Toward a Science of Pitching Injury?

I began thinking and writing about this subject weeks ago, but life happened and I couldn't finish the post then. Life continues to happen, but I don't care. I haven't been able to stop thinking about it, so I have to stop and write it down.

It started simply enough. In an Open Thread, snotboogie posted this link.  It's about the long toss debate. Several folks made comments along the lines of Osfan21's "nobody really knows yet whether certain mechanical traits are more likely to result in injury." That's true. But I thought ‘why not?'  It seems like an empirical question with an on-going natural experiment. Some organizations are using 360' toss, others not -- compare the outcomes. There are too many uncontrolled variables for something that rough to lead to any conclusion, but if there is a significant difference it could be a pointer toward the value in studying it more.

This glance toward experimental analysis would likely have passed in a moment, if a day or two later 2632 hadn't linked to Jon Shepherd's Camden Depot post on the Science of Baseball.  I was pleased to see that some folks are thinking in a scientific way about injury, but noticed that the subjects are all at the HS and college level. Why isn't MLB injury the subject of scientific study? Publication of such studies would give away a lot of proprietary information, so realistically I don't expect to see it such things show up in journals. But I don't hear anything that leads me to think it happens at all.

The moment that fixed this subject in my mind came when, in response to my last pitching post, the right honorable higgins pointed me to Buzz Bissinger's 2007 Times article on Kerry Wood. I was just knocked out by this quote:

To get a better idea of how much time pitchers lose to injury, consider that 244 of the pitchers who played in the majors in 2006 have been on the disabled list at least once in the past five years. Their injury time adds up to 27,351 days, the equivalent of 149 lost seasons.

149 seasons! To me, that says one thing. The money to improve the situation is stacked all over the table and falling down onto the floor. Given the average 2011 player salary is $3.3 million dollars, that's roughly $492 million wasted over 5 years. Call it $3 million per team per season.

So try to avoid wasting some of that money.  Invest some comparable amount of money in developing an improvement program to reduce your injury losses on pitchers. Say it's your extra 2%. If you have $100 million payroll, set aside another $2 million toward a science of pitching injury.

Since we're starting from "nobody really knows," the initial investment would be all about data collection. I'm thinking that you would want to collect the following, for each pitcher at each level of the minor leagues:

Throwing program information - kinds of throwing used (long, short, bullpens), frequency of sessions and number of throws

Training habits - have players or training staff record the types, reps and frequencies of non-pitching workouts

Pitch f/x data - if these cameras are not constantly running in the minors, they should be

Data on mechanics - this would be the hardest. But I think you could crack it by videotaping every pitcher in every outing from standard camera angles. Then have trained staff play the video in slo-mo/freeze frame on touch screen computers, using a stylus to tap-enter the positions of the pitchers joints at standard intervals through their delivery. Resource limitations would determine how much of this data you could input, but you would at least have mountains of primary data available.

Pair those input variables with outcome measures of the type and severity of the various injuries (e.g. exactly positions where the fabrum bear nested and the depth of his den in millimeters) and have statisticians run regressions like mad. Somewhere in there you'd have to find inputs that appear predictive of increased injury.

I'm sure some of you are thinking ‘sure, and you'd have a list of correlations, but nothing about causation.' Exactly right. The predictive variables would be starting points for experimental manipulations in your farm system. Can you change, in a controlled fashion, the mechanics or training programs of pitchers of similar build and see a change in injury rates compared to the historical baseline? I have no idea how such experiments would actually be designed. I'm sure nobody in an MLB front office does either. So you would need to partner.

Imagine a city where there is a research-oriented university hospital within walking distance of the major league stadium. If it's not too hard to picture, imagine baseball fans use their parking lot. The MLB club offers access to this injury-science program to that university's physical medicine department to use in teaching research methods. Undergrads in the primary data collection practicum are tapping your computer screens in the basement. Grad students are writing up the results for simultaneous submission to the front office and the prof. The prof is shaping the design of the trials by suggestion to the grad students in the experimental methods class. And the cost of the whole program, especially the mechanics data, comes down a bit. $2 million plus a supply of free labor would get a lot done.

To me, it's an article of faith that careful observation like this would yield actionable knowledge that would save valuable arms and lower team ERAs. Maybe not, but if you run it for a decade and there's no indication of that, then you've spent what we're paying Brian Roberts to be dizzy for a year and you cut bait. If my faith is warranted, it could go radical places. Cue the fuzzy-screened dream sequence.

Once you had the pitching mechanics technology in place, you could push it both up and down from the minors. On the upstream side, when a pitching coach is scratching his head about a pitcher's struggles and can't eyeball any change in delivery from video, taking new data points and running them against historicals could improve in-season corrections.  Or someone monitoring changes to tell you when a pitcher is compensating for soreness even while succeeding.  On the downstream side, take James F's CamdenCast suggestion to videotape a lot more work by prospects before the draft. Use this same process and maybe you would draft smarter.

The biggest benefit of all could be that by reducing injury, you not only protect your current investment in your pitchers, but also make pitchers cheaper over time. Those 149 lost seasons inject a lot of scarcity into the pitching market, driving up prices. Reduce it and, fantasy of fantasies, maybe Kevin Gregg's price starts to align with his value. Maybe he can't get a job at all. Isn't that a goal worth investing in?