100k hands.... open for inspection/selection/detection/interjection

[Nick Royale]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]

If I'm at a table where I get significantly ahead, I'll often quit and move on. Stupid, I know, but that urge to "lock in the win" is at times overwhelming.

[/ QUOTE ]

Dude, it's extraordinarily stupid if the fish is still there, as you know. But usually I bust the fish, and if he doesn't reload, then the table sux, so close it out and move on.

[/ QUOTE ]

I think he means over a whole session, not just on a single table of a multitabling session.

[/ QUOTE ]
I actually think 1 table = 1 session.

[Jay Riall]

[ QUOTE ]
Wow. Maybe we shouldn't have been looking at your STATS for the problem.

[/ QUOTE ]

ZZZZZZZZZ at looking primarily at stats to find leaks anyway. Stats would be one of the least important things to look at imo unless there are obvious glaring holes in them.

[SirFelixCat]

I'd love to hear from Pokey on this...esp his formula to calculate SD...this has ALWAYS befuddled me.


eta: Wow, ok...so can you (Pokey) or anyone else explain how to calculate SD like I'm 5? [img]/images/graemlins/grin.gif[/img]

[FreakDaddy]

Hire a coach... like me... [img]/images/graemlins/smile.gif[/img]

Really good coaches will know how to understand what kind of glasses you're seeing poker through, and can highlight your leaks/weakness, and open you up to new ways of thinking about the game of poker. In a nutshell, that's a major reason why people aren't as good as they could be, and they will never be able to really break out of that until they talk about their thought process and work with someone who thinks and sees the game from every angle.

This is the simple and most direct point about being exceptionally good at poker. The really good players are not just linear thinkers, they are artists by nature, and know how to use multiple ways of approaching a problem.

Every decision you have at the poker table is a problem, so rationally it's in your best interest to be able approach the problem from multiple perspectives in order to come to the best conclusion. Everyone has the ability to do this, but, in my experience, very few people can. So the are contionually limiting their potential with each decision (and not just at poker) they make.

Finding a good coach (or teacher), in poker, or just about anything is more difficult than it sounds to be honest. Most teachers have no developed plan, they don't know how to present questions to the student so thay they can come to their own understanding, and most don't seem to really understand the nature of a problem. That's just my experience in working with multiple music teachers, earning my MA, and working with students.

Good luck...

[Pokey]

Nick Royale said:

[ QUOTE ]

I'm not really sure I got how PT calculates SD though.


[/ QUOTE ]

Say you have five sessions:

Session #1: 100 hands at a winrate of 20 PTBB/100.
Session #2: 200 hands at a winrate of -20 PTBB/100
Session #3: 300 hands at a winrate of +10 PTBB/100
Session #4: 400 hands at a winrate of 0 PTBB/100
Session #5: 500 hands at a winrate of 1 PTBB/100

Your average winrate is then:

100*(20) + 200*(-20) + 300*(10) + 400*(0) + 500*(1) = 15/1500 = 1 PTBB/100.

Poker Tracker would calculate your standard deviation as follows:

100/1500 * (20 - 1)^2 + 200/1500 * (-20 - 1)^2 + 300/1500 * (10 - 1)^2 + 400/1500 * (0 - 1)^2 + 500/1500 * (1 - 1)^2

It does NOT chunk your game into 100-hand sessions artificially and weight them; rather, it just calculates the weighted average of each SESSION'S deviation from your usual winrate.

Theoretically, you could therefore "zero out" your PT SD: start every table and play until your winrate on that table is exactly equal to your "true" winrate. Then quit the table and start a new table. When PT goes to calculate your standard deviation it finds that your winrate is always the same on every table, so your PT SD = 0. Note that you could have the wildest of fluctuations within a session; all this will be irrelevant from Poker Tracker's perspective. All that matters is the winrate you banked over the session.

SirFelixCat said:

[ QUOTE ]

so can you or anyone else explain how to calculate SD like I'm 5? [img]/images/graemlins/grin.gif[/img]


[/ QUOTE ]

Sure.

What you're looking for is how far your typical hand's return is from the overall average.

Say you play 5 hands:

Hand #1: you win $2.
Hand #2: you lose $10.
Hand #3: you win $6.
Hand #4: you in $3.
Hand #5: you win $4.

Your average return per hand is:

Average = (2 - 10 + 6 + 3 + 4) / 5 = $1 per hand.

Of course, you didn't win $1 per hand; you didn't even win $1 in ANY hand. That's just your average. How far away was your typical hand from that $1 average?

