How many hands is enough to make an accurate sample size?

Maggiesdad

Trying to take an honest view on how I am playing/running on another site. I want to see if the decisions that I am making and the results I am getting are in tandem. I realise that the sample size will be a major factor in the accuracy of my findings and so am looking for some guidance as to what is a fair and reasonable sample size. All help appreciated.

RLT16

500,000 hands

Maggiesdad

In Response to Re: How many hands is enough to make an accurate sample size?:

500,000 hands
Posted by RLT16

Thank you for your reply. What do you base this on?

RLT16

well i think it depends what game types you play tbh, if your playing mtt's your going to playing alot of hands per tournament so how ur running is going to fluxuate alot per every 10,000 or so hands, so to get an overall long term accurate 500,000 is probably abit much but would definitely do the job

Maggiesdad

In Response to Re: How many hands is enough to make an accurate sample size?:

well i think it depends what game types you play tbh, if your playing mtt's your going to playing alot of hands per tournament so how ur running is going to fluxuate alot per every 10,000 or so hands, so to get an overall long term accurate 500,000 is probably abit much but would definitely do the job
Posted by RLT16

Again, thank you for the reply. I guess that I have got a lot of hard work to do in analysing all those hands!

jdsallstar

The eternal poker dilemma - I am playing well, running well or both or neither or vice versa lol

Even with huge samples it's probably still difficult to fully determine if it's luck or skill that's having the biggest impact. For instance let's say a player gets to 100 final tables in a year it doesn't take a genius to appreciate that the hands on those tables (and the the run good/bad you experience) will have a huge impact on your overall profits. 

I don't really know if there's a right answer but all you can is keep trying to improve and keep making the best decisions you can make. Other problem is of course  that the "best decision" is often widely debated

In summary, just to try to play the best you can because you'll not be able to determine how much luck is affecting your results. 

Maggiesdad

In Response to Re: How many hands is enough to make an accurate sample size?:

The eternal poker dilemma - I am playing well, running well or both or neither or vice versa lol Even with huge samples it's probably still difficult to fully determine if it's luck or skill that's having the biggest impact. For instance let's say a player gets to 100 final tables in a year it doesn't take a genius to appreciate that the hands on those tables (and the the run good/bad you experience) will have a huge impact on your overall profits.  I don't really know if there's a right answer but all you can is keep trying to improve and keep making the best decisions you can make. Other problem is of course  that the "best decision" is often widely debated In summary, just to try to play the best you can because you'll not be able to determine how much luck is affecting your results. 
Posted by jdsallstar

Wow! Thank you for your reply even if it is most perplexing. 

If we cannot accurately, honestly and objectively work out how we are playing then we are only left with results. But what are results? If we only measure in money (in the end, the thing that matters the most) than when we are constantly outdrawing (and by definition - getting it in with the worst of it), we are playing well!

Anybody else got any ideas?

Lambert180

In Response to Re: How many hands is enough to make an accurate sample size?:

In Response to Re: How many hands is enough to make an accurate sample size? : Wow! Thank you for your reply even if it is most perplexing.  If we cannot accurately, honestly and objectively work out how we are playing then we are only left with results. But what are results? If we only measure in money (in the end, the thing that matters the most) than when we are constantly outdrawing (and by definition - getting it in with the worst of it), we are playing well! Anybody else got any ideas?
Posted by Maggiesdad

The one thing we always have to assess how we're playing is our HHs.

It depends what games you play as to what are the most important hands to look at... 

For example, in MTTs, it might be pointless (or rather, not the best use of your time) looking at loads of exit hands when you had <15bb if you are vv good at identifying spots to jam/3bet jam. Fwiw, in MTTs I pretty much wouldn't use results alone to make a decision on how well you're playing at all, the sample of MTTs you'd need is pretty ridic to get a proper meaningful idea, and even then you might be donig some things realllly well and some things reallly bad.

Whatever format it is, find the spots where you think you're losing alot of chips/£££ and look into them. 

But either way, the best way of knowing if you're plaiying well will always be to look at HHs, and work it out. If you're value betting a river and you're not sure about if it's correct or not to do so, you can go work out his range, see what % of it we beat, what % of the hands we beat would call a vbet etc then youll know yourself straight away if you played that street well or not.

In answer to the actual question...

SnGs/MTTs - I'd be more interested in a game count that no. of hands

Cash - Probably 500k+ if you want a meaningful idea of winrate

jdsallstar

That's what I was, poorly, trying to say.

You cannot use results alone to determine how well your playing. Longer term upwards or downwards profit trends will tell us if we're doing something right or not of course but short term results will be heavily weighted to how well we've ran.

By posting hands and reviewing hand histories we can analyse certain spots to see if we're playing optimally or if there are any glaring errors. Hands in isolation though again won't give us a 100% reflection of how we're playing and as i say there is always a multitude of opinions on the best way to play particular hands.

All you can do is try to improve and the results will be the best they can be IMO.

There are many others on here who are way more qualified than me to answer your question though lol I could, and probably am, talking out of my Arsenal.

TeddyBloat

actual results are the last thing to converge. as such - and depending on sample size and format - they can be a poor indicator of skill level.

if you are playing on another site you can buy poker tracker very cheaply and it will allow you to see exactly how lucky you are. you can run poker tracker without using a HUD and can look at results by position, starting hand etc etc.

here's my stars chip graph -

http://imgur.com/TR0CcRD


i've ran well on stars but also have a high confidence that i am a winning player also.

if you are simply looking at results then your true winrate will be harder to discern depending on what format you play.

HU cash, DYMs, HUSNGs will need far fewer games / hands than full-ring, big field MTTs and spin and gos

remember variance is a function of EV winrate.  better players will in general experiance less variance. we can visulaise this with software:

for example here is 1000 simulations of a 52% ev ITM player playing 10k husng games on stars

http://imgur.com/ZBWJ1JK

if the player slightly increase his winrate to 53% evITM then look at the difference in variance:

http://imgur.com/r76LVro

note that he goes from a 10% chance of breaking even over 10k games to never breaking even.


notice also how many simulations are  losing actual $$$'s after 3k, 4k 5k games etc compared to none after 10k

add another half of a percent ev ITM%. so 53.5% evITM:

http://imgur.com/OMskUqo

now no chance of not being a significant actual $$ winner over 10k games, but after 5k games also

small increases in winrate and sample size can reduce variance in actual results.