Tuesday, May 29, 2007

Traveler's Dilemma

Read this:

http://www.sciam.com/article.cfm?articleId=7750A576-E7F2-99DF-3824E0B1C2540D47

Now think about no limit poker. IF this article is correct, players will always bet the wrong amount at all but the higest level of competition. And even then they might.

What does this mean?

I means that betting the "correct" amount can be enough to outplay our field. Wow. I was hard on myself last night in the Hoy because I bet POT everytime I hit top pair in an effort to avoid variance and players catching up. But do I need to except the fact that I did not maximize my return?

2 comments:

Michael Albert said...

Nice link, unfortunately broken (but the title let me find it). I think this one works.

As a person with some credentials in the area I'd caution anyone about drawing conclusions about the play of real games from these abstract models.

For instance in the traveler's dilemma, a person whom the man in the street (as opposed to a game theorist) might regard as rational will simply write down the actual value. This is non-optimal in the game sense, since it risks being punished by a low-balling opponent, but few of us would be willing to accept our guaranteed $2 payout from optimal play, when the actual value was something like $80.

The more interesting form of the dilemma is when you play it as a freeroll -- now the theory does make some sense. But, even here, the analysis treats the game as an isolated event, and optimal play in that context does not necessarily match optimal play globally (if we know that a whole sequence of such freerolls will be played with the results publicised, it's not hard to guess that most people will converge on the $100 strategy).

Oh, too much to say ... stay tuned for a blog post.

Anonymous said...

GREAT BLOG