Theory of Poker and Theoretical Win Rate

[Megenoita]

In TOP, Sklansky says that a pro should be able to estimate what his hourly wage should be when he sits down at a table. He gives an example estimating how much of the money he will leave with. I was thinking that we should be able to do this with online poker as well. The standard that has been accepted at most levels is 2 BB/100, but I'm wondering where we get that figure and how we can quantify a WR for a given game.

Take 5/10 (6) Limit Hold Em for example. There are usually 2 big fish (44/3, 57/20), 1 moderately bad player (40/20), 1 overly aggressive player (30/23), and a tag (21/15). How much should I earn assuming I have position on one fish and only 1 agg player to my direct right? There should be a way to estimate...

Thoughts?

M

[eisanm]

I'm in no way good at this, and it might be unnecessary to even post this since I don't add much.

But my opinion is that you cannot be that precise. Even if you could determine a good approximation for your winrate, there's nothing saying you will actually get that winrate this specific time (variance, that is).

So I think it's pretty much meaningless trying to figure such things out.

I'd say that if you have a certain win rate after a large enough number of hands to determine your win rate, you can expect to earn that in a typical game.

For example, in typical games your win rate has been 2bb/100 over the last 200k hands. Then if this game seems typical you can expect to earn that much.

If this game looks better than a typical game you expect to earn more (this seems to apply to your example). Then maybe for this specific game you'll estimate 2,5 bb/100 or maybe 4bb/100 if it's a really good game.

I don't think you can be much more precise. And even using the kind of reasoning I do above, variance still means that I still don't expect to end up with the expected amount of money, even if I knew my EXACT win rate.

So my conclusion is there's no point in thinking like that.
Maybe Sklansky has a different opinion.

[AaronBrown]

I agree with eisanm. No one has a good enough model to answer this, there are too many different types of players. Plus the table make-up is important, it's not just the sum of the components. With a table of fish and one world champion, a good player might go broke even though he's above average for the table.

You'd need a large database of players playing against each other, then a lot of data analysis work to try to come up with a model that fits the data. Unless you work for an online Poker company, the only source I know is the IRC Poker Database. Before there was online Poker, people played on the Internet. The money wasn't real, but people took the games pretty seriously and the data are useful for a lot of theoretical work.