Winrates, bankrolls, and finances
This thread will basically be a containment thread and will stock pile all of the questions and answers about winrates.
What are you asking/arguing? I can't tell if I am just not understanding or if you're being difficult for no reason. This all started with ''is 500 hours enough'' and the answer is kind of but also no, then GG posted a few times giving fairly good insight into variance. I have no idea what you're playing devil's advocate for.Do you need someone to tell you you're doing well? Be
no it didnt "start" with is $500 enough. I just said my argument against sample size has nothing to do with me.
i am basically asking where the idea of smaple size comes from. honestly. cause i wouldnt play if i wasnt a winner. i like money too much
if 6000 isnt enough to be accurate. my question is why keep stats at all then. this has nothing to do with me
The reliability of a series of observations is bigger the more of them you have seems like a pretty easy concept to grasp.
If you are winning a 13bb per hour clip after 500 hours, it's highly likely you are a winner. It's not as certain you are a 13bb per hour winner.
Going to pokerdope and plugging in your observed win rate and the hands you played -and assuming a standard deviation of 100BB/100-, your true win rate should hover between 9bb per hour and 17bb per hour... 95% of the time, if I am not mistaken.
Gonna point out I think there's a bit of a misconception that having more hours is strictly superior in terms of judging your true winrate, which I don't think is fully true. For most people, 2000 hours is going to represent 2+ years of play so you're going to end up with extra variables to consider, like player pool changes, game changes, and personal/player improvement. Insignificant over just a few months but grows in significance the longer you're tracking.
It's probably best to always give yourself at least a +/- 2bb/hr margin of error to your winrate, even over thousands of hours. I also think there's a lot of value in qualitatively estimating your winrate, but requires a lot of self awareness.
Double Board PLO bomb pots are a done in so many games now every dealer change. It's not even the same game and it often takes up 25% of an hour. Winrates must be taken with a grain of salt
Recently hit 1k hrs 2/5 - 15.3 bb/hr
Offering coaching too
https://www.twoplustwo.com/post?postId=59298895#post59298895
What variance (standard deviation) do people usually assume for live play?
I have started a mini project to try to get HUD-like statistics on my live play, and happen to have been on a particularly hot streak. The "typical" figure I have seen for Holdem is 100 bb/100 on the upper end, but am forced to question that when it would say I have a 95% confidence of winning at least 83 bb/100** (29 bb/hr) which is patently ridiculous. Yes, it's a small sample size, but the entire point of a variance figure is to control for that.
However, I figure that all the analysis is tailored to online play which opens for standard sizes, and so might be more realistic to consider something like a 1/2 live game with a common $10 open to essentially have a $5 big blind that the hero is never obliged to post. In this case, it would suggest the standard deviation should be more like 250 bb/100, dumping the low end of a 95% confidence interval to a comfortable -44 bb/100.
**Using /100 because I have exact hand count on my play, and it's more common for poker stats.
If you do a search ITT, you'll find a bunch of folks reporting their SDev/hour and 95% confidence intervals from a few years back. Almost all of us used hours instead of hundreds, due to how we track.
