leaves a pitcher's hand to the time it arrives at the plate, a 6 mph difference amounts to a difference of ~0.03 seconds in when the pitches arrive. Maybe that's enough to disrupt a hitter's timing, but I'm skeptical. But hey, pitchers can always improve, so who knows?

Another interesting thing about the pitchf/x data is that it also shows two discrete breaking balls, which he hasn't mentioned in his interviews:
They both come in at about the same speed, but one has negative vertical break, while the other has "positive" break. I think this means that the two pitch types must have distinctly different spin. Kalk's algorithm classifies the negative breaking pitch as a curve, while the "positive" breaking pitch is classified as a slider. Their horizontal break is (mostly) the same.

Anyway, I just find it fun to compare what players/teams tell us about their pitches, and what the pitchf/x scouting data actually report.

Outfield Arm ratings, 2004-2007

John Walsh has released a spreadsheet with his estimate for outfield "arms" ratings for all players from 2004-2007. A huge thank you to him for this, as I'm sure it took a fair bit of extra work to get all the data ready for release. I'm steadily working up my '07 review (though free time has been hard to find since the start of the semester), and it will include those data in fielding assessments. I'm planning to calculate my fielding estimates for outfielders as:

[(ZR-runs + PMRorRZR-runs)/2 + WalshArms]*0.75 + 0.25*FansRuns

Infielders will continue to be:

(ZR-runs + PMRorRZR-runs)/2*0.75 + 0.25*FanRuns


Do Pitcher Abuse Points work?

With Dusty being the Reds' new manager, I'm sure that we'll be seeing a lot about Baseball Prospectus's pitcher abuse points. The question is, of course, can pitcher abuse points be used to actually predict injury? Tom Tango did a small study on this and found absolutely zero indication that high PAP predicts reduced workload in the future (due to injury, or due to lack of effectiveness via hidden injury). In fact, in several of his variations, he actually found that more "stressed" pitchers threw more pitches the subsequent year!

Certainly, there's some selection bias here--pitchers who can take it tend to throw more pitches year after year. The Harang/Arroyo comparison from last year is sort of a case in point. But the key finding here is that whatever effect PAP has on pitchers, if any, isn't large enough to make clear predictions about future injuries, at least in a coarse way (which is typically how folks try to apply them).

My guess is that the answer to this lies not in a stat like PAP (population-level assessments of workload), but rather in very individualized approaches. Maybe just including information about pitcher height and weight in a PAP model would be enough. But I continue to think that the best approach is a more "biological" approach, like what Boston does with their young pitchers. Hopefully, the Reds do or will do something similar to this in the future.

Nevertheless, let's try to keep a level head when we see the pitch counts mounting on Bailey, Cueto, etc, this year. We can be concerned, but we don't have to immediately call for Dusty Baker's head.

Reds popping up in various articles

A few highlights from some recent studies or features that featured Reds or former Reds:
  • Kevin Goldstein released his list of the Reds' top 11 prospects. There are a lot of similarities with John Sickels' Top 20 prospect list, but some modest differences of opinion as well.
  • Dan Fox released some historical baserunning work, and it's fun to see where some Reds fall out:
    • Barry Larkin ranked 10th among all players from 1956-2007, according to the sum of all of Fox's various baserunning stats. Joe Morgan came in 14th, Eric Davis 20th.
    • Pete Rose ranked among the worst all time on stolen base plays and advancement on groundouts, but among the best all time on advancement on fly outs.
    • Joe Morgan was the best in history (1956-2007) on advancing via balks, pb's, or wp's. So he must have really freaked out pitchers.
    • Tony Perez was the worst in history on advancing via fly balls, mostly because he got thrown out at the plate a lot in 1980-1981.
  • Xeifrank has been trying to predict division outcomes using his custom simulation and ZiPS projections. Amazingly (at least to me), he has the Reds near the top of the pack in the NL Central.

5 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. According to Josh Kalk, the average change-up is thrown about ten m.p.h. more slowly than a fastball. So it looks like Affeldt has some more work to do.

    ReplyDelete
  3. I wonder what a week with Mario Soto would do for Affeldt's change up....

    ReplyDelete
  4. Justin,

    Just incase you didn't see... Fangraphs just changed their leader boards so that you can see more than 50 names per page. Good thing I didn't start copying and pasting any data yet :)

    dan

    ReplyDelete
  5. @Ken, I should have checked that article. 10 mph would result in a 0.05 second difference, which still doesn't seem like a lot to me. I was thinking ~0.1 seconds might be what would be needed. But Johan Santana's change-up is "only" 10 mph different from his fastball, and his sure works well...

    @Dan, I hadn't seen that--thanks! Sure improves how useful his site's output can be! I should send him a thank you...
    -j
    -j

    ReplyDelete