Can someone please help me understand Net Expected Won convergence?
Posted by thk3421
Posted by
thk3421
posted in
Gen. Poker
Can someone please help me understand Net Expected Won convergence?
My question is whether the Net Expected Won value should converge to the Dollars Won value after a sufficiently large enough sample? According to PT website the answer is that in the long run the two will converge.
http://www.pokertracker.com/guides/PT4/tutorials/all-in-equity-graphs
The longer I play though, the more the two seem to diverge. Is this just short term variance? Is this something that other players see in their charts? Perhaps the Net Expected Won doesn't account for rake? My last 8k hands are shown below for 6max 25NL cash games, with no other games or tournaments and always 100BB buy-in. (queue the violins for sad feelings when viewing this chart :) )
As long as I have your attention: Based on these results I figured I was just sucking at poker while learning, but is there hope that I am just running bad? The Net Expected Won is positive, but I have my doubts that the green line will converge upwards...
.
Loading 3 Comments...
Hi! Variance is tough in NLHE, 6-max. Over 8k hands everything can happen. EVERYTHING. 8k is nothing. I recently read a great article on variance - was in German though - which explained, why you likely need to play between 0.5 and 1.0 million hands before you get an IDEA of your true winrate.
Have a look at your standard deviance - it likely will be somewhere between 80 and 100 I guess? This means, your expected winrate after 100 hands will be - say 2bb, but the real winrate can swing between -98 and +102.
So again, 8k is nothing. You are running unlucky - in regard to all-in-situations. Unfortunately that does not mean that you "actually" were a winning player - even your "expected winrate" is bound to massive variance, the yellow line just shows the all-ins. It does not show if you constantly suck out with your 3-outers on the river. :)
Cliffs: short-term variance is brutal, the lines will (very likely) converge eventually, given enough (!) time.
Thank you for the reply. I completely agree that 8k is too small of a sample to infer anything meaningful about win/loss rate. Nonetheless, I find some comfort in knowing these two lines are supposed to converge eventually.
you could easily go another 1 million hands of them continuously spreading apart
Be the first to add a comment