Question About Changes in Skill Over Large Samples???

[Christian_Peters]

For math inclined people;

Something new players really want to know is if they are improving and how much they are improving over their first 20,40, or 60 K hands.

I am curious if you could figure this out by taking that plain old win rate/100 graph over the sample, then getting a regression equation for the graph, then finding the derivative for the regression equation, and graphing the derivative function. It would seem to me that this would be a good graphical representation of how your skill level is changing over time (obviously this procedure assumes a few things, like the average skill of all your villains remain constant). But still, does anyone think this could be useful?

[kerowo]

I'm not sure what looking backward does for your game, but then again I'm not the mathiest mathy player.

[Xylocain]

This is very easy to do with Poker Office, I made that with my vpip and pfr in the aug results thread. There is however no infomration to extract from this [img]/images/graemlins/laugh.gif[/img]

[neurotiq]

this post made me dizzy [img]/images/graemlins/tongue.gif[/img]

[NIX]

I don't know if this would be possible or really useful.

I think there is too much variance in your winrate for it to converge to something useful in a sample size that small. For example, if you're playing in really loose games and go awhile without hitting your draws, then suddenly, you start hitting them, you didn't improve at all as a player, you're just running better. But that could really impact your winrate. Same things with downswings. If you're improving as a player and hit a bad downswing and compound your error by tilting off a ton of money, your general skill could've improved, but your winrate sure went the other way.

Edit: I actually just checked my winrate graph. It's been steadily decreasing for the last 50k hands. I'd hate to think that I'm getting increasing worse at poker in that timespan.

[Aaron W.]

[ QUOTE ]
For math inclined people;

Something new players really want to know is if they are improving and how much they are improving over their first 20,40, or 60 K hands.

I am curious if you could figure this out by taking that plain old win rate/100 graph over the sample, then getting a regression equation for the graph, then finding the derivative for the regression equation, and graphing the derivative function. It would seem to me that this would be a good graphical representation of how your skill level is changing over time (obviously this procedure assumes a few things, like the average skill of all your villains remain constant). But still, does anyone think this could be useful?

[/ QUOTE ]

BB/100 has too much variance to pull information out of that stat alone. In fact, as of right now, I'm pretty sure that nobody has a really useful quantitative measure for true skill at poker.

[Christian_Peters]

Yeah, I guess you would have to assume zero variance over the sample for the derivative to really mean anything.

ug. Thanks all for your comments.

[Xylocain]

Well, actually no, if you are happy with 80% confidence then this would be fine.

[bozlax]

[ QUOTE ]
I am curious if you could figure this out by taking that plain old win rate/100 graph over the sample, then getting a regression equation for the graph, then finding the derivative for the regression equation, and graphing the derivative function.

[/ QUOTE ]

Well, sure, you COULD do that...if you want to do it the EASY way.

[/confused dog]