Sample sizes to analyse performance

M

matiusaa

Visionary
Silver Level
Joined
Jun 29, 2013
Total posts
779
Chips
0
Hi I would like to know if anybody knows what a good sample size would be in order to rely on the winrate?? 100k hands? 200k? I play zoom poker. PLEASE: justify your statements.
I noticed that on the HM2 on the reports tab it tells you the winrate (bb/100) and std deviation (bb/100) so, a question to anybody who knows about statistics:
Lets say I have a winrate of 3bb/100 and a std deviation of 80bb/100 in 100k hands. If I want to know the std deviation of the sample I should make 80/(sqrt(1000))?
When you have the std deviation and mean of an aleatory variable, and want to know the std deviation of a sample of the variable, wasn't it calculated doing the std deviation/sqrt(sample size)?
Sorry for my bad english
 
R

RakeMyLife

Rock Star
Silver Level
Joined
May 1, 2013
Total posts
208
Chips
0
Hi, I've researched this question before and it seems there is no hard line, except that the more hands the better. Here are some ranges I've read/heard that may be useful:

25,000 hands: a mediocre sense of true winrate
50,000 hands: a fair sense of true winrate
100,000 hands: a good sense of true winrate
200,000 hands: a great sense of true winrate

As u can see, 25k would be bare minimum I would rely on to get any reasonable sense of ur winrate. However, the tricky thing is that if u are doing well enough through 25k hands, it's very likely u will have made enough money to move up in stakes rather than play to 50k hands at the same level just to find ur winrate.

The standard dev example/question is a little confusing to me, sorry! Not sure why u would need this info (standard dev) though, as ur winrate is the most important number. If ur trying to analyze ur swings as a function of ur playing style (in which u might need the standard dev), then u should recognize that this only provides descriptive information. It won't exactly show u how u can improve.
 
P

ParagonPoker

Enthusiast
Silver Level
Joined
Jan 7, 2017
Total posts
52
Chips
0
200,000 hands for the best sense, 100,000 is still an ok representation, but don't go basing your bankroll on it.
 
M

matiusaa

Visionary
Silver Level
Joined
Jun 29, 2013
Total posts
779
Chips
0
The standard dev example/question is a little confusing to me, sorry! Not sure why u would need this info (standard dev) though, as ur winrate is the most important number. If ur trying to analyze ur swings as a function of ur playing style (in which u might need the standard dev), then u should recognize that this only provides descriptive information. It won't exactly show u how u can improve.

Well maybe even 200k are not enough. Your winning rate has a mean and a std deviation, just as every aleatory variable, if not you would win the same amount every hand. Like any aleatory variable, the bigger the std deviation the more variance you have, and hence the more hands you need to play.
 
R

RakeMyLife

Rock Star
Silver Level
Joined
May 1, 2013
Total posts
208
Chips
0
Well maybe even 200k are not enough. Your winning rate has a mean and a std deviation, just as every aleatory variable, if not you would win the same amount every hand. Like any aleatory variable, the bigger the std deviation the more variance you have, and hence the more hands you need to play.

Not exactly. No matter how many hands you play you will always have a standard deviation. As alluded to earlier, it will more or less show u how swingy ur sessions are. This doesn't depend on the amount u play, but HOW u play...
 
Top