You answered your own question: existing rake rate is too high to be sustainable. Sites would lower the rake to keep generating revenue in the future.

oops: http://en.wikipedia.org/wiki/Laffer_curve

it's perfectly sustainable, but it's not necessarily fair OR optimal.

IMO our best hope is for blizzard/world of warcraft to use their huge bankroll/brain trust to tap into the gaming industry, offering a 60 dollar client and 14/month type fee that they use with their wildly popular video games

why would they care? People still keep playing..

i couldn't find one for this year with a quick search, but this is the winrate distribution of players who played 1k+ hands in 2010 according to ptr.



that was 2 years ago and % of winning players was nowhere close to 20-40%. granted that is all cash games, but adding in tournaments wouldn't raise it anywhere close to that much.

Thanks for the links. My intuition was backwards. Seems that the survivorship bias of winning players being more likely to continue playing substantially outweighs the higher "variance" for short-term players creating more winners.

Looks like this data fits 20-40% winners reasonably well, actually. You have to compare the areas under the curves on each side of the axis, not the peak points of the curves.

edit: er, wait, perhaps not, we can't see the entire left tail here, but it should still be close I think?

Looking carefully at this graph, at full ring:

3.8% of players win 0-0.5 bb/100 (pro-rated equally between -0.5 and +0.5)
6.1% of players win 0.5-1.5 bb/100
4.2% of players win 1.5-2.5
3.3% of players win 2.5-3.5
2.1% of players win 3.5-4.5
1.7% of players win 4.5-5.5
1.2% of players win 5.5-6.5
1.0% of players win 6.5-7.5
0.8% of players win 7.5-8.5

and so on...

The above equals 24.2% of players winning between 0 and 8.5bb/100, and there's some more winning above this level. It will end up being about 1/3. So I think that your graph really hammers home my point. Thanks for confirming.

Cheers

DC.

I think he answered the 2nd bit with this statement;

"For instance sites could charge a rake on withdrawals inexcess of deposits."

If a player is withdrawing more than he is depositing he cannot be a fish and will understand the rake method.

i don't think it works quite like that since you could put an infinate number of points on the graph. i think its more like you would have to compare the area of the underside of the lines on one side of the y axis to the area of the underside of the lines on the other side of the y axis. i think you would still be right though, although i'm not sure how many players bust before reaching 1k hands.

it is quite possible mathematically. moreover, even a much higher percentage of winners is mathematically possible, even with the rake and no rakeback

This is the most perfect argument why pokersites would benefit from more winning players;



A small increase in winning players results in a huge increase in amount of hands played.
Imagine you got the first 4 groups -95.5% of the population- playing 5k+ hands.
All the other numbers at the bottom would move to the right as well.

The alternative is we lose most of these players and all the numbers will move to the left.


By the way, this graph shows roughly 20% are winners, however small.

This true. My own data shows this as well

My blog has some data (my nick @ blogspot)

The statistics book also shows this data

according to this guy's research:

http://www.sharkscopers.com/blog/the...rs-are-winners

26% of people who played online tournaments and at least 100 games made a profit. (this was 2010 and also updated in 2011 without much change) i would guess that would be higher than the % of cash game winners since sites offer a lot of guaranteed prize pools and things like that. also, a good player is more likely to play 100 games without busting than a bad player since the bad player not only will play bad but also is much more likely to have bad br management.

tbh i'm not really sure how you could come up with an accurate number. over small samples variance is very high so -ev players will still only lose so often. over large samples, high %s of -ev players disappear from the results because they are already busto.

We should be careful using this chart.

(1) It was produced hastily without verification and the original author warned it might be incorrect.

(2) The methodology is not fully clear (I don't even know whether it is net of rake for example).

(3) Bear in mind that each bar includes all the winning players to the right. e.g. the winners in the 1k+ bar include all the winners in the 50k+ category. If the bars were discrete groups then the slope would be steeper from bottom left to top right.

