|
|
#12
10-28-2007, 07:45 PM
|
|||
|
|||
Re: Variance question
< 10^-29
|
|
#13
10-28-2007, 09:32 PM
|
|||
|
|||
|
Re: Variance question
Under a Normal distribution, you will be more than 10 standard deviations from the mean with probability 1.5 x 10^-23. Since this isn't going to happen, if you think it did, either (a) it isn't a Normal distribution, (b) you computed the mean or standard deviation wrong, or (c) you're psychotic (the chances of being too insane to know you're insane is about 0.001, so any time you claim a probability is smaller than this, you should consider the chance that you're crazy).
No more than 1/100 of the observations can lie beyond 10 standard deviations from the mean, regardless of the distribution. Suppose you have a distribution with 0.99 chance of 0, 0.005 chance of -10 and 0.005 chance of +10. The mean is zero, and the standard deviation is 1. 0.01 of the observations are 10 standard deviations from the mean.
|
|
#14
10-29-2007, 12:48 AM
|
|||
|
|||
|
Re: Variance question
[ QUOTE ]
Under a Normal distribution, you will be more than 10 standard deviations from the mean with probability 1.5 x 10^-23. [/ QUOTE ] Where did you get this? =normdist(10,0,1,true) in Excel just gave me 1.000.. (a total of 30 0's). Is Excel unreliable for this many decimal places?
|
|
#15
11-01-2007, 02:59 PM
|
|||
|
|||
|
Re: Variance question
[ QUOTE ]
Where did you get this? =normdist(10,0,1,true) in Excel just gave me 1.000.. (a total of 30 0's). Is Excel unreliable for this many decimal places? [/ QUOTE ] If excel's normdist function can be used to determine the how likely you are to be within X std deviations, then shouldn't =normdist(1,0,1,true) give .68? It gives me .8413.
|
|
#16
11-01-2007, 04:45 PM
|
|||
|
|||
|
Re: Variance question
Answered the excel question for myself, after doing a bit more searching.
Excel's normdist function gives the portion that is less than the argument. So normdist(1,0,1,true) isn't the distribution from -1 to 1, it's the distrubution from -infinity to 1. Which would certainly explain the 7 decimal place difference.
|
|
#17
11-07-2007, 11:16 PM
|
|||
|
|||
|
Re: Variance question
[ QUOTE ]
Normal distibutions are quite common because of the Central Limit Theorem. Even if the results of your individual hands do not follow a normal distibution, the results of your 100 hand sessions will. [/ QUOTE ] How many of these 100 count hand sessions are needed before your results become normally distributed?
|
 |
|
|
Linear Mode
