Percentage of online players who are profitable

[Voltaire]

One more point. As you, runout_mick, suggest winners at one level will tend to be losers at a higher level, and that may result in a higher percentage of players being losers overall. Hard to figure that with any precision though, except by those who have access to the players' accounts.

[T50_Omaha8]

I think online at any given moment around 25% of players are winners. A number like 5% being winners would suggest that every time you play a 6-max cash game, there is a 75% chance you are at a table full of losing players. This seems low.

I have played certain games long enough and with small enough player pools to where I knew who all the winning regulars were, and it was always around 1/7-1/8 or so of the competition at any given time. I'll bet another 1/5-1/6 or so were marginal winners or people who weren't all that regular, etc.

Thus I think it's reasonable to estimate that if you click open a random table, you can expect around 1/4 of the people there to be lifetime winners at poker. This doesn't mean, however, 1/4 of ALL people are lifetime winners at poker.

Those who win keep on playing. Quitting is generally the loser's move, so I think if you count losing players who got frustrated and quit poker, you'll get closer to 5% of players being lifetime winners.

[HesseJam]

[ QUOTE ]
I just don't buy it, but that's fine.

If the best long-term NL winners are under 10PTBB/100 and breakeven is the 95th percentile, where is all the money going? Put differently, I would expect to see much higher win rates among winning players if 95 percent of the player pool were losers.

Lucky

[/ QUOTE ]

All the money is going to the Poker Sites.

[HesseJam]

[ QUOTE ]
[ QUOTE ]
I have similar stats for around 200k hands at $25-$200NL. At PL, around 58% are losers.

[/ QUOTE ]
This type of figure always comes up in these discussions (multiple times on this thread), and it is always garbage.

Maybe 40% of players would win if they played 50 hands and then quit. That's what a 200k hand database is showing you when the typical players have only 50 hands. The median hands played by your opponents doesn't change much when you double the size of your database.

That 60% of your opponents are losing after so few hands actually indicates that the vast majority are hopeless. It takes a substantial loss rate (about the rake), to be down 60% of the time after a few orbits.

[/ QUOTE ]

The first two paragraphs are correct.

[HesseJam]

Several problems with this question have already been pointed out:

1) It is not really meaningful to cluster players into losers and winners.

a) if you go through your unfiltered PT database, you will always end up with a 45/55 ratio or something like that because you will have many players with a few hands which distorts the results. Variance dictates that almost anybody can be up after 100 hands.

b) You can also not go with the statements by the Poker sites. There are many players who might lose interest in a particular poker site after less than 3000 hands because they just played until they got their bonus. After 3000 hands I suspect the winner/loser ratio to be at 30/70 or something like that - just because of variance! Also, there are many players who bust with their initial deposit at that particular site and never return.

c) many players are/become quickly profitable at lower levels and bust on higher levels.

2) It is much more meaningful if you try to establish an average table and see how many winners one table can sustain. If you assume at 9-seat NL25 table you have only one winner with 6 PTBB and two break-even players with 0 PTBB at the table and also assume that the poker site eats up about 0.5 PTBB per hand (a bit high, I know) then you just can calculate how big losing the remaining 6 players HAVE to be:

They have to fund 100 hands x .5 = 50 PTBB
They have to fund 6 PTBB for the winner

So, in total, 6 players have to bleed 56 PTBB/ hands, which means on average they have to be -9 PTBB losers.

Now, you might say the ratio is 1 winner to 6 losers = 14% winners in this scenario. Again, this does not give you the whole picture. The winner is likely to play many many more hands than the losers. Thus the constant supply of new losers has to happen to keep up these game conditions for the winner (and the break even players).

This leads to the reason why the games are getting tougher: For every player who makes the progress from loser to winner you have to constantly supply 6 losers more. At one point in time the sites can no longer keep up with constantly supplying an ever growing amount of losers. Then the formerly break even players become losers. Players who had won at one level have to step down a level to keep their head above the water which in turn increases the pressure on the players a level below. -> dog eat dog.

3) Conclusion: There cannot be more than 1-2 winners at any one particular table. Having 3 big winners at one table - and they have to stay big winners at this particular table - requires too many players who go busto within a couple of 100 hands.

[lucky_mf]

[ QUOTE ]
2) It is much more meaningful if you try to establish an average table and see how many winners one table can sustain. If you assume at 9-seat NL25 table you have only one winner with 6 PTBB and two break-even players with 0 PTBB at the table and also assume that the poker site eats up about 0.5 PTBB per hand (a bit high, I know) then you just can calculate how big losing the remaining 6 players HAVE to be:

They have to fund 100 hands x .5 = 50 PTBB
They have to fund 6 PTBB for the winner

[/ QUOTE ]

I like this – but first let us agree that PT calculates win rate net of rake. So basically the win rates still have to sum to zero if everyone is playing the same number of hands.

