UTK Special 3/16/22

Predicting Injury is Impossible

For someone with a title like “Director of Bioanalytics”, saying that something like predicting injuries is impossible raises some eyebrows in pitch meetings. For years, people - including myself at times - have been trying to find some model that accurately models injuries or at the very least injury risk. Thus far, no one has been very successful.

My model, which I used for Team Health reports for better than a decade, was pretty good. Not great, but pretty good. The system was accurate for large numbers - I can tell you about how many pitchers will have Tommy John surgery this year, I just can’t tell you who - and much less accurate for individuals. The risk adjustment was worthwhile, but was based on publicly available injury data and history, team context, and a few other factors.

Today, we have far, far more data, from the semi-public* Statcast system to far more sophisticated internal systems, wearables, and modalities. We still can’t project injuries, largely because the systems in place are failures, overcomplicated and yet far too simplistic to take on a task that should be so holistic.

I can tell you how it could be done - give me all the data - but what I’m going to tell you today is, in fact, the impossible. I’m going to predict injury in 2022.

In 2022, I predict there will be 49,274 days lost to the IL and if it hit 50,000 I would not be stunned. That’s only a four percent increase**, but there’s a couple factors.

First, the league is shifting back to 15-day DL stints. While teams were more inclined than ever to use the List since it “only” cost them 10 days the last couple years, there were not a bunch of phantom injuries. While there will surely be some situations where a team will choose to play short-handed for a couple days, there’s not going to be many. Roster construction is thin when it comes to position players and for pitchers, teams are already trying to figure out how to deal with new options limits when it comes to the bullpen.

I spoke with a couple team officials who agreed with this idea. One front office exec feels that “arm soreness” will be on the rise this year, in a real way. “Say a guy comes in, tells the doc his arm is sore. That happens all the time and [the Athletic Trainers] get him through. Now, you’re thinking about a couple of these guys who are sore and you put one of them on the IL for a while, bring up a guy. When one comes off, another goes on. You’re saving options and saving the pen at the same time.”

Another exec thought we’d see longer stints for starters. “Not the top guys but if you lose your four or five and you will, there’s not as big a drop-off so you IL him, bring up your six or seven or where you’re at and you leave him there a bit longer. Make sure that four is one hundred percent, rested up, maybe an extra rehab start.”

Those two will add to days lost, as well as the general tendency to ‘protect’ players with an IL stint. I’m not overly concerned about the lockout’s effects, as most players work out on their own in the off-season anyway. Sure, the medical staff couldn’t check in, but I don’t think this is a major issue. If there’s any sort of hard spike, it’s likely to be in muscular injuries (backs, hamstrings, and quads) for hitters and different muscles for pitchers (shoulders, biceps, and lower back.) Those tend to come in camp and seldom are serious enough to carry through a normal camp, but with less time, that’s possible.

The rest of the increase is simple trend. It’s a trend that’s been rising and without significant change, there’s no reason to think that the red line and green losses won’t continue. We’ve seen increases in arm injuries, increases in rehab times from those arm injuries, and even when you take out COVID, the trend is clear that in any way you slice the data, injuries are going up. I think this also takes into account that teams and doctors have gotten better at managing injuries; had they not, I think the numbers would be up by twenty percent rather than four.

If the average salary stays roughly similar to 2021, then that would mean that we could be looking at another record for dollars lost. Yes, I know injuries don’t care about averages and that losing Max Scherzer for a 15-day stint would cost much more than anyone else in the game, but that median salary ($1.15m) can be divided to the normal 180 days of the season, then multiplied by the days lost. That would be $314,811,586, which is under the estimated totals I have seen from last year, which were closer to $400m.

If the total ends up closer to the average salary ($4.17m), it gets much uglier. I’ll save you the math - that total would be $1,141,481,484. The exact numbers and days will of course be different.

While it is impossible to predict injuries, it’s easy to see where this is going. Baseball hasn’t made significant changes to the system and is likely to have the same issues. The changes all seem to be coming from outside of teams and when teams do make moves, they seldom wait for the long-term changes, acting from day to day as much as they do year to year.

The advantage of managing team health, reducing injuries to near-zero, and to save hundreds of millions of dollars over the course of the next decade on a per-team level still exists and the first team to get there has a chance to hold that advantage for years. Now, who will be the owner bold enough to actually do it?

*By “semi-public” I mean that much of the data from StatCast is available to the public, but not all. We know that teams are contracting with third parties to read the data - the best known is Reboot Motion - and that there may be more data from Hawkeye that is not put into the data stream. MLB should get full credit for making any of it public and keeping it there, which has led to a lot of innovation.

**According to Derek Rhoads of Baseball Prospectus - who is flat out a data wizard - last season had 47,353 days lost to the DL.