Where does the money we win come from?

[gdaviet]

On more than one occasion I've pondered where the money we extract from these games actually comes from. I decided to look through 6 months of pokertracker data and see what I could come up with. Your thoughts and comments are welcome and appreciated.

The sample: July - December 2006 100NL UB -> 81,846 hands.
Unique players: 4,503
Winners / Losers: 1,802 / 2,701 (40% / 60%)
Average win rate: -3.18 PTBB / 100
Total $ Won: -$45843.58
(Turns out this works out to almost exactly what PT reports for rake)
Total rake: $45,871.25

Ok, that all makes sense, as a our group of players is loser equal to the rake. Exactly what we would expect from a zero-sum event. Now I add a filter of 2,500 hands to the same data.

Unique Players: 26 (Wow, 99.4% of players get filtered out here. For reference, I played 32K of the 82K hands in this sample, and the 2nd most active player played 20k hands)
Winners / Losers: 13 /13 (50% / 50%)
Average win rate: 0.48 PTBB / 100
Total $ Won: $1,400.72

Now we have established that the less frequent players have contributed $47,244.30 to our game. Now let's shift the other way in our filter. Let's examine players how have less than 100 hands in our sample.

Unique players: 3,077
Winners / Losers: 1,161 / 1916 (38% / 62%)
Average win rate: -14.95 PTBB / 100
Total $ Won: -$33,940.19 (72% of all the losses)

And finally, let's drop our filter to 50 hands or less.

Unique players: 2,161
Winners / Losers: 765 / 1,396 (35% / 65%)
Average win rate: -27.64 PTBB / 100
Total $ Won: -$26,320.70 (56% of all the losses)

These results were truly eye opening to me. 72% of the money flow comes from people who lose all they can stomach within 100 hands. An astounding 56% of the money comes from a group that doesn't even make it 5 rounds.

That donk you always play with regularly 4 times a week ... he's actually only small potatoes in terms of contributing money to the game.

Here's the usefull idea I've taken away from this. We've all played with some of these aweful players. Everyone is trying to get