is it possible to beat higher stakes over a decent sample, and to not beat lower stakes for over 200k hands?

i.e:

X guy does not beat 5/10 over 200k hands, but he does beat 10/20 and 25/50 over 200k hands for each level.


If it's possible, can we conclude that 5/10 is harder than 10/20 and 25/50 at least for X?

Yes. You can calculate the chance of it, given some sample parameters, although there will always be some guessing involved.

Not based on this evidence alone, no. Basically for any period of hands, no matter how long, you can estimate how likely particular outcomes are. If you want to imagine having a "true" win rate at each stake, there is a probability that you will run above your true win rate at the lower stake, and below your true win rate at the higher stake, and thus make more at the lower stake in bb/100, even though your true win rate at the higher stake is lower.

There might also be logical reasons you make more at a higher stake, depending on what the stakes are and how far apart they are. You might simply focus more on a higher stake game because the money matters more. Or you might only play the higher of your stake levels when the games are good, and grind the lower stakes games the rest of the time.

yeah, an example of this is ADZ124

Guy is down 30k over 200k hands in 5/10 but at the same time he crushed 10/20 and 25/50 winning over 500k from them and having played over 700k hands in those stakes.

$10/$20 NLH SH 508,011 $339,008 1.67 $67,304 $22,210 Apr 5th, 2010
$25/$50 NLH SH 285,696 $170,200 0.60 $39,654 $13,086 Apr 3rd, 2010
$5/$10 NLH SH 204,165 $-30,178 -0.74 $24,895 $8,215 Apr 5th, 2010


At first it didn't make much sense, but it is possible.

What are the columns here? In the 10/20, is that 508k hands, winning $339,000? Because if so, $339,000 is 16,590bb. That comes out to 1.6ptbb/100, which I think is probably a decent rate but is that counted as "killing" the level? It's probably small enough, with a relatively large standard deviation, to allow a lot of variance even on these time scales.

Try out pokervariancesimulator.fr and enter in 1.67bb/100, 50bb/100 variance, 500k hands. When I did it, there was a huge range, from barely winning to killing it.

I did so, but I should have done something wrong cus it appears as it has only 12k profit.

I put these values: 1.67 50 500000 1

altho it's a positive graph it doesn't show the real profit he had.

When I first started playing online cash games, I was crushing 1000NL over a couple thousand hands, but I couldn't beat the lower stakes at all.

Quite frankly, I don't know what I was doing to beat 1000NL when I was playing, I guess I was just getting lucky.

but yeah, variance exists, and being able to beat higher stakes over a small or even decent sample does not mean you can beat the lower stakes.

200k hands is more than just a decent sample.

Lol, I wish I ever crushed 1000NL over a 200k sample...!

First of all, when you put in those numbers, you're getting a graph in ptbb, not dollars. So you'd need to multiply by $40 to get $

Second, don't put "1" in the number of players entry, use something like "100". The point of the tool is to show you the range of possible outcomes - it simulates the range of results a lot of players with similar stats will see over a given # of hands.