THE BOOK--Playing The Percentages In Baseball

Hmmm ... he said more than $1 billion over a decade.  Let’s say $40 million per team.  The gap between best and worst is $100 million, so let’s say teams go from $10 million to $110 million, with an average of $40 million.  Makes sense that it’s asymmetrical, since one superstar out can cost a lot of money in one shot.

The difference between best and worst is $10 million a season.  That’s 1 win plus, or 1 win minus.  Doesn’t seem like that big a deal.

But, as you say, we’d need to test that.

Phil - The difference I’ve seen is nearly $50m in an individual season (2009 Mets to 2009 Phillies, tho the Mets are an admittedly extreme example.) I also normalized salaries to “average” and used percentage of salary to estimate what a team would have saved over average (or not saved.) Over time, the effect might regress to the mean a bit, but in an individual season, I’d argue 3-5 wins is pretty meaningful. Ask the ‘10 Red Sox.

But that $50 million is an extreme case, right?  Over time, it evens out?

Yes, pretty extreme, but not outrageously so. Think of the Mets ‘09 as the Mark McGwire hitting 70 of injuries. Teams pretty regularly lose a real 20-25 million in a season.

Evening out is also a tough term. I’m NOT a data guy, so I dont want to mis-state anything. There seems to be three groups - one that’s good, one that’s bad, and one that varies randomly around the midpoint.

The thing that strikes me here is that even if we’re talking about one win, that’s valued at ... what $5m, if you’re signing a free agent? Making a medical staff WORLD CLASS would require maybe a million bucks. Something as simple as hiring a third trainer would cost maybe 50k. Yet less than a third of teams do that, even the good ones.

What’s the evidence that this is a skill as opposed to random variation?  Is it that the year-to-year correlation between teams is high?  Because there are a lot of things that could lead to year to year correlations without implicating training staffs.  For one thing, the players on the team correlate year to year.

The article alluded to some players improving injury wise after moving teams, but couldn’t that also be just a regression to the mean type effect?

These two examples contradict each other, they can’t both be true, but it wasn’t clear from the article how big either effect is

Phil - Here is all the amounts I got teams losing from 2002 to 2009:

Range is from .5 m to 50 million
Average = 11.7 M
Standard Deviation 8.8M
Median = 9.4 M

https://spreadsheets.google.com/ccc?key=0ApDc5PGsBzgVdG1VYjE4WmZheS0td1ZqeXBaZkxiVnc&hl=en

My info is similar to Jeff’s, so fair to go with his info.

Offhand, that doesn’t look too far from random to me.  The distribution is going to be skewed because the salary distribution is skewed.  I bet if you rephrase it in terms of seasons, instead of salary (that is, weight all full-time players the same, regardless of how good they are), you’ll get a normal distribution.

But that’s just a guess.

P.S.  If injury-prone players sign for less money than reliable players, it could just be a difference in risk-taking among teams.  Some like new cars, and some like cheaper used cars with repair bills.

Phil - here is an article I did on days and trips per team = http://www.beyondtheboxscore.com/2010/3/10/1365703/disabled-list-trends-and-team

The White Sox should be the team to examine as the trend continued into 2010:

http://www.fangraphs.com/blogs/index.php/2010-disable-list-spreadsheet-and-team-information/

Tell you what Phil—you give me the top two teams (ChiSox, Brewers) and I’ll give you any other two. I’ll bet that mine cumulatively come out ahead.

Ah, a bet.  I like it.

Phil: I would love for you to take the bottom two teams (especially since it’s not my money).

The person offering the bet typically has to give odds (because they are the one who is so certain), say, 6-5…

I took Jeff’s data, adjusted the yearly salary and lost money based on each season’s total payroll (eg 2002 bumped up, 2009 bumped down) and then adjusted each organizations totals to the average team payroll.  Here’s how it looks, over eight years:

Team DL$
LAN $168 (hello, Dreifort and Schmidt)
MON $142
NYN $129
ATL $125
TEX $122
ARI $121
OAK $118
KCA $117
SDN $116
DET $114
BAL $109
SLN $105
LAA $102
CLE $102
TBA $102
MIL $91
COL $89
CIN $88
CHN $84
TOR $84
SFN $81
PHI $76
HOU $75
NYA $73
BOS $68
MIN $66
PIT $66
FLO $55
SEA $51
CHA $35

Chicago’s the leader by far, about $17M ahead of last place per year.  Mean is $12M lost per year from a payroll of $78M.

Well, I have no reason for betting ... it doesn’t look that much more than random to me, but if Will thinks it’s real, that makes me think maybe it is.  I was just speculating.

What if you tried a random simulation?  Take all the players, randomize them among 30 teams appropriately, and see the distribution of DL dollars.  Then you could see how unusual real life is.

Will, can you tell us more about why you think this is real and not just random?

I’ll give you 7-5, MGL.

Phil—wouldn’t you see that with fantasy teams? With that, the distribution is almost completely random, tho there’s an underlying bias we can’t see - the player drafting.

Maybe we have differing definitions here. I think there IS a major element of randomness. I just think its only part of it unless a team is good or bad. Like I said earlier, there’s that “middle third” that tends to randomly fluctuate around the midpoint. When I see teams constantly at the top, constantly at the bottom, that tells me there’s more than just random ... but I could be wrong.

Sure, fantasy teams, that might work!  Good idea.

Might age have something to do with it?  Teams who spend a lot of money on free agents should have more injuries, because their players are older ... no?

Will, I have no opinion one way or another on this topic.  I was just trying to instigate a bet between you and Phil!

Looking at it in terms of dollars lost is less interesting than looking at it in terms of wins lost.

A team losing young and cheap productive talent to injuries is more costly than a team losing an overpaid older FA signing who is unproductive at the time he gets injured.

As an example, the Yankees would have been better off if AJ Burnett went on the DL for 2 months even if the dollars lost to injury would have increased (he cost them wins by staying healthy) , while a team like the Red Sox losing young and relatively cheap players like Jacoby, Pedroia, Youk and Buchholz hurt a lot even if the dollars lost was low.

If, as Phil suggests, teams choose to sign injury risks because they’re cheaper, the rankings might not change much.

Will could bet on Chicago, but since they don’t like signing injury risks, they pay up for healthy players.

It won’t mean they are any better at keeping them healthy, though.

Age may be a factor. But the White Sox are not, and have not been, a particularly young team. And I’m not sure about the White Sox not signing injury risks, either. Maybe they don’t seek out injury risks like some teams but it doesn’t seem like they avoid them, either.

Phil/17—yes, age is a factor, but not as much as you think. There are “age/injury types” that are related. Young players get injured more, but heal more quickly. Older players are the reverse and it flops at a certain age, which I’d kill to do more work on. It essentially evens out in a large population, so I tend to say age doesn’t matter. There’s a definite survivor effect too.

20—that’s demonstrably untrue. The Sox took on a TON of injury risk like Dye, Thornton, Jenks, Thome, and others and STILL kept them healthy. They’ve use their medical staff as a decided advantage.