online players have a much lower winrate, and consequently a much higher relative standard deviation, and that means they need a much larger sample size to get meaningful results, no?

Total for the year 2013....





Just 2/5 NL...clearly playing mixed games and PLO is killing my win rate!

^^^^

16.5 bb/hr, sweet 2/5 NL winrate Slim!

Too bad your other games you're "only" winning at ~$20.50/hr.

GniceresultsG

Can't imagine PLO ever killing a win rate. Limit games are of course understandable.
Very nice work.

Thanks obv fish on a heater

Thanks. Yeah I ran poorly in some big PLO pots the last 2 months or so and yeah I played some 10/20 and 20/40 limit games through the middle of the year.

Trying to figure out the mathematics of a winrate in poker is damn near impossible because the variables are always changing. What we're actually looking for are common sense approximations and even then, what does 500 hours really tell you?

Do you know how I know I'm a winner in my games? It's not based on hours played, profits, standard deviations, or any formulas. It's because players say things like "I always move allin with aces so they don't get cracked" or "I know you got a full house, but I got to see it" followed by me showing them a full house and taking most of their stack and online I've had two different donks CALL and lose their 100bb stack allin on a double paired board because they couldn't fold a flush. You get the idea ...

Truth be told, with the right players, there's probably a live 1/2 game somewhere at some point in time that's so tough, 99% of the players here couldn't beat it on average. On the other side of the coin, there's probably a 10/20 game somewhere at some point that's so easy that 99% of the people here can beat it. Poker is about beating players or in some cases just A player.

Again, unless you play in live or online games where the variables don't change much, you're just making a best guess outside of a vacuum.

The key to winning imo is #1 having a bankroll and #2 game selecting. You may be rolled for 2/5 and a winner in that game, but the best game may be a 1/3 game. 5/10 may be your standard/winning game and TWICE you've tried to take a shot at 10/20 and failed, but if you have the bankroll and the RIGHT, good, juicy game opens up, you should take a seat imo.

Bankroll is just so important so you can #1 handle the swings and #2 not be affected by moving up or down. If you have $50k, there's no 2/5 game that should ever intimidate you since you have 100 buy-ins. Now, if you only have $10k ... well, it better be a very juicy game and even if it is, variance in the short term can sometimes take out a player making all the right moves.

wow the average IQ level ITT is definitely a good 20 points higher than in the strat threads.... let's start posting hands here and having a password to access this thread imo

Just stalk Avaritia and pay attention to everything.

Wonderful post.

So a couple questions for you or anyone. I imagine the minimum meaningful sample size should change depending on if you are playing NL vs Limit correct? What are your thoughts on this? I presume you are saying 500/1.5k-2.5k for no limit correct?

I'd have to presume that getting a meaningful sample at limit is much easier as your win/loss per hand is much much lower. You can't have -200BB hands in limit (well, you can but that just seems stupid...) and obv you won't have +200 BB hands either. Therefor variance is lower, so your sample size has to be smaller to get a more accurate result.

And yes, the 1.5k+ is for no limit.

ok so. for a limit how many are we thinking?

Specifically for me Im thinking of my 2/5-100 Spread Limit game. It's a lower variance game (with expected lower winrate... more on this later maybe)... any ideas? I started keeping good records with pokerjournal only 3 months ago <300hrs and I have a feel that I am a winning player.

Variance is not lower in LHE fwiw

Yeah - limit is higher variance per win rate compared to NLHE... so it actually takes longer to determine any meaningful stats.

what? no limit by definition has more variance... if variance has anything to do with SD of returns... 1/2 limit will have less variance then returns then 1/2 NL... you can't compare 2/5 NL with 30/60 limit and say that 30/60 has more variance? its apples to oranges... I know for a fact that my 2/5 100 spread returns certainly show less variance then my 2/5 NL returns just from a life of poker without having really a huge data set.

Limit is much higher variance ainec

Like the swings for 30 $/hr for nlhe vs limit, limit is way way higher

hmmm. i guess im wrong. i don't know how. It seems I face a few sick all in situations in day seemingly every time I play NLHE and seem to take massive beatings and massive wins in NL.... I've played way more hours in spread games and for it to go bad huge one way or the other its really an outlier... not sure what im missing though

Slippery....Read again... "per winrate".

Variance by itself means squat... variance / WR is what determines downswing potential.

20/40 limit player with a $50/hr WR is going to require a bigger bankroll than 2/5 NL player with a $50/hr WR.

NL is not a swingy game when compared to many other forms of +EV gambling. Although, LO8 is about the most BR friendly game...

ok, maybe i get it now... so a 2/5 100 Spread game is a lower win rate, lower bankroll game then 2/5 NL... so while the SD of 2/5 NL will be larger then 2/5 100 spread... the 2/5 100 spread will have a higher variance then a comparable maybe 1/2 NL game?

Yes.

You're really talking Coefficient of Variation http://en.wikipedia.org/wiki/Coefficient_of_variation, standard deviation over mean.

cool. thanks.

3 months. 280 hours. ridiculous win rate.



i know its not that much data... but it seems to follow a decent curve.

fyp

I played limit for over 10 years. Switched to NL just 1.5 years ago. Limit is way more swingy than NL. Bank roll management is much easier with NL.

(sd^2/2wr)*(1/ror)=br

if you look at the formula, youll notice that the effect of doubling the winrate is that it halves the required bankroll for the same ror (or alternatively you could say it doubles the ror for the same bankroll)

what is ROR?