good prob or still to small ??

[Lary18thS]

What u thinking about this prob - i mean almost 3k(45man) games , it's good or too small to get far going conclusions ...??

[Chancard]

This should probably in the brags section.

BTW when i grow up i wanna be just like you, Lary.

[Lary18thS]

trolls not welcome imo ..........

[TYggShipIt]

Coaching lessons how much?

Great job Karol.. But seriously... lets talk.

[rryu87]

heh sdfdfdfdffsdfsdfdsfsdfdsf

[Lary18thS]

...... dosn't mether ... thank's for conscrutcive anserwers ....

[TYggShipIt]

Karol your a killer, if you can maintain that for 3k games i dont see it fluctuating more then 5% lower if you keep it up and dont change anything, very impressive and well done. Wish i had your skills insight and help. I'll be looking for you signing up for the 45s and stay out of yours.

Well Done sir.

[Peke]

Karol

see this post by a statistician (that's what he says at least )
How to calculate confidence intervals on your ROI

By his own account for 45 man goes something like:

- after 4500 games your error would be +-10%
- after 9000 games your error would be +-7%
- after 18000 games it would be down to +-5%

That's considering the same buyin level and same playing factors (ie you're always 16 table & such)

[FrankJones]

There are two possibilities: You will either continue to post a 31% ROI and go on to be regarded as one of the best to ever play these games; or, not. Although that is an obvious and seemingly pointless statement, it still has some value.

The confidence intervals only tell you so much. Even if we say that there is a 95% likelihood that your next 2000 games will have an ROI between 24 and 36 %, what does that really tell you? Not much. As poker players, we defy 95% probabilities on a daily basis.

What matters is that you keep playing solid poker, because the rest is out of your control. I am not sure how to construct confidence intervals on non-normally distributed results that are skill based and not random. That seems very difficult, and was not covered in my three years as a math major in an actuarial track. But as you can see from that link that was just provided, the volatile nature of the game makes predictions very difficult.

In my first 2730 games I had an ROI of 22.6 %
In the 1560 games since, my ROI is 8.5 %

So, you just never know.

[FrankJones]

Karol

see this post by a statistician (that's what he says at least )
How to calculate confidence intervals on your ROI

By his own account for 45 man goes something like:

- after 4500 games your error would be +-10%
- after 9000 games your error would be +-7%
- after 18000 games it would be down to +-5%

That's considering the same buyin level and same playing factors (ie you're always 16 table & such)

I have a question. When you say the error is +/- 10%, do you mean to say that if a person has a 20% Roi after 4500 games, it means his range is 18% to 22%?

That is how I would interpret it, though without clarification, someone might misinterpret that as meaning that the range would be 10% to 30%.

I have seen this confusion happen in cases where someone tries to refer to a percent change in a number that is already a percent to begin with.

By that, I mean if something changes from 20% to 15%, that is a 25% decrease, though I have seen people mistakenly refer to it as a 5% decrease.


MASSIVE EDIT:

There was some confusion, my fault mostly, which resulted in some pointless debate. I have removed my non-constructive posts related to the debate, so I apologize if some posts seem to be random or out of context.

The author of the sharkscope was in fact dealing with absolute error, and not error rate, which should clear any confusion that there may have been.

I apologize to all for the silly debate. It turns out we were discussing apples to oranges and I was just being stubborn.

On a positive note, we branched off into an interesting discussion about the possibility of trying to prove mathematically that poker is a game of skill.