What do I mean by Redd's replacement value?

I got a couple of emails asking what I meant when I wrote yesterday that Andrew Bogut was more valuable than Michael Redd because he had a higher "replacement value" than Redd did. Let me expound.

The concept of "replacement value" is the creation (as far as I know) of the writers at Baseballprospectus.com. They created it to determine the relative value of each baseball player at each position by comparing each player to a "readily available player" at the same position who could be acquired "at little or no cost" to the player's team.

A couple of weeks ago I decided to apply the same concept to value professional basketball players. So using my own "win profile" statistics, I established"replacement costs" for each of the 5 traditional basketball positions. Here, in brief, is how I did it.

Basically, I followed the cookbook recipe set out by baseballprospectus with a few adjustments. Since baseball and basketball differ in their basic substitution patterns, I could not simply take a survey of all basketball reserves as the baseball writers did when they determined the replacement value of baseball's position players. I had to take an approach closer to the approach the baseball writers used to value bullpen relievers, meaning I had to gauge "readily available" according to some minimal use standard.

So I arbitrarily picked 14 minutes of playing time per game at the high end of my definition of a "readily available" player, and I required the player to have played at least one percent of his team's total player minutes as my low end. My rationale was that if a player is used less than 14 minutes a game he's probably not in his team's "regular rotation" and therefore he can be acquired by any other interested team at, probably, little cost to that team (with exceptions, obviously, for special cases such as "projects" and others), and I wanted him to have played at least 1% of his team's overall minutes to avoid sample size distortions (Joe Useless comes in for 3 minutes at garbage time, plays like Wilt Chamberlain and is never heard from again).

Finally, I only wanted to measure the production of players who would be considered suitable by most NBA team's to man the given position. So I required that each played I measured played at least 30% of his minutes at the given position.

With all that done, I came up with the following replacement costs for each position (expressed in terms of "Marginal Win Score per 48 minutes"):

PG: -1.4

SG: -0.9

SF: -1.0

PF: -1.6

C: -2.6

Those numbers are about as you would figure, given the physical and skill demands of each of the respective positions. After all, centers and power forwards have certain physical requirements that necessarily limit their pools of talent, whereas point guard has certain skill requirements that limit its.

And as you can see, the two main positions manned by Michael Redd (SG/SF) can be replaced at the lowest cost to the Milwaukee Bucks. Therefore, there is a higher burden upon him than there is upon, say, Andrew Bogut, to deliver above a certain premium in order to justify his labor cost.

Fantasy owners understand this concept well. If you can replace a certain position easily, you will be less willing to expend resources on that position. Instead you will marshal your resources and spend them where the talent is most scarce.

So even though Michael Redd generally delivers a slightly better Win Contribution each season than Andrew Bogut, Bogut is still the far more valuable commodity, because the difference between Bogut and "some random center" is far greater than the difference between Redd and "some random swingman".

This is why I would argue that Redd ought to be shopped. Not only is he not worth what we are paying for him in absolute terms (the market value of a win in the NBA last season was about 1.6 million. Thus Redd owed the Bucks 9.1 wins, but he delivered only 5.5 -- if you except my numbers), he is not worth what we are paying him in relative terms... we could replace him with Mr. Scrub and not be that much worse for wear (indeed, the team's trajectory was not that much worse with below replacement level Charlie Bell than it has been with Redd).