st dev bb in HM2

[Daniel]

How is st dev bb calculated in HM2? In HM1 it was in the range of 70 and per 100 hands, now it is around 10, even after a short session of only 60 hands with a 100bb win. Looks like something is wrong. Anyhow, please explain how this is calculated.

[DogNamedBluff]

Standard Deviation is how far your winrates deviate from the average per hand. There is more information about how this stat is calculated in poker here; http://www.evplusplus.com/blog/2010/...ard-deviation/

[Daniel]

Well, I know what it means and how to calculate it, but I don't see how the difference between HM1 and HM2 is caused? Which calculation do you use for HM1 and HM2?

overall stats today:

HM1: hands: 1387, bb/100=35.1, st dev bb=92.3
HM2: hands: 1383, bb/100=34.3, st dev bb=9.31

[udbrky]

HM1 = per 100

Hm2 = per hand

This is in discussion currently.

To convert, you just multiply or divide by 10 (square root of 100)

[PlusEV]

HM1 = per 100

Hm2 = per hand

This is in discussion currently.

To convert, you just multiply or divide by 10 (square root of 100)

but if i try to run a simulation like on evplus or calculate confidence intervals myself, which value to use? the winrate is /100 also, so i should use hm1 value? (although it doesnt say bb/100 in the stat there)


also: is the standard deviation of the sample or of the population calculated?

i have 2 hands in my database. one i lost 17.2bb and one i split the pot and won/lost 0. mean is -8.6bb
HM1 says for standard deviation: 120.21

how is that value calculated? i just dont find any formula that fits

[fozzy71]

I have escalated this to our 2nd level support techs and have explained the issues you are having and they will be in touch shortly to help resolve the problem.

[PlusEV]

I have escalated this to our 2nd level support techs and have explained the issues you are having and they will be in touch shortly to help resolve the problem.

great, thanks!

[udbrky]

HM1 used the STDEV function. In HM2, the STDEVPA is used.

The STDEVPA function returns the population standard deviation for a population whose values are contained in an Excel worksheet and specified by the argument(s) to STDEVPA

- Differences between the Excel statistical functions STDEVPA and STDEVP

It is changed because we feel it to be the more accurate value.

If you run both functions in Excel, for STDEVPA, you get 8.6, for STDEV, you get 12.1

Converting that for 100 hands (x10 for the square root function in the SD calc), that would be the 121 you get in HM1.

I'm pretty sure that EVPlus uses /100 st dev, so to be consistent, use this value when using their site.

[PlusEV]

sorry for coming back to your reply a bit later...

thanks for the detailed answer. i read the link, yet i dont really understand what the difference between stddevpa and stddev numerically or lets say in this example. to me it seems stdevpa and stddev only differ in such a way:

if in one cell there is a text string, std dev doesnt treat it at all and stdevpa treats it as a value 0.

this shouldnt make/explain any difference in the example of mine mentioned above or should it?

can you explain the exact calculation/difference in the simple example of mine?

also for evplus i should use the hm2 value then rather than the hm1 value?

[udbrky]

Here are some more. STDEV uses sample, STDEVPA uses population:

Calculating Standard Deviation In Excel

Standard deviation - Wikipedia, the free encyclopedia

In the research I did - this has been about a year ago - I agree with our decision that it is more accurate using STDEVPA, I would use it.

When we've changed stat definitions in HM2, it has been for accuracy.