FAAB, Vickrey and the cost of automation

Anyone who has been reading me for more than a decade probably knows that I hate the FAAB process for in-season free agent acquisition. In fact, my first “Why I Hate FAAB” column ran in 2005.

In short, I can’t come to terms with a blind bidding process that wastes valuable resources, yet we spend so much effort obsessing over projective precision on Draft Day. Who cares if I overspend by $100 for the Flavor of the Week but it’s not okay to draft a 10th round ADP pick in the 7th round? Really?

I wrote about it most recently here and here. It’s fixable; nobody cares.

The Vickrey Method, which is described in one of those links, is the closest to an agreeable solution for many leagues. Vickrey awards the winning bidder with his player, but at the cost of the second highest bid, plus $1. It’s based on sound economic theory. Tout Wars used Vickrey for about a decade, then abandoned it two years ago.

Part of the reason was some owner pushback. There was a groundswell of discontent over gross overbids that were grossly reduced. If you bid $200 for a player and the next highest bid was $12, you bought him for $13. That seems inequitable at its face – why reward an owner for misreading the market? – but how can you really read the market in a blind bidding process?

Several weeks back in Tout Wars, I bid $213 for Franmil Reyes (pictured). Needing an offensive boost, I calculated that amount by looking at what he went for in other leagues, but in Tout, I was the only bidder. Vickrey would have awarded me Reyes for $1. That seems wrong (was everyone else sleeping?), but so does having to eat so much of a scarce commodity like FAAB. There are workarounds to mitigate situations like this, but you can’t satisfy everyone.

The Tout Board of Directors was mostly in support of Vickrey. Tout discontinued FAAB for a different reason.

Back in the days before full automation of the FAAB process, league SWATs were often faced with handling email requests like this:

CLAIM Chad Curtis $43
If I don’t get Curtis, then claim Greg Vaughn $43.
If I DO get Curtis, then claim Greg Vaughn $21.
If I don’t get Curtis OR Vaughn, then claim Carlos Baerga $21 or Scott Rolen $13 (but ONLY if I get neither Curtis nor Vaughn).
If I DO get either Curtis or Vaughn, but not both, then claim Carlos Baerga $12 or Scott Rolen $8 (but ONLY if I get either Curtis or Vaughn).
And finally… Claim Trey Beamon $4 and Derrick May $2.

To make space for these claims, in order…
Reserve Fonville, Cut Belk
DL Carlos Garcia
Cut Dennis McCarty
Cut Manny Martinez

Good times.

This was a real FAAB claim from LABR in 1997. Let’s call this owner Ron (but it wasn’t me). Thankfully, today’s automated systems allow for a more organized structure to bidding, but the extent to which owners still try to make nested conditional bids often requires multi-page decision trees.

The above bidding could be summarized as:

Claim
Chad Curtis $43
Greg Vaughn $43
Carlos Baerga $21
Scott Rolen $13
Trey Beamon $4
Derrick May $2

Claim
Greg Vaughn $33
Carlos Baerga $12
Scott Rolen $8
Trey Beamon $4
Derrick May $2

I think.

Now add 11 other owners, all placing similarly convoluted bids, and Vickrey.

Let’s simplify this just a bit.

Suppose I… um, Ron has a starting FAAB balance of $100. He bids his entire bankbook on Chad Curtis and a similar $100 bid on Greg Vaughn. He hopes to get at least one, but would be thrilled to have Vickrey give him a shot at both.

Let’s say that the second highest bid for Chad Curtis was $50. Ron would win Curtis and have his winning bid reduced to $51 by Vickrey, leaving him with a FAAB balance of $49. But the online engine can’t do this adjustment on the fly without considering all the other bids for all other players. So Ron is initially (though temporarily) charged his full $100 bid, effectively zeroing out his budget.

Concurrently, another owner wins Greg Vaughn with a $35 bid. Had Ron been credited correctly, he would have had $49 left to spend for Vaughn and should have been the rightful owner. But his balance is $0 when Vaughn is awarded.

Per Tout board member, Peter Kreutzer: “The problem with fixing these issues by hand is that they’re not easily spotted by the SWAT. And the fixes can get quite complicated, affecting multiple teams.”

So we canned Vickrey. Might be the only time a Nobel Prize winner was so brazenly dissed.

With every effort to simplify processes, often there are sacrifices elsewhere. I would imagine some other engine might have been able to figure this out — maybe multiple runs, I don’t know — but you can see how this could get complicated pretty quickly.

In lieu of Vickrey, Tout Wars increased each owner’s FAAB budget from $100 to $1000. That gave the perception of a deeper pocketbook and softened the pain of overbids. It’s all in our minds. I sure feel richer being able to toss out a bunch of $8 or $14 token offerings that barely make a dent in the budget.

But losing $212 of $1000, or $20 of $100, still stings.