In fantasy sports it is all the rage for experts to tout their “world renowned statistical model headed by the industry’s brightest minds” that have apparently won millions for themselves and their clients just “make sure to sign up for the mid-season special by clicking here!”
Obviously, I’m being a bit facetious as I’m on board with advanced analytics. Math is a long-neglected tool that all fantasy players and real football teams should use to their advantage as it has shown to be more effective than being on team #EstablishTheRun.
The problem I have with statistical models is when they bleed into my specific line of work, which includes injuries, recovery, and human performance. There’s not a person on this planet, myself included, that can take a player’s previous injury history, plug it into a system, and spit out a quantifiable injury risk profile. I’ll give you the top four reasons why injury risk profiles aren’t useful.
1. Statistics don’t tell the whole story
Statistics play a role in predicting missed time, don’t misunderstand what I’m saying (see my hamstring article here). However, if statistics are bastardized and used to tout “injury risk profiles” I tend to draw the line. Keenan Allen is the perfect example, as most injury risk sites would have considered him a “high injury risk” prior to the 2017 NFL season. He broke his collarbone, lacerated a kidney, then tore an ACL. No medical professional with any semblance of a scientific background would even pretend that those injuries are related in any way anatomically or otherwise.
Another example is Leonard Fournette, who since 2016 has had ankle sprains, hamstring strains, and a thigh contusion and is subsequently considered by some models as a “medium to high injury risk””. However, there’s not way to quantify a correlation between hyper stretched ankle ligaments, a hamstring tear, and a thigh bruise.
So, what exactly makes these two players an injury risk?
2. We haven’t figured out general injury risk
Listen, I’m not saying I have all the answers, but I am saying that models aren’t it. For years now researchers have been in search of the holy grail that is “injury prevention”. Highly intelligent people like Tim Gabbett have conducted research on acute: chronic workload ratios and injury rate, and have had only minimal success. (reference 1)
This study lumps together contact, non-contact, and even “freak” injuries that occur in players. The result? Players with “high” levels of physical work may be more resistant to injury than their counterparts. Notice this says nothing about lower extremity injury risk, contusion risk, or hamstring strains. It’s strictly a study on relative injury resistance specifically for elite rugby players.
3. We haven’t figured out specific injury risk
Physical therapists like Mike Reinold are constantly tweaking the best criteria to help patients return to their sport without tearing their ACL again (or tearing it in the first place). Despite these efforts, ACL tears still occur often, especially in female athletes. Additionally, some injuries like first time concussions and shoulder dislocations are simply a matter of luck. So, if we haven’t figured out how to predict a single common athletic injury, how would a static model even begin?
4. We’re better off evaluating individual players
But if we’ve already established that “injury” is an umbrella term with several different factors at play, how can we evaluate players? The answer is to start with the individual’s issue and work backwards. In my injury article discussing Aaron Jones, I use medical data, anatomy and physiology, and a past injury history to form my injury thoughts on Aaron Jones’ specific situation. The outcome? Aaron Jones is completely healthy through Week Five and just went off for four touchdowns! The point I’m trying to make here is that even an individual analysis and clinical assessment of past injury history cannot correctly predict a player’s injury history, but it’s the best chance we have.
My point isn’t to promote myself or disparage others. I legitimately do not mind if you unfollow me on Twitter and go follow Stephania Bell or Matthew Betz. Why? Because a trained individual’s clinical opinion with player specific context is your best shot at coming close to predicting injury risk.
Reference 1: https://bjsm.bmj.com/content/50/4/231.short#
Edwin Porras is a Contributor for the Unwrapped Sports Network website. Follow him @FFStudentDoc on Twitter.