(4) It just feels wrong imo. A few hundred hands is nothing in poker. The percentage of winners in the lower categories should be close to 50%, whereas, if the adjustment in (3) is made, the chart would indicate more like 10%.

(5) Note also that the chart is based on a sample of hands across various sites at a particular time (not later than 2009). So the players with small hand numbers are not neccessarily less experienced. They may play most of their volume on other sites or may not have happened to play much in the period of the sample. Also, the sample covers a range of stakes.

I don't have a particular axe to grind but just pointing out that we shouldn't draw conclusions mainly on the basis of this source.

In my data winners was close to 50%. I guess that's because most players I have in my db played very few hands. Then it's basically just variance if u win and lose.

However when u play 100 tourneys u are most likely to lose. If u play 1000 hands of cash ur still pretty likely to win even if ur bad.

So everybody is still talking about the rake being to high and that's a problem.


But why?

You can always argue that a price for something is too high.

It's not about the rake. It's about the ratio of money won vs what the sites make.

I never see anyone thinking about this. I think the fee a site should charge should never be higher than 20%.

something interesting to try would be to take all the players that have a sample that is under the minimum required all the way down to 1 hand. put all of their results together into a significant sample and i think that you will find that this group of people that is left out of the results has a much higher rate of loss than the people that have over the minimum sample. i think that would hold true nomatter what you set the minimum sample at because of the effect of -ev players playing less hands on average.

http://2.bp.blogspot.com/-iuh3uC5BEw...o%2Bfilter.jpg

These are the winners/losers across all random data I have in my db.

The WPA winners would win if they did not pay rake.

The number of winning players can theoretically be anything <100% over a decent sample, but is this really the point? More realistically and for arguements sake lets say 25% of players are winners over a 60k hand sample, the graphs in the OP show that a player of average poker skill has <1% chance of being in this 25%!

I think that with this serious lack of heaters among average-skill players over what today is a pretty small sample, they will be so aware of their lose-rate individually but more damaging, communally, that they will be constantly leaving the games or convert to sitout-bumhunting.

As fees would be taking as a % of profits, I can't see how it would encourage mass-multitabling, only depends how much you win throughout the year

After doing a few sums I realised that indeed 50% withdrawal fees would equate to current site takings. This highlights how disgustingly high and damaging the rake currently is and could only be taken in a hidden way such as raking winning pots.

well. if u play 1 hand ur chances of winning or losing is kinda all luck. so its not a good way of determine if u are a winner.

the problem is to know if folks are winners requires too many hands. you are not going to play 50k hands if you are playing just for fun.

what i'm saying is that you could combine all the hands played by players who did not meet the minimum requirement as if they were one player and compare their rate of loss to those that did. these players i'm sure account for a very high% of the player pool and i would be very surprised if they weren't much more -ev then the players that meet the requirement.

It looks like a large part of the population doesn't even get to the 100 games.
They lose and don't like it and look for something else.
If only their first experience had been more positive they might have been hooked instead.

The OP has a great point and pokersites should be aware of this.
There may not be much direct price elasticity, but the indirect effect is enormous.

Also the point made about mouth to mouth advertising is important. 10 Years ago me and all my friends used to play live every day. Then one of us joined a pokersite and told the rest about it. Most of us were pretty sceptical, but he told us enthusiastic stories and in the end all of us joined the site. The stories you hear today aren't that inviting. Poker is not sexy anymore.

Rake catagories alongside rake reduction

Charge rake price depending on players profits:

example:
Cat A - Big Winners = rake fee of 9bb/100
Cat B - Small Winners = rake fee of 4bb/100
Cat C - Breakeven = rake fee of 1bb/100
Cat D - Losers = rake free

Rake would be taken as normal with excess instantly being refunded to player accounts or bonus account depending on their profit catagory

Rake prices would still have to drop for cat A rake to remain competitive with other sites

As rake is charged per 100 hands this would discourage big winners from mass multitabling