Comments on: Injury Chances for Strike-Throwers

By: rgathman1115

rgathman1115 — Tue, 30 Oct 2012 01:58:08 +0000

A multitude of variables come in to play when you talk about the reason for a DL visit. Personal fitness and strength, lifestyle and diet habits, lifetime pitch/throw counts, throwing mechanics, freak injuries and many more. Due to the countless variables that can’t be quantified it is very tough to say that any logical relation can be made between throwing strikes and DL visits.
What I believe Beane is indicating is that he prefers pitchers that focus on executing pitches and working the zone instead of pitchers that try to light up the radar gun, or overthrow. Overthrowing can certainly lead to increased injuries due to added stress on the body.
Consider pitchers in the early 20th century that would throw both games of a double header with a pitch count reaching the 300s. Their style was focused on location over velocity, with a heavy reliance of defense. Those pitchers lasted because they weren’t maxing effort out on every pitch. Fireballers today, that Beane tries to avoid, seem to try to throw maximum velocity from pitch 1 to pitch 100. That maximum effort in every pitch is going to add an undeniable strain that a “strike thrower” wouldn’t face when focusing on location over velocity. To randomly quantify that logic, a fireballer is throwing roughly 1.1 pitches to every 1.0 pitches a “strike thrower” throws. At the end of a game both pitchers throw 100 pitches, but with the added stress a fireballer endures after max effort on every pitch, it is as if he has thrown 110 pitches. Now, these numbers obivously have no real statistical backup, they are used simply to help illustrate my point.
Also a statement Kurkjian made that interested me was “younger players, in theory, are less likely to get injured than older players”. Is there any statistical validity to this statement?

By: Jeff Zimmerman

Jeff Zimmerman — Sun, 28 Oct 2012 19:34:13 +0000

Viva – I just ran the numbers. and there is some signs that show high K pitchers are a bit more injury prone.
K%: DL%
>21%: 39%
18% to 21%: 41%
14% to 18%: 36%
12% to 14%: 32%
< 12%: 34%

I really like the Pitchf/x Zone% for looking at the values.

By: vivaelpujols

vivaelpujols — Sun, 28 Oct 2012 11:33:18 +0000

Something I noticed was that pitchers who threw more *strikes* actually had a slightly higher DL% than average while pitchers who had a higher Zone% had a lower chance of injury.

I think what you’re seeing here is that pitchers with better stuff have more out of zone strikes and are more likely to be injured than pitchers throw an equal amount of in zone strikes but fewer out of zone strikes.

So I guess Beane’s theory holds up, but I’m also guessing that injury rate is positively correlated with a high K rate, which is probably the best indicator of performance.

Jeff, how does strikeout rate in year n project injury rate in n+1?

By: Patrick

Patrick — Sun, 28 Oct 2012 00:19:53 +0000

Nice write-up. Definitely adds some clarity, but I agree more granularity is needed before we make an absolute conclusion (which may never happen :))

First, I agree mechanics can be a major factor in reducing or eliminating injuries. Its clear from watching the likes of Andy Pettitt that having a clean repeatable delivery can keep one healthy and effective over a long time. That being said so many other things go into being able to do that-that we may become overloaded with data and start assuming all sorts of biases.

Ultimately, we should also account for TYPE of injury. It would be incorrect to assume that poor mechanics would lead to something as freakish as a hip injury incurred when running the bases (Chen Ming Wang) vs. someone blowing out their elbow multiple times as a result of overthrowing (Zumaya). I dont know if that was accounted for here. The same would go for liners back to pitchers that cause injury and perhaps other types of injuries that may be construed as not mechanics related.

The last point here is that it may be incorrect on Beane’s part to assume young pitchers have a lower chance of injury than older pitchers. I would use opposite logic to posit that older pitchers have a better chance of surviving injury for the following reasons:

*I have no data to back these up……

1: Older pitchers probably do not overthrow or rely on “Stuff” to get guys out, but rather are more capable of using the strike zone to get guys to chase.

2. Can get out of jams better by not giving in, thus limiting pitch counts.

2: Are already older pitchers, thus having already established themselves as reliable pitchers, thus meaning less time on DL. What I mean is, a pitcher who spends time on the DL probably doesnt last as long as someone who doesnt spend time on the DL, so being an older pitcher, one would assume they didnt spend a lot of time on the DL….right?

By: Ben Hall

Ben Hall — Sat, 27 Oct 2012 02:29:10 +0000

This is bizarre. Are you actually suggesting that Jeff’s large sample that supports the idea is nullified by your one example?

By: Nathaniel Dawson

Nathaniel Dawson — Sat, 27 Oct 2012 00:59:21 +0000

Yeah, I don’t know. I look at the data you present and it looks to me like more noise than any kind of indication of anything. The numbers are all very close, and seem to be somewhat contradictory when looking at the “strike-oriented” metrics. The one result you singled out as most interesting isn’t that far off from the norm. ~38% of all pitchers in the study hit the DL the next season, while only 34.5% of the high zone% did so. You chose equal numbers in each group, so the “others” hit the DL at a 41% rate. Or, the difference between that group and the others is about one in eighteen pitchers. Maybe that has some significance, and is worth looking into further, but there’s nothing from what I see to corroborate or confirm anything. To me, it looks likely to be randomness more than any particular pattern.

By: Kyle Boddy

Kyle Boddy — Sat, 27 Oct 2012 00:53:55 +0000

Interesting case study!

Aside from the comments mgl pointed out, have you done more rigorous statistical analysis on the data to see how likely these effects are occurring compared to random chance? (t-test, regression, etc)