How many hands is a "good sample size"

[sammyfive]

discuss

 

[thepokerkid123]

Depends what you're trying to prove.

If you just want to know if you're winning or not, 20-50k will give you a very good estimate.

Significantly positive or negative results are enough to make an educated guess if you're overall winning or losing on a much smaller sample. Not sure on what the numbers would be, but if you burn through 15 buy ins in 10k hands it's a pretty good hint.

If you want to calculate your winrate, >100k.



This really all depends on your standard deviation and how precise you want your conclusions to be.

 

[sammyfive]

Alright so that is quite a few hands.

so I'm currently 7 tabling 5NL at 30 BB/100 over about 4000 hands but I have done nothing spectacular and feel like it is realistic to keep winning at near at this pace. How long before I can say that i beat 5NL?

 

[thepokerkid123]

You probably can already. It just becomes a more reliable statement the bigger your sample size gets.

 

[DietDave]

I'd say about 10k is a good start, but 20k-30k will give a better idea. 100k is great but that takes a long time.

 

[shrtstakatak]

If you want to calculate your winrate, >100k.


This really all depends on your standard deviation and how precise you want your conclusions to be.

When you say winrate, do you mean BB/100 for example? What kind of standard deviation would you consider acceptable? +5%, +10%?

 

[No Brainer]

30BB/100 is running super hot. 4000 hands is not a large sample size by any means and your 30BB/100 winrate will definately drop substantionally by the time you hit 50k hands.

Basically to answer your question, more is always better. 50k will give you a decent sample to see how well you are doing at micro stakes.

 

[joeaugie]

Considering top pros go through 100K-150K break even stretches but boast a 2.5-3ptBB/100 winrate, you have to get in about 1 million hands to be certain you are within .5 of your winrate.

It's a tough one to answer completely, but for purposes of improving, I think playing 20K hands, then reviewing, then 20K, etc is a good way to go.

I know I've had 9K break even stretches at 5NL followed by 11K stretch with a near 10ptBB/100 winrate.

The game just has far more variance than most people realize, which is why bankroll management and emotional control are essential to success.