Standard Deviation

[Pvnuts]

Hey every one... I have a few questions about StdDev as it is recorded in my Poker Office software.

1. My SD is 25 and my hr rate is ~$7 at the $1/$2 limit games. The bankroll I have in the site I play at is ~$350. Can someone interpret what this all means? Does it mean that my flux is +-$25 of my bankroll? or my hr rate?

2. Also, I punch in the above numbers at this site that calculates the Risk of Ruin and the results are less than 1%. Does this mean that assuming that I continue to play basically the same way at the same level and don't tilt any more than I have while accumulating my stats, that my risk of ruin is less than 1%?

3. One more question... Does anyone know of a site I could plug in all my states and a report would come back that would suggest ways to fix my leaks?
I came accross a site that did it for free for a year, but I lost it... [img]/images/graemlins/frown.gif[/img]

Thanks

Al

[rubberloon]

Unless you know number of hands played cannot tell. Take number of hands divide by 100 take square root and divide into standard deviation to get standard error then divide standard error into your win rate. If greater than two you're doing OK.

[uDevil]

[ QUOTE ]

1. My SD is 25 and my hr rate is ~$7 at the $1/$2 limit games. The bankroll I have in the site I play at is ~$350. Can someone interpret what this all means? Does it mean that my flux is +-$25 of my bankroll? or my hr rate?


[/ QUOTE ]

I'm not familiar with PO. Does it give units for those numbers? Is it $7/hr winrate with a SD of $25/hr? Given those numbers, I calculate a RoR much less than 1%.

What does it mean? Here's one way to look at it: Suppose you play 100 hours of poker. For 68 of those hours, you'd expect to win $7 +/- $25, i.e. between -$18 and $32. That's a range of one standard deviation about the mean. For 95 of those hours, you would expect the result to be that you win between $7 +/- ($25)*2. I.E. between -$43 and $57. That's a range of 2 SDs about the mean.

[ QUOTE ]

2. Also, I punch in the above numbers at this site that calculates the Risk of Ruin and the results are less than 1%. Does this mean that assuming that I continue to play basically the same way at the same level and don't tilt any more than I have while accumulating my stats, that my risk of ruin is less than 1%?

[/ QUOTE ]

Yes, if the numbers are accurate. Your confidence increases as you play more hours.

[ QUOTE ]

3. One more question... Does anyone know of a site I could plug in all my states and a report would come back that would suggest ways to fix my leaks?

[/ QUOTE ]

I don't. You can post your numbers here and people will have a go at it. In my opinion, this is very superficial and only marginally helpful though.

[Pvnuts]

Rubberloon... Thanks for your reply...

I did the math you suggested and the number I come up with is 1.2. What does that number represent.

The deviation of 25.95 is based on 2688 hands.

[Pvnuts]

Hi uDevil,... thanks for taking time to explain the deviation issue, it helped alot...

The SD is hourly... However I've only played 2688 hands... Does this mean my SD is not accurate at this point? I've read onlie that you have to have 100000 hands under your belt to have an accurate SD...

Also, is 25 representing that I'm too volatile?

One more question... When should I consider moving up to the $2/$4 game. I didn't use the 300BB rule for the $1/$2, should I consider waiting for my bankroll to go up to $1200 before making the leap?

Thanks

Al

[Runthemill]

PT is your friend.

[uDevil]

[ QUOTE ]

The SD is hourly... However I've only played 2688 hands... Does this mean my SD is not accurate at this point? I've read onlie that you have to have 100000 hands under your belt to have an accurate SD...

[/ QUOTE ]

I don't know how to quantify it, but as I understand it, SD converges quite quickly, much more quickly than winrate. It is true that you haven't played many hands (to be consistent with your other numbers, consider hours, not hands), so you can't have much confidence in your winrate, but the number 100,000 is pretty arbitrary. There is no magic number at which you suddenly have 100% confidence in the numbers. Even if you currently cannot have high confidence, the numbers you have are the best estimates available.

[ QUOTE ]
Also, is 25 representing that I'm too volatile?

[/ QUOTE ]
Well, that's 12.5 big bets/hour. That might be on the low side, actually. More important is the ratio winrate/SD^2. Yours is high, which is why you have a low risk of ruin even with a relatively small bankroll of only 175 big bets. By the way, I miscalculated RoR in my previous post-- I think your RoR is actually ~2%. But don't rely too much on that. Because you have played so few hours, these numbers could change quite a bit.

[ QUOTE ]
One more question... When should I consider moving up to the $2/$4 game. I didn't use the 300BB rule for the $1/$2, should I consider waiting for my bankroll to go up to $1200 before making the leap?

[/ QUOTE ]

I would play more hands at your current level before worrying about this. Maybe you've just been running good. Sometimes the worst thing that can happen to a new player is to have good results at the start. I can't really give you a number, but $350 is only ~88 big bets at $2/4, and you can expect your winrate to drop when you move up.

[pzhon]

What rubberloon was trying to calculate was the number of standard deviations you are away from breaking even, to try to establish whether you are a winning player or not. If you are ahead of win rate x by 2 standard deviations or more, then you have strong evidence that you are winning x or more.

He assumed that the SD was per 100 hands. If it was hourly, then you need another calculation: observed win rate per period / (SD per period / sqrt(# periods)). Although the SD per period is just and estimate, it converges much faster than your win rate does, particularly in limit.

Anyway, a critical assumption of the ROR calculation was that the win rate you entered was your true average. After 2700 hands, you don't have a very good idea what your average win rate is, and in fact, it's not clear that your win rate is sustainable by anyone. It looks like you were running well, although whehter by a little or a lot will take a lot more hands to determine.

Another assumption in the ROR calculation is that you never cash out. If you plan to cash out periodically, you need to reduce your win rate by the amount you are cashing out.

[Pvnuts]

Hey everyone, thanks again for your replies, I'm trying hard to absorb it all... [img]/images/graemlins/smile.gif[/img]

When you refer to winrate, you do mean the hourly rate right?
Also Pzhon you were refering to winrate and SD per period... is a period a playing session? an hour?

Lastly, is there a book I could buy that explains all this stuff or do I just have to keep reading posts and buggin everyone untill I get a grasp of it all... lol

Thanks again

[pzhon]

[ QUOTE ]
Hey everyone, thanks again for your replies, I'm trying hard to absorb it all... [img]/images/graemlins/smile.gif[/img]

[/ QUOTE ]
You're welcome. You have come to the right place.

[ QUOTE ]
When you refer to winrate, you do mean the hourly rate right?

[/ QUOTE ]
You don't play a fixed number of hands per hour. Most people express win rates in BB/100 for online play, and BB/hour for live play when there is no count. The number of hands per hour is not 100 online. When I play NL 6-max, I expect to play about 80 hands per hour per table.

[ QUOTE ]
Also Pzhon you were refering to winrate and SD per period... is a period a playing session? an hour?

[/ QUOTE ]
You can use period to mean an hour or 100 hands. The calculation should be the same for either. If you use it to mean a session, this will introduce other factors, such as whether you vary the session lengths.