PDA

View Full Version : Sample Size question - SD

Cerril
10-22-2004, 03:04 PM
So after doing a search I'm still confused on this one... I mean, really confused. What exactly is the meaning of Standard Deviation in my PT stats? I feel like I should be able to grasp it pretty intuitively but it completely escapes me.

Is this a number that should be lower (or at least lower than a certain point), or is it a measure of how insufficient my sample is, or both? What's the correlation between it, my total hands played, and my winrate in BB/100?

dogmeat
10-22-2004, 11:22 PM
The standard deviation is what you can expect your numbers to be with regard to your win rate. If your win rate is 1 big bet per hour, and your standard deviation for 1 hour is 10 big bets, then after an hour of play, your win/loss will be anywhere from -9 bb to +11 bb.

If you want a true estimate of your hourly rate, you need to understand that this number will fluctuate greatly over the course of a year. If your SD is 10BB and you play 400 hours, then your standard error will be very small and you can expect your numbers to be accurate.

If you want to calculate your SE:

SD/sqrt(N) where you plug-in your standard deviation for an hour, and N is the number of hours played. Assuming 1 BB/HR

SD=10 bb, you play 400 hours, then your SE is 10/sqrt(400) = 0.5 bb This allows a confidence of 90% that your win will be accurate to within 1.6*0.5 = 0.8 big bets. Or your win will be 1 bb +/- .8 bb Poker is all about the long run, and the long run is a long, long time.

Thanks to BruceZ, I can also figure my risk of ruin for my bankroll:

BR = -(SD^2/(2*WR)}*LN(ROR)

FWIW, most of the SD's I have seen posted are between 10 and 20BB. Mine happens to be 21bb - but this includes all my games, from .50/\$1 NL to \$5/\$10 limit.