In hand #1 you were $1 off ($2 instead of $1).
In hand #2 you were $11 off (-$10 instead of +$1).
In hand #3 you were $5 off ($6 instead of $1).
In hand #4 you were $2 off ($3 instead of $1).
In hand #5 you were $3 off ($4 instead of $1).

To calculate the standard deviation, what we do is this:

1. Square the deviations of each hand from the overall average.
2. Sum those squares.
3. Divide by the total number of observations.
4. Take the square root of the result.

In this case, it would be:

Sum the squares: 1^2 + (-11)^2 + 5^2 + 2^2 + 3^2 = 1 + 121 + 25 + 4 + 9 = 160.

Divide by the number of observations: 160/5 = 32

Take the square root: 5.7ish.

A standard deviation tells us a few things about the "shape" of the distribution of results:

1. About 2/3rds of all observations will occur within one standard deviation of the mean. (in our example, about 2/3rds of observations should be between 1-5.7 and 1+5.7, or between -$4.7 and +$6.7.)

2. About 95% of observations will occur within two standard deviations of the mean. (In our example, 95% of observations would be between 1-11.4 and 1+11.4, or -$10.4 and +$12.4.)

All the standard deviation is really there to do is tell you roughly how "spread out" the results are. Poker Tracker tells you how "spread out" your SESSION results are; what we actually care about is how "spread out" your HAND results are.

[Xanta]

[ QUOTE ]
I'd love to hear from Pokey on this...esp his formula to calculate SD...this has ALWAYS befuddled me.


eta: Wow, ok...so can you (Pokey) or anyone else explain how to calculate SD like I'm 5? [img]/images/graemlins/grin.gif[/img]

[/ QUOTE ]

OK, I'll try to explain what standard deviation is. It's nothing big or scary, it's just a number. Basically, every hand of poker we have has an average result, say +$.10 per hand. To get a handle on variance, we want to know how often any random hand is going to be within some amount of that 10 cents, plus or minus. I won't go into the maths, but basically any random event is 68% likely to occur within ONE standard deviation of your $.10. %95.4 of hands are gonna yield a result within TWO standard deviations of your $.10. You can see pretty easily that the bigger your standard deviation is, the bigger your swongs are going to be. If your standard deviation turns out to be $8 per hand, then there's a 68% chance any random hand will net you somewhere inbetween $-7.90 and $8.10
Make sense?

Now, onto calculating standard deviation. Sorry, but we have to go into a little maths here.

Let:
N = the number of hands we have
x = the result of a measured hand (+$5, -$100, 0, whatever)
X = the AVERAGE of all of our N hands (we said this was $.10 eariler)

Now, for each and every hand you have, take (X-x)^2, and add all those terms up. Then divide by the number of terms you have, N (remember?) and then take the square root of the whole thing. Bingo bango bongo, you have your standard deviation. In the end the whole formula looks like this.

The O looking thing on your left hand side is your standard deviation, the N is the number of measurements, (xi - x with a bar)^2 is the same as (X-x)^2 and the big giant E means take the sum of all of whatever's to the right.

COLES NOTES: Standard deviation is a number that comes from the spread of each your winnings of individual hand from your average winnings/hand. The bigger it is, the more likely it is that you'll have crazy swongs.


HERES THE CATCH!
1)PT doesn't calculate standard deviation like I descibed above, it takes a shortcut that pokey described.
2)The forumla saying there's 68% chance of any random hand getting you one standard deviation away from your average only works if the results of each hand are effected by only random error. This is called a NORMAL function. Poker hands are a NON-NORMAL function, because you're gonna have a lot of hands where you win or lose 100BBs, but it's very unlikely that you'll win or lose 85bbs. You're almost always potstuck to call once you've invested that much, meaning that the distribution of winnings is not just subject to random error, so standard deviation doesn't quite work.

God damn it it came out mathy. SUNUFAGODDAMNBITCH!

[Xanta]

Lol, I got ownnnnnnned.

[blah-blah-blah]

is it just me or is the graph that he posted really insane that he ran that much over expectation and he's losing w/o showdown?

[SirFelixCat]

Pokey and Xanta....<3. Thanks fellas. I truly appreciate it and will print this out. Many, many thanks fellas.

[Xanta]

Disclaimer: Pokey is a million times better at this kind of data managment math than I am, and I might have a few small errors in my response, particularily about normal and non-normal functions.

blahblahblah, it's not insane that he has such a high W$SD since it's not a completely random stat, it's effected by how you play, how tight you are on the river, etc etc. My W$SD is 54 over 100k hands and it's a serious leak of mine that I'm trying to plug, not just running good.