THE BOOK--Playing The Percentages In Baseball

That Mark Prior got hurt before reaching free agency undoubtedly kept him from the top of this list.

Jeff had a bunch of articles analyzing injuries in different ways about a year ago, but had to take them down because another site argued BtB infringed on their data rights.  We disagreed, but didn’t want to push it because of other relationships with the site.

Dodgers have quite a tally at the top of that list.  I believe insurance did cover some, but not sure how much.  When you add to that the horrible contracts given to Pierre and A.Jones - what a waste!  The Dodgers were so lucky to have so many good players making nearly league minimum.  The party is over though and the margin for error for Colletti is very slim.  The young-ins are making more money now.
vr, Xei

Sky: well, now Jeff can re-run his analysis, using the new (and updated) data.

Email me which site said this.  They’re going on my “on notice” board.

I like the idea, and this looks like a good starting point, but I’d like to see the same thing done with a few refinements.  I replicated Jeff’s results by linking Josh’ injury database to the salary table in the Baseball Databank, dividing the total days on the DL for each season by 180 (appr. number of days in a season), and then multiplying that percentage by the player’s salary for that year.  I get exact matches with Jeff’s list for the dozen or so names I randomly spot-checked.

The first suggestion I would have would be to remove off-season DL stints, since those wouldn’t count toward the percentage of the 180 in-season days missed.  I looked at the same query with DL stints Josh marked with the offseason flag removed, and that made a difference in the rankings for some players.  To do this accurately, you’d have to include a start date for each season and then split stints at that date to account for DL stints that overlap the off-season and in-season.

The other considerations I’d like to see are a bit more subjective.  I think there are cases where the percentage of in-season days missed multiplied by season salary is not a very good estimate of the money cost to a team by a player being on the DL.  In particular, I think this is true for players with back-loaded contracts and players signed at some kind of discount for injury-risk-related reasons.  For example, Jeff Bagwell’s salary was high in his later years when he was missing a lot of time, but it wasn’t really costing the team his full salary, because, in essence, his high salary in those years was paying for the production he gave them in the earlier, more productive years of the contract.  There are a few ways you could account for this.  I’m thinking maybe something like redistributing salary, for the purposes of the study, between years to follow Tango’s .5 win per year decline model to estimate the production from each year the team is paying for, or maybe something like calculating a total expected production for each contract using a dollars-to-wins conversion and then seeing how far short a player was of that value, and then splitting the shortage into value lost by level of performance vs. value lost by being on the DL.  If you were to do it that way, I would suggest prorating the WAR produced to cover the days missed on the DL, and then subtract that from the expected production to get the value lost to level of performance, with the remainder of the difference attributed to DL time.

Example:  Player A was paid for an expected 10 wins on one contract from 2002-2009, was on the DL for 288 in-season days, and produced 6 WAR.  288 days is 20% of his total in-season days, so extrapolate his 6 WAR to cover those extra 288 games:

6/.8 = 7.5

So his rate of production was for 7.5 WAR over those 8 years with the DL stints removed.  That’s 2.5 wins less than what was expected from his salary, so that’s 2.5 wins cost by level of performance and 1.5 wins cost by being on the DL and not providing his average rate of production over that time.  Use the same dollars-to-wins conversion to value that in terms of salary.  Obviously, this kind of analysis would depend on what method you use to calculate dollars to wins.  For this, I’d probably go with the method Sky used a couple months ago.  Also, you’d run into problems with players who missed most of their contract on the DL because their WAR rate production won’t be very reliable, so maybe you would want to regress the observed WAR production rate to the value the team paid for based on how much time was missed vs. how much time he actually played.

This would be a more complicated question than Jeff is answering here and take a lot more subjective decisions than Jeff’s look takes, so I think it’s good to have a simpler, more objective place to start from like Jeff gives.  Now, it would be nice to see some follow up work, be it from Jeff or whomever else.

Ditto Kincaid.

I don’t know if it’s actually worth it to do it that way, if you’ve got a huge group of players to begin with.  At the player level, sure.  AT the team or league level… ehhh, probably not.

But, excellent point on the “days lost” in-between seasons.  If Jeff is counting those, then he’s got a problem, and that might explain why his numbers might be a bit higher than they should be.

I didn’t account for them them the first time, but did with this final run.  i just totally missed that field the first time.

I counted about 11 million (rough calculation) for JD who is ranked 100 at 7.6 million using B-Refs salaries and games played (2008, 2005 and 2003 season only).  But not all games missed in these seasons are due to being on the DL which is likely the reason for the difference. With injuries in September for example teams don’t even bother with the DL.

Of more interest would be the dollars of lost value to a team due to players missed games.  An All Star not making much money in year 3 of his career for example, but costs his team more in value (runs/wins) if missing games due to injury or rest (rest to protect brittle players from injury).

Back to JD as an example, he has made 42 million over 3 years, and was paid 8.6 million in salary for missed games.

His value per Fan Graphs was 46.2 million over 3 years, so he cost the team 12 million in performance (wins or runs) by not playing 162 games. 

IIRC, DC mentioned that value for a given season should not translate into actual salary paid (market value) in response to Chone Figgins signing, so I don’t know if JD’s calculated value should traslate into what he is worth on the market in a 5 year deal.

In any event, not everyone plays 162 games so this could be corrected by using an expected number of games by a bonafide regular. Say 10% of games missed is the norm, so JD cost 5.3 million in salary for missed games over 3 years with the Red Sox and 7.2 million in lost value compared to players who play at an equal level and play 145 games per year.

For the 2002-2006 non-Red Sox years, it was 5.6 million in salary paid for missed games (using an average salary for the period to reduce calculation), and 13.6 million in lost value.

Over 2002-2009, JD received about 10.9 million in salary for games missed above average, and 20.8 million in lost value for games missed above average (as mentioned, not all games missed are due to injury, many missed games are to protect him from injury).

Not picking on JD here, just using him as an example of what would be a more useful stat to evaluate the cost of a players not playing as much as the average bonafide regular.