Small +EV edges: Simulation vs. Reality

[gulon]

Playing with Prego and SNGPT today and thinking about how it relates to real game play. It's nice how the simulations put villains on exact ranges but in a real game this is an estimate at best.

Although mathematically pushing here is +EV it is < 0.50%. In a real game, obviously I can't guarantee villains will play an exact 11% and 25% range like a simulation will do nicely for you. Does the margin for error of putting someone on a range make this a fold instead of a push at gametime? Do you err on the side of pushing or on the side of folding?

75/150
Hero (t2282)
SB (t2313) - 25% (22+, A2+, KTo+, K8s+, QTs+)
BB (t3139) - 11% (44+, A9o+, A8s+, KQs)
UTG (t3386)
CO (t2380)

Preflop: Hero is Button with 66o
2 folds, Hero... ?

EV Diff: 0.17%

[DevinLake]

Prodigy made a really nice post on this topic. Search his posts.

[gulon]

[ QUOTE ]
Prodigy made a really nice post on this topic. Search his posts.

[/ QUOTE ]
I searched all forums in the last 1 year for user "Prodigy" and got 11 posts and none of them seem to be about poker strategy. Are you sure it's "Prodigy"?

[DevinLake]

Prodigy54321

[gulon]

Okay I read the thread: http://forumserver.twoplustwo.com/showfl...rue#Post4870127

It's a very interesting thread, but it focuses on the question of whether you should make a push given an exact EV value. My question, I think, is a little different:

My premise is that you can't really put a villain on a "perfect range" the way a quiz-program does like Prego. There is a gap between the quiz-program and reality. Given the margin of error for putting a player on a range, can you really trust that a small EV number is "in reality" +EV?

I would think that a range is more like a Gaussian distribution, with a standard deviation that accounts for error about a mean. If your +EV edge is smaller than your standard deviation, is it even +EV in the first place?

[crzylgs]

[ QUOTE ]
If your +EV edge is smaller than your standard deviation, is it even +EV in the first place?

[/ QUOTE ]

Yes. If the mean EV of the distro is positive, this would be a push. The chance that it's -EV is balanced by the chance that it's more +EV than you're expecting.

[NHFunkii]

I don't understand how you know what a gaussian distribution is, but you don't understand that it's the mean that counts...

[DevinLake]

gulon, keep searching his threads. That's not the one I was thinking of. There was one that addresses your question exactly I believe.

[microbet]

Do the math and figure out how big a mistake it is if you are off on your range in either direction. Usually it is not symmetric.