Two Plus Two Older Archives  

Go Back   Two Plus Two Older Archives > General Gambling > Probability
FAQ Community Calendar Today's Posts Search

Reply
 
Thread Tools Display Modes
  #1  
Old 03-14-2005, 06:29 PM
CallMeIshmael CallMeIshmael is offline
Senior Member
 
Join Date: Dec 2004
Location: RIP Mitch Hedberg
Posts: 1,097
Default Confidence interval question

I'm looking to compute a confidence interval for the number of hands a player plays and rasies with preflop, after X number of hands.

Does anyone know a stardard deviation for this data?

Thanks
Reply With Quote
  #2  
Old 03-14-2005, 06:50 PM
gaming_mouse gaming_mouse is offline
Senior Member
 
Join Date: Oct 2004
Location: my hero is sfer
Posts: 2,480
Default Re: Confidence interval question

If there true preflop raising percentage is p, then the sample percentage after N hands has a binomial distribution with parameters with mean p and standard deviation sqrt(p*(1-p)/N).

You can now apply the normal approximation to the binomial (ie, assume a normal distribution with the same mean and SD that I calculated above) to obtain confidence intervals.

For example, if you observed a PFR=.1 after 50 trials, a 95% CI for the true PFR is about [.015,.185], which is a fairly large interval.

I think in general people give too much credence to stats after for small sample size.
Reply With Quote
  #3  
Old 03-14-2005, 07:21 PM
jason1990 jason1990 is offline
Senior Member
 
Join Date: Sep 2004
Posts: 205
Default Re: Confidence interval question

[ QUOTE ]
If there true preflop raising percentage is p, then the sample percentage after N hands has a binomial distribution with parameters with mean p and standard deviation sqrt(p*(1-p)/N).

You can now apply the normal approximation to the binomial (ie, assume a normal distribution with the same mean and SD that I calculated above) to obtain confidence intervals.

For example, if you observed a PFR=.1 after 50 trials, a 95% CI for the true PFR is about [.015,.185], which is a fairly large interval.

I think in general people give too much credence to stats after for small sample size.

[/ QUOTE ]

If we do this with X=0.1 and N=50, then (approximately)

P(-2 < |X-p|/sqrt{p*(1-p)/N} < 2) = 0.95.

There is a significant amount of algebra needed to convert this to a statement of the form

P(A < p < B) = 0.95.

I think what you have done is to use

P(-2 < |X-p|/sqrt{X*(1-X)/N} < 2) = 0.95.

While this may be close enough in many cases, the nitpicky part of me wanted to point this out. [img]/images/graemlins/wink.gif[/img]
Reply With Quote
  #4  
Old 03-14-2005, 07:42 PM
gaming_mouse gaming_mouse is offline
Senior Member
 
Join Date: Oct 2004
Location: my hero is sfer
Posts: 2,480
Default Re: Confidence interval question

eh. while you are technically correct, what i've done is completely standard practice and is what most people mean when they say 95% CI. so there. [img]/images/graemlins/tongue.gif[/img]
Reply With Quote
Reply


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -4. The time now is 09:18 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.