For the past four seasons, I have written team health reports in the preseason, giving fans and fantasy players an advance look at the risks their teams face from injury. Those reports have been among the most accurate available and, along with those produced by Sig Mejdal, now with the St. Louis Cardinals, are considered the industry standard. Once again, we head into the breach of what is both a time-consuming labor and a labor of love all at once. I’m taking the task on in a slightly different fashion this year–hopefully this way of looking at health and risk will guide us to new knowledge.
The one thing that is often misunderstood by people about these reports is that they actually don’t measure health–they measure risk. While it may seem like those are two sides of the same coin, it’s not quite that simple. Risk is a measure of probabilities rather than possibilities. It is quite possible to have a high risk of injury without suffering an injury, just as it’s possible to drive home with your eyes closed. While the measure of that risk is injury, a binary result of “is he injured or isn’t he?”, the actual measure of the success of our system is the prediction of injury over a population.
We define risk in bands: normal risk is , elevated risk is , and high risk is . It’s a simple enough system, at least for those readers who aren’t colorblind. Each band is different for each position/age combo, while the risk bands are based on standard deviations. For instance, Scott Rolen is part of the 3B/32-34 band; this change from age-31 to age-32 adds nearly 3% to his risk baseline. While this may or may not accurately reflect anything about Rolen himself, the actuarial baseline holds true. When you add in his other factors, Rolen comes up as a yellow. I don’t think anyone will be surprised by that.
But what are the other factors? While position is by far the best single predictor, other factors such as injury history (functional and traumatic), body type, career path (as calculated by PECOTA’s attrition function), team factor, park factor, pitcher workload and mechanical profile (where appropriate), a scouting-based effort rating, and three other “secret sauce” factors go into the calculations. The factors add percentage points to the actuarial baseline, or in some cases actually subtract. Players at the respective positions are then compared to the non-age adjusted positional whole. That is then adjusted once more for team and league variables.
One of this year’s major changes came from a lunch I had with Sig Mejdal. Some players actually have enough reductive factors to move one standard deviation down from the positional norm. Looking back at my 2006 calculations, there were actually a few players that made it under that mark. Even better, all of those players made it through the 2006 season without significant injury. While this is a small subset of the players, I think the possibility that there is a predictable positive factor for some significant players is enough to denote them. When you see the blue designation this year, you’ll know that these are among the least-risky players in the game.
The other major change for this year is the inclusion of a new set of injury data that includes minor league and day-to-day data. While the accuracy of the information is sometimes less than 100% due to the vagaries of reporting, the expansion of the data set makes huge leaps in helping understand the risk for young players. One of my long-held assumptions was that there was an inherent danger in the leap between levels, especially to the major leagues. The fact is that the risk remains linear when all other factors–such as workload and the normative risk–are considered.
This year, we’re looking at positions rather than teams. The reason for this is that all along players were being compared to other players at the same position, much in the same way that they are for VORP. Unlike VORP, a player’s defensive position doesn’t actually change the risk. Injuries are binary, in that a player is either able to play or not able to play. Yes, some injuries are simply limiting, but there is no way to quantify that in any standard manner. The change in format is mostly to take a look at the players that are being compared against each other, which I think will be both more useful to readers and perhaps illuminating for me.
The other factor that went into the positional format and some new formatting within the reports is that many of the people that use the reports do so to help with their fantasy teams. Hopefully, we’ll get these to a point where they can stand alongside PECOTA in the pages of BP someday, making your fantasy prep even easier, but for now we’ll make this small step.
You’ll also find that we’ll be making some changes on the site that will help not only these reports, but make the daily Under the Knife columns an easier source of fantasy info. It won’t replace Rotowire anytime soon, nor will it be ready to launch with these reports, but I think you’ll find it was worth the wait.
You’ll note that I won’t write about every player. Heck, we won’t even list every player on the positional reports. For that, you’ll be able to download the handy Excel chart that gives a quick look at every single player that was run through the system. We’re getting this out there now so that those of you with early drafts will be ready, though you should be warned that sometimes I disagree with my own system. There are reds that I think are too pessimistic and greens that make me smack the monitor. The colors give you solid guidance, but the hows and whys are why I spend hours writing these reports.
Finally, one other new feature this year comes at the suggestion of Joe Sheehan. Normally, I write about the position players and pitchers projected to be the starters at the beginning of the season. This year, we’ll expand that a bit by “catching up” at the end, covering players that either won jobs in spring training, had a major change in risk due to trade or position change (but not injury–it wouldn’t be fair to soak up a couple gimme red lights), or that simply were overlooked in our initial run through the leagues.
I can’t–and won’t–cover every player. In large part, I ignore bench players and relievers because they are inherently replaceable. If a team loses a middle reliever, studies have shown that there’s usually very little bottom line effect; certainly nothing like losing a closer or even a mid-line starter. Most teams can replace backups, at least one or two deep. It’s the deep, interconnected risk that goes beyond the purview of the system and remains one of the most difficult “if/then” scenario for a team. You’d be surprised at how few teams actually work through the scenarios.
The PHRs aren’t a solo effort. They wouldn’t be possible without assistance from the entire Prospectus team, especially Ben Murphy and Bill Burke. I also need to thank Danny Romo from Global Sports Services, Sig Mejdal of the Cardinals, and Brandon Chizum for their ideas and assistance.
Without further ado, we begin the journey. Download the Excel spreadsheet by clicking here.
For the absurdly low price of $39.95 for a full year, a
Baseball Prospectus Premium subscription offers you access to Will Carroll’s exclusive Positional Health Reports, Nate Silver’s groundbreaking PECOTA projections, and daily content from BP’s group of experts on baseball talent. You’d be crazy not to subscribe.