Poker Tracker SD question

[jobro2]

If your standard deviations on poker tracker are listed:

SD/HR $7.28---14.46

SD/100 $8.09---16.1

Does that mean that for any given session that 68% of the time your SD/HR and your SD/100 will lie between the two numbers provided?

[TALLBrad]

The standard deviation gives you an idea of the variance of your winning rate in either BB/100 or BB/NR.

You can perform a confidence interval test that will give you a 95% ("Confidence") that you winrate resides within a range.

your winrate +/- ( 1.96 * SD) would be your limits.

For more info check out your stat book or:

Rice Virtual Lab in Statistics

[Catt]

[ QUOTE ]
your winrate +/- ( 1.96 * SD) would be your limits.

[/ QUOTE ]

Careful! It is actually WR +/- (1.96 * SE)

SE is SD / SQRT(# of hands / 100) . . . [assuming you're using the BB / 100 earn rate and not hourly].

In OP's example, if he played 10,000 hands for an actual WR of 1 BB / 100 (just keeping the numbers easy here), he could calculate his "true win rate" range with 95% confidence as follows:

Low end of range = 1 - 1.96*(16 / SQRT(10,000/100))
or 1 - 1.96*(16/10)
or 1 - 3.13
or (2.13) BB / 100

High end of range = 1 + 1.96*(16 / SQRT(10,000/100))
or 1 + 1.96*(16/10)
or 1 + 3.13
or 4.13 BB / 100

Meaning Hero could believe, with 95% confidence, that his true win rate falls between (2.13) and 4.13 BB per 100 hands.

[TALLBrad]

realized that after i posted....was going to fix after lunch.

Thanks,
Brad.

[jobro2]

is that -2.13 and +4.13?

[Catt]

[ QUOTE ]
is that -2.13 and +4.13?

[/ QUOTE ]

Yup.

(And I should have used 16.1 as the SD in the forumla instead of the more straightforward 16).

Here's a great thread on SD and what it means to understanding variance (with instructions on building a simple spreadsheet model to track your play).