PDA

View Full Version : Std Dev Bug ?



Ollives
08-07-2011, 01:59 PM
"Std Dev bb" stat seems to be wrong...it is 10x less than in HM1 ie I have 4.6 instead of 46...
Probably easy to change...!

sideshow
08-08-2011, 12:10 PM
Thanks we have verified the bug and it is being written up.

Ollives
08-09-2011, 11:25 AM
Cool!

Ollives
10-05-2011, 02:25 PM
Up....Bug still here!

udbrky
10-05-2011, 08:34 PM
Actually it is a changed stat - you can verify using Excel's STDEVPA stat.

It is per hand, not per 100 hands - which makes it 1/10 HM1 because standard deviation's calculation is the square root.

Ollives
10-07-2011, 07:04 AM
Ok got it...!
But why changing that ? Is there a very good argument to make it /hand ?
We are used to /100 hands data for a lot of things and in that case the /100 hands is, I think, more relevant than /hand...don't you think so?

udbrky
10-07-2011, 03:09 PM
Well it's measuring how much your amount won is from hand to hand, so it doesn't really make sense to measure /100.

Sometimes over time, the way we look at stats change, for example, for a long time, big bets/100 was standard, but now, everyone looks at big blinds/100.

Ollives
10-08-2011, 03:37 PM
Sure! Things are changing with time! But not always for the better! But I got your point...
I am not really convinced though but will go with it...;)
Keep up the good job!

Ollives
10-08-2011, 04:54 PM
Thinking more about it....I am sorry but I feel it really doesn't make sense to have this stat "per hand".
The StdDev directly refers to my winrate and my winrate is "per 100 hands" so StdDev should be "per 100 hands".
It would be useful if we used a winrate on "per hand" basis but thats not the case...You have to keep the same frame of reference...

The sttdev to my opinion measure the variance of my game and I dont really care about variance on a "per hand" basis (not speaking about the fact that i am not playing every hands!)...
Just think about it it would not come to your mind to use winrate on a "per hand" basis....

So although there is nothing wrong with having it "per hand" I deeply feel it is useless and counterproductive...

udbrky
10-09-2011, 09:02 PM
It would, if it were measuring winrate, but it's measuring winnings from individual hands. So it really doesn't make sense to do /100.

Ollives
10-10-2011, 06:08 AM
:D:D:D
What is "winrate" if not "winnings from individual hands" ???

Then you expressed it "/100 hands" but you could do it "/10 hands" , "/hands" or "/125485 hands" that's just some kind of representation...but to make things easier to our brain it is easier thus to have StdDev on the same scale.

But...obviously you dont feel it that way! :D

Fine....

Edit: Actually...we can have a look at the definitions you put here for HM1:
http://forums.holdemmanager.com/custom-reports/3595-formulas-every-hm-stat-updated-nov-12-2009-a.html

udbrky
10-10-2011, 04:44 PM
I'm asking around to see if my understanding is off on it.

Ollives
10-11-2011, 02:26 PM
That is really smart from you!
I will be really interrested to get the final word on this, whatever it is...!
Thanks a lot for your time and consideration!

MezCalito
10-11-2011, 02:32 PM
It makes no sense NOT to have it in the same "time frame" as the winrate. Besides the fact that ppl probably are used to the variance numbers in /100 hands, you obviously also use it in analysis of bankroll requirements and crude confidence intervals, thus it's much easier to have them in the same time frame. I really don't see the point of changing it to std / 1 hand.