ITH Poker Forum

[jfletcher]

Through my first 2786 hands in pokertracker, my SD is 15.4 BB/HR.

Can someone explain exactly what standard deviation is. I understand that it measures how consistent your results are, but I don't know what would be considered good and what would be bad (good being more consistent, I mean).

[JRF2k]

I cannot explain it, but I can give you some helpful links!

http://www.bio.davidson.edu/Courses/Bio ... b%207.html

http://www.robertniles.com/stats/stdev.shtml

http://mathworld.wolfram.com/StandardDeviation.html

http://davidmlane.com/hyperstat/A16252.html <-- this one has info on variance too.

Hope these help.

JRF2k

[jfletcher]

I clicked one of those and got a headache. There's gotta be someone out there who can explain this to me, in a poker tracker context.

Thanks.

[SirNicksolot]

It has been a long time since I took stats, but ignorance has never prevented me from giving advice, so I will take a stab. (I did not read the article posted because I didn’t want the facts to influence my response)

That stat is saying that you are earning 15.4 Big Bets per hour and that there is a 66% likelihood that it will continue.

[MXRider]

The way that I understand standard deviation is that if you have a 15.4bb/hr standard deviation, you are seeing an up or down swing of 15.4bb each hour. I could be wrong about that, but that's what I get out of it.

MX

[SirNicksolot]

The way that I understand standard deviation is that if you have a 15.4bb/hr standard deviation, you are seeing an up or down swing of 15.4bb each hour. I could be wrong about that, but that's what I get out of it.

MX

That makes more sense.

[Deacon]

My statistics course was very long ago, but from what I remember

If you are earning 1BB/Hr with a Std Dev of 15BB/hr. For any given hour you have a X% chance (with X being 90% I think, it probably says in those articles what X is)
of being within 1BBr +/- 15 BB, or -14BB to +16BB

[darvon]

Gather 'round people and I will tell you about SD.


Imagine a set of glass tubes standing vertically. Each tube is set vertically next to one another, making a line 20 foot long. Each tube is 4 foot high, open at the top, with an inside diameter of a ping pong ball.

Got the image? Good.


Now let us imagine you were an exactly break-even player, i.e. your average BB/100, averaged over 100,000 hands, is $0.00.


Now take the tube in the dead middle and mark it -$.50 - +$.50 with a grease pencil.

Mark the one to the right of that one as "$.50-$1.50" and the one to the right of that "1.50-2.50" and so on all the way to the right. SImiliarly mark the ones to the left in negative amounts.

Got it?


OK. Now play 100 hands. Find the BB/100 and put a ping pong ball in the tube which corresponds to the value of the BB/100 of your 100 hands.


Then do it again for another 100 hands, generating a new average for ONLY those 100 hands and put in another ping pong ball.


Repeat 1000 times.


Now you as you look at the tubes the top ball in each tube will define a "bell shaped curve" Remeber that from high school? If not, what you will see is the highest point being in the middle as it is the most probable result (because you actually are a break-even player). The closer to the middle the higher the stack. Your average result is $0BB/100, but there is another component to describing the curve of your results. Is your curve "narrow" or WIDE?

i.e. If your middle value is a 4 ft stack, how far do you have to go left and right before you have a 2 ft stack. you could have a tight cluster where you only have to go a couple of tubes to drop to 2 ft or a dispursed curve where you have to go dozens of tubes over to drop to 2 ft.


This "tightness" of your curve of results is your Standard Deviation of your BB/100 results.

If I recall exactly, the Standard Deviation of your result is how many tubes over (left and right) do you have to go before you have accounted for 63% of all the ping pong balls. So if you have a "tight" curve, your SD might be 3 tubes. If you have a wide curve, you might have a SD of 10 tubes.

So if you have an SD of $10BB/100 then if you play 100 hand sets and play 1000 of them then your result will be with -$10 and +$10 of your average about 63% of the time. The other 27% will fall outside of that range with lesser probabiliy the farther out you go.


Now if that was clear, please clean up those tubes and ping pong balls and get them back to the ITH warehouse.

[SirNicksolot]

Nice explanation Darvon. Obviously you used those tubes for this demonstration in college and not to build "recreational paraphernalia".

[jfletcher]

OK Darvon, but what does that mean to ME? My SD is 15.44 BB/HR, so what does that mean about my play?

Does is mean that in 63 percent of my hour-long sessions, I will win at least 15 big bets? That doesn't seem right.

[niin]

Nice explanation, Darvon.

fletcher, it means that 63% of the time you should be up or down a maximum 15.44BB/HR. 27% of the time you might be up or down more than that.

If you're averaging, say, 1BB/hour over 10,000, if you were to look more closely at some point in time during those 10,000 hands, 63% of the time you would be up or down a max of 15.44BB.

Standard deviation is the potential deviation from the norm. The higher your SD, the bigger swings you will encounter. The lower your SD, you'll see smaller swings both up and down.

Note that that works both ways... a player with a high standard deviation will be up *or* down their SD amount. So if you had a 100 BB/HR SD, you could have 100 BB swings up or down at any point in time.

As for your specific number... That means in any one hour session, you will likely not win or lose more than 15BB. I'm not sure what a typical SD is.

[UnderARock]

So what is a "normal" or "good" SD? Or is it simply not useful to deal with for less than a huge number of hands?

[darvon]

OK Darvon, but what does that mean to ME? My SD is 15.44 BB/HR, so what does that mean about my play?

Does is mean that in 63 percent of my hour-long sessions, I will win at least 15 big bets? That doesn't seem right.

Exaclty what NIIN said. It's the MAX size of MOST of your swing. So if you are at 1BB/hr and your SD is 15.44 then MOST of your hours will be between 16.44 and -14.44.


I think a "normal" SD is low teens. If you have a SD considerably higher than that, I think it means you are not folding soon enough. Sometime you get lucky and get pots you shouldn't and sometimes you aren't lucky and you lose more than you should on a hand. But this would also affect your BB/hr average.


I can't tell you what a big or small SD mean as far as play types.

Anyone else???

[NYIsles]

Where is the standard deviation number on poker tracker? I've spent the last 15 minutes poring over every page, and I can't seem to find it...

[The Waterboy]

Laymens terms, SD is the average zone you can expect to be in. First you need average BB/100 hands say 30. Your SD is 15. This means that your average session is between 15 and 45 BB/100. What this tells us is that you may be a loose player and you can expect your stack to flucuate widely as you are + or - 15 bb/100 with more than 1/4 of your sessions exceeding that in either direction.


For example a rock may have a SD of 10BB/100 with 80% of sessions within 10BB/100 hands of average BB/100.