How many BB/100Hands are considerd good?

[Sounded Simple]

Awesome pokerboy! Thanks.
I tried to work this out with an old textbook once and got seriously confused.

This should be stickied (or paste it into a new post and sticky).

[holdem2000]

[ QUOTE ]
[ QUOTE ]
FWIW currently have a standard deviation of 1.14 ptBB/100 on that winrate

[/ QUOTE ]

[ QUOTE ]
the 1.14 number I gave you is the standard deviation on my winrate for the 117K hand sample.

[/ QUOTE ]

So 1.14ptBB/117k

By the way, did you convert that the right way? You know you have to convert from SD to variance, multiply, and then convert back?

Also, most statistics problems are written within 2 standard deviations (95%) as opposed to one standard deviation (68%).

[/ QUOTE ]

Edit: pokerboy did a lot of this in a different post while I was typing this, I'll leave this up though.

So when I'm saying a standard deviation of 1.14 ptBB/100 over 117K hands the ptBB/100 is still the units I'm measuring winrate & SD in even if my sample is larger... reporting it as something like total profit with the SD of the profit in dollars is probably clearer... maybe SD = 1.14 ptBB/100 (117K hands) is a decent way to write it.

Yea, I converted it right... could find it by taking my SD for a 100 hand sample, squaring it to get the variance, multiplying by 1170 to get the variance of a 117K hand sample, square-rooting to get the SD, but so far we have the SD measured in just big bets instead of BB/100, so we divide by 1170:
SD = sqrt(38.84^2 * 1170) / 1170 = 1.14... or once you've done it enough times you remember that to convert a SD from a 100 hand sample to a 117K hand sample you just divide by sqrt(1170):
SD = 38.84 / sqrt(1170) = 1.14

The +/- 1 SD notation is used in physics and a couple other fields for reporting the standard deviation and then something like a 95% confidence interval is often provided separately. Not sure why +/- 1 SD is ever used, it is pretty misleading, but some fields use it as convention.

OP, sorry for the bit of a hijack here.

[1p0kerboy]

holdem2000

Sounds like you know your stuff. I was just checking because it's common for players to convert it wrong. I wasn't trying to insult your intelligence or anything!

[holdem2000]

[ QUOTE ]
Awesome pokerboy! Thanks.
I tried to work this out with an old textbook once and got seriously confused.

This should be stickied (or paste it into a new post and sticky).

[/ QUOTE ]

If you want something like this stickied it's probably good if we also include a disclaimer that all this supposes that your win/loss on any given hand (or at least every sample of 100 hands) is normally distributed... while this certainly isn't correct most of the things we do like finding a 95% confidence interval are still reasonably accurate. We should also include how to use a z-table to do things like find a confidence interval for any level of confidence you like, and do things like finding the probability that your 'true' winrate is bigger/smaller than some given value. If there's demand for it I'm sure pokerboy or I can write something like this up sometime this weekend.

[Sounded Simple]

Yeah I think that would be useful.
The formulas I was lookig at ("Mathematics of Poker" and an old school book) confused me a lot because of the terminology but I can understand you guys posts much easier.

BTW - Whats a Z table?

[holdem2000]

[ QUOTE ]
holdem2000

Sounds like you know your stuff. I was just checking because it's common for players to convert it wrong. I wasn't trying to insult your intelligence or anything!

[/ QUOTE ]

Thanks, same to you [img]/images/graemlins/smile.gif[/img] I know what you mean about players often converting it wrong (or even more often making pretty inaccurate guesses about variance from not knowing how to do this stuff).

[holdem2000]

[ QUOTE ]
BTW - Whats a Z table?

[/ QUOTE ]

At a glance I think wikipedia has a decent explanation (with a link to a Z table at the bottom) http://en.wikipedia.org/wiki/Z_table , I've got to run now but I'll explain it in a post sometime this weekend with a couple examples of calculations using PT stats.

[Sounded Simple]

Sorry that was a stupid question.
Sometimes I is lazy with the interwebs.

[inverted]

time for a hodem2000 howto? or holdem2000 well :P

[GiantBuddha]

-2PTBB/100 is average, FWIW