So there is one guy winning 6 PTBB/100 and two players with 0 PTBB/100 (all net of rake). Because rake is only paid on pots a players win, the win rates of all players involved would be higher in absence of rake.

[ QUOTE ]
So, in total, 6 players have to bleed 56 PTBB/ hands, which means on average they have to be -9 PTBB losers.

[/ QUOTE ]

Here is where I start to disagree. If there is only one winning 6 PTBB/100, and two breakeven players, the remaining players 6 players must be losing 1 PTBB/100 (net of rake). This is the point I was making earlier: If the pool of winners is small (in this case top 11%) then the remaining players must be bleeding slowly. In this case 6 players are losing 1 PTBB (net of rake) and 2 players are at 0 PTBB net of rake. If we had one (the same) winning player over 2 tables 17 different opponents (top 5% winners) the results would be 6even more dramatic.

As we know there are big fish who lose at a fast rate, the winnings have i) have to be concentrated heavily at the very top of the distribution (meaning very high win rates for winners), or ii) spread out across more of the distribution. I vote for the later.

Lucky

PS – and I don’t think the games have gotten harder.

[Dima2000123]

Here's a simple example of how PT can bias upwards the % of winners.

Let's say that we have two sessions. In first session, player A wins $50, and player B loses $60. The other $10 goes to the rake.

In the second session, player A loses $60, and player B wins $50. Again, the other $10 goes to the rake.

If your PT only captured either of the two sessions individually, it would say that 50% of the players are winners. Taken together, 0% of the players are winners.

Statistically speaking, high variance doesn't bias the estimation of the mean. If you want to estimate the average win rate for all the players, you can do so with a 200k hand PT database. However, high variance does bias the estimation of fraction of means beyond an arbitrary point (like winrate>0).

[jogsxyz]

[ QUOTE ]
95-5 would not surprise me.

Pokertracker databased are biased.

[/ QUOTE ]

PT numbers are not biased. Winrates are -5bb/100 to +3bb/100. Standard deviation is 15-20 bb/100. Until the ppl in the sample reach 10,000 hands the s.d. has a greater effect on winrates than skill. There needs to be 100,000 to 160,000 hands before the real winrate of each player emerges. 10K hands is plenty for recognizing a hopeless player.

[bernie]

All of them.

Just ask them. [img]/images/graemlins/smirk.gif[/img]

b

[HesseJam]

[ QUOTE ]
[ QUOTE ]
2) It is much more meaningful if you try to establish an average table and see how many winners one table can sustain. If you assume at 9-seat NL25 table you have only one winner with 6 PTBB and two break-even players with 0 PTBB at the table and also assume that the poker site eats up about 0.5 PTBB per hand (a bit high, I know) then you just can calculate how big losing the remaining 6 players HAVE to be:

They have to fund 100 hands x .5 = 50 PTBB
They have to fund 6 PTBB for the winner

[/ QUOTE ]

I like this – but first let us agree that PT calculates win rate net of rake. So basically the win rates still have to sum to zero if everyone is playing the same number of hands.

So there is one guy winning 6 PTBB/100 and two players with 0 PTBB/100 (all net of rake). Because rake is only paid on pots a players win, the win rates of all players involved would be higher in absence of rake.

[ QUOTE ]
So, in total, 6 players have to bleed 56 PTBB/ hands, which means on average they have to be -9 PTBB losers.

[/ QUOTE ]

Here is where I start to disagree. If there is only one winning 6 PTBB/100, and two breakeven players, the remaining players 6 players must be losing 1 PTBB/100 (net of rake). This is the point I was making earlier: If the pool of winners is small (in this case top 11%) then the remaining players must be bleeding slowly. In this case 6 players are losing 1 PTBB (net of rake) and 2 players are at 0 PTBB net of rake. If we had one (the same) winning player over 2 tables 17 different opponents (top 5% winners) the results would be 6even more dramatic.

As we know there are big fish who lose at a fast rate, the winnings have i) have to be concentrated heavily at the very top of the distribution (meaning very high win rates for winners), or ii) spread out across more of the distribution. I vote for the later.

Lucky

PS – and I don’t think the games have gotten harder.

[/ QUOTE ]

PT deducts the rake from your winnings, thus somebody has to come up with the rake. At PL 1 at Stars all players together lose 4.3 PTBB/100 on average and this is the rake. If you have only 1 winner with 4 BB/100 at a table you have a shortfall of 8 BB/100 to fund the rake and all others have to be 5.3 bb/100 losers.