February 5, 2010
Under The Knife
Team Health Reports 2010
It's the eighth year for the Team Health Reports and its newly named PIPP system. As always, the system is designed to give a risk rating, one of three color-coded categories-red, yellow, or green-that correspond to the risk of injury. Of course, "injury" is a broad category, and unlike Sig Megdal's system from years past, I don't try to predict the location of the injury or the severity. Instead, it's binary: either a player hits the DL at some point during the year, or he doesn't. I'm no doctor or athletic trainer, as I often say in UTK, but I'm certainly no database jockey or statistician either. Luckily, I've consulted with all four of those things-some of the best in the world, to be honest-in helping me put these together. It ends up as complex as the plot of Lost, and some days when I'm staring at row after row on a spreadsheet or talking with a scout about a player's off-season workout, I'm just as confused. I am happy that MLB has started to collect injury data on a more granular level, something they haven't done before. That data, I hope, can be used to reduce the number and severity of injuries in baseball. For pitchers alone in 2009, there was more than $250 million wasted, yet no team has put even a million dollars into their budget for research to reduce this.
The system continues to be made of up two things: a baseline risk and then additions or subtractions based on several factors, such as injury history, team injury history, and PECOTA's attrition factor. The baseline risk is an actuarial table, based on 10 years of MLB data and clustered according to age and position. That clustering often creates a bit of an outlier problem, such as with Tim Wakefield or Jamie Moyer. In the "SP 40+" category, one injury could change the baseline from yellow to red, making it very difficult to get individual reads on players. Sadly, it's a flaw I can't get around without introducing subjective factors that would defeat the purpose of having a system.
Like everything, we're hoping to improve the system year by year. While the results from last year were solid in the aggregate, there are lots of misses on the individual level. That's not going to change. One of the apparent flaws of the system is actually its strength. Risk is not certainty and just as any prediction is looking for the least error rather than the most likely, that's what the PIPP system underlying the ratings does. Let's take, for example, a "red" risk with an underlying 50-percent risk of injury, based on the baseline and the factors. That's a literal coinflip.
That player is very risky and if he's a critical player, a team is taking a very binary chance with success. But at the same time, despite his inherent high risk, there's an equal chance at health. Just under half of the riskiest band will be healthy this year, avoiding the DL and putting up numbers that will help their teams win. While that is, by definition, wrong, I'd ask you whether or not you'd like to bet the house on a coinflip or worse. That's what teams-real and fantasy-are doing with red-rated players. For a coin, heads or tails really has no upside, but a player like Ben Sheets might. One of the things that PIPP does not do is take any value into consideration. Ben Sheets at $10 million is the same as Erik Bedard at $1.5 million, as far as the system's concerned. There's obviously a real-world difference.
One of the biggest changes this year is the baseline. The actuarial table underlying the ratings was revised, as it has been each year, but this year, the bands changed significantly. There are more, creating a more individualized approach while introducing some level of sample-size error. Due to the extended actual experience involved in putting the table together and because its real use is to price the insurance teams are required to purchase on all their players, I'm pretty confident in its accuracy. For my purposes, we've seen some changes in how the baseline affects ratings. Catchers are now not automatic yellows, though they are still quite risky. There's some other more subtle changes you'll notice if you've been following along and as I write up each team, I'll comment more on these.
One thing I do want to comment on here is something I've been jokingly referring to as the "suck green." One of the factors that the system takes into account is the projected playing time that any player is expected to have. This is their exposure, for the most part, and for pitchers, it's a huge factor. Some players are projected to play such a low amount of time or are expected to be replaced that they end up green. This doesn't make them good, but their lack of talent leads to a lack of exposure and by proxy, a lack of risk. That same player, if exposed to a full season, would likely be significantly more risky. We see this a lot in catchers or in positions where a player is expected to be replaced at midseason by someone coming up.
For pitchers, the exposure is a huge factor. As we know that pitchers facing innings increases have an increased risk of injury, a pitcher who is going to be asked to go out and make 30 starts after having less than 150 innings the prior season is going to face increased risk. The system doesn't try to predict whether an organization will protect a player by shutting him down, whether they'll trade for some help, or any other possibility. It looks at the projection and says, "If he goes this many innings as projected, he's risky." The A's, with a number of young pitchers facing big increases, look very risky and they are. They also have a "shotgun defense,"a number of pitchers, just a notch down talentwise, that can come in and take spot starts or allow the team to shut guys down if need be. It's a blind spot in the system that I can't figure out how to close without making subjective guesses about what a team might or might not do, so know it's there.
The other key thing we don't know about any pitcher is his joint load. Few teams use motion capture to know this themselves, so I'm telling you now that there's no way I'm going to know. (This is about to change, however.) For years, we've guessed at what good and bad mechanics look like, hoping that we could see. We see "inverted Ws" and "picking up the elbow" and a hundred other folksy things that have no basis in reality. I'm guilty of having done this, but the fact is, every pitcher is different.
With each pitch, they're putting a load of force on different parts of their kinetic chain. It might look beautiful and be too much for the arm to take, or it might be ugly and it holds up because that player has some sort of genetic advantage. I don't know why Francisco Rodriguez has held together, but he has. B.J. Ryan held together for years. Mark Prior didn't, and when you hear people say that they know what Stephen Strasburg or any other pitcher is going to do in the future, know that they're guessing. We're nearing a point-perhaps very near-where we can get this kind of information and when we do, it might be the PitchF/X for the medhead set.
There's one pitcher, however, who comes out as such an outlier that I have to mention it. As I've explained, there's a baseline risk, then several factors that can add or subtract. Zack Greinke, last year's AL Cy Young winner, has such a ridiculously low rating this year that it bears watching. I've seen position players this low, but never a pitcher. The underlying percentage for Greinke is 12 and while Royals fans are screaming "don't jinx it!", I just have a hard time believing that any pitcher could be this low. No, Greinke's off-field issues are not factored in, but in essence, they worked in his favor.
Greinke didn't rack up big workload totals while he was below the injury nexus and now that he's passed it healthy, who knows where he could go. I often wonder if we don't have pitchers out there who could be the kind of four-man rotation workhorse that Bob Feller reminds us he was. I'd start with Greinke if I were tasked to find him. And yes, look above. We have no idea what Greinke's joint loads are and I guarantee you, neither do the Royals. Think about that when the inevitable new contract comes up for him.
The final thing I want to discuss is why I don't change the ratings as spring training progresses. For the past month, I've been working on these rankings and the writeups, putting everything I possibly can into them, knowing that there has to be a line in the sand for information. This year, it's February 1. Things will change, but the ratings won't. For one thing, I'm sure some would accuse me of "cooking the books." Honestly, it's just because I want to put these out there and let them stand, for good or bad. It's risk, not reality, and putting them down once in virtual stone works for this purpose. Injuries will happen, risk factors will change and that's what I talk about every time I write a new Under The Knife.
Fact is, most of you didn't read what I wrote above. You just want the spreadsheet so you can start getting ready for your fantasy drafts. Well, here it is. Those people won't understand the strengths and flaws of the system and will use it blindly … and usually wrongly. They'll accuse me of being wrong about an individual player, and I'll nod and say that I'll try to do better next year. Then again, you're a BP subscriber and the vast majority of you are smart. If you did skim down here, I'll advise you to take the time and not only read the explanation and introduction above, but the team reports themselves, where there's far more information than a three-color rating system. All that said, here it is. Let the risk begin.