Thursday, March 26, 2009

The Winner's Curse

mulkowsky wonders if a common auction phenomenon also applies to Rotisserie League Baseball.

one thing I struggle with when I try to postmortem my draft performance is the concept of "the winner's curse" -- the idea that when everyone has a different estimated value for an object, the person to win that at auction is likely to be the person who is overvaluing it the most. In terms of Roto, my drafts always look good in retrospect viz a viz my estimated prices, because I'm using them to guide my draft, but I've got no idea whether they are really better than others' estimated prices.
I've provided the link to the Wikipedia definition of The Winner's Curse above. Part of the definition is:

The winner may overpay or be 'cursed' in one of two ways: 1) the winning bid exceeds the value of the auctioned asset such that the winner is worse off in absolute terms; or 2) the value of the asset is less than the bidder anticipated, so the bidder may still have a net gain but will be worse off than anticipated.
I'm skeptical that The Winner's Curse applies to Rotisserie League Baseball. Most of the examples I found cited on Wikipedia and elsewhere discuss an auction where the currency available is theoretically unlimited. The example I liked best Initial Public Offerings (IPO). Bidders are purely guessing what a stock will be worth. While it is next to impossible for the cost of a newly available stock to reach an extremely high dollar amount, that doesn't mean it won't based on bidder perception.

Where the Winner's Curse could apply to Rotisserie is that we don't know how each individual player will perform. However, there are a lot of data available to us: prior statistical knowledge, injury history, role and how that will translate to playing time (and accumulation of statistics), and possible position depth or scarcity.

What we do know is that each league will produce $3120 worth of stats for 12-team A.L. only leagues and $3380 worth of stats for 13-team N.L. only leagues that use standard $260 budgets per team. This isn't esoteric or mysterious, but rather a fact that is as old as the game itself. Free agents we acquire during the season will produce a greater or lesser amount of statistics from season to season, but the amount of value that we can acquire during the auction is finite.

While there are obviously variations from player to player and from season to season, I don't believe that these fluctuations adequately fit the definition of The Winner's Curse. If a Rotisserie owner could spend an unlimited amount of money out of his own pocket to put together a $700 team while the rest of the owners in the league could only afford a $260 team, I believe that this would apty fit the definition of the Curse. However, the closed auction format means that you're not seeing a Winner's Curse but rather variance on player to player from year to year that generally falls within an expected range based on prior performance and all of the other factors cited above.

1 comment:

Rodger A. Payne said...

In my AL auction, I believe the "winner's curse" applies to young star-level prospects and closers. These players often have very high upsides, but also disappointing potential downsides that can be triggered by poor starts (in small samples of performance), injuries or other vagaries of the game. In short, there's a lot of uncertainty, which is the prime cause of the "winner's curse."

In the case of prospects, everyone knows that many players grow (as Sandler would put it) through their early 20s, but nobody knows how much any given player will grow. In recent seasons, my league has greatly overbid for guys like Alex Gordon and Chris Davis. It's a keeper league and nobody wants to let another owner have a long-term bargain. So the auction winner often ends up cursed.

Many closers are also doomed by the problem of uncertain information. As we see every season, it doesn't take that much for many closers to lose their jobs. Even the handful of reliable closers usually yields at least one turkey by the end of the season.

Incidentally, you are absolutely right about total league value. The over-valuing of certain players is directly linked to under-valuing of other players. It's zero sum. Often, the most undervalued guys are former highly touted prospects who are now perceived as busts. After a year or two (or three) of mediocre performance, young prospects lose their luster. Michael Cuddyer was like this a couple of years ago.

With high levels of information, the winner's curse is less likely. Thus, proven star players are often priced the same by most bidders. Ironically, at my auction, the scarcity of roster space (saved for desired young prospects) creates an early window of buying opportunity when the best players in the game are sold in the first two rounds. Usually the bargain is only a couple of bucks, but I'd much rather buy a star for a couple bucks under price than pay top dollar for the most prized rookies.