I’ve been working on what we’ve known as Team Health Reports for the last couple weeks, really digging in over the last few days. It’s mostly an intellectual exercise for me, since I don’t have a place to publish them. At worst, I’ll put the matrix up somewhere.

The biggest issue is that one of the components of “The System” — and I’ve taken to calling it that — is now gone. Note that I didn’t say this was a problem, because it actually improved the accuracy over the last two seasons. That component was PECOTA’s attrition number. For various reasons, PECOTA has been … less than ideal … over that period. Just removing it helped, but I’ve replaced it with a very concocted number based on the risk of a 50% falloff in stats. Granted, that can be for a lot of reasons (including retirement, which I can’t figure out how to factor out …) but it seems to work appropriately, if not well. I’d love for one of the competing projection systems to come up with something on this, since I’m sure they could do it better than I could. 

The biggest improvement, as it is every year, is in the baseline. This year’s baseline for the first time includes some level of minor league data. The injury table doesn’t change, but age-risk data does include minor league players. Backtesting was only available for one year, but it did improve accuracy noticeably, so I’m including it.  Having this actuarial table underlying the system is really the key to the whole thing. Since no one else has it, it gives me a massive advantage. 

The biggest problem is still in place and that is that it measures risk, something that’s poorly understood. The System doesn’t say “this player will strain a hamstring in May”, just that there’s an elevated risk of injury. If a player has a risk of 50% (and yes, there’s an underlying number that looks like, but isn’t actually a true percentage. I created the system thinking it was and Nate Silver explained to me a couple years back that it wasn’t.) then half the time, The System will look “wrong.” One baseball season isn’t a very big sample size for injuries, but over the eight years I’ve done this, it’s come out pretty well. 

I don’t have a clever acronym for it, but maybe you do. Whatever you call it, it is something that fans, fantasy owners, and even the teams themselves look forward to. I only hope I get to show you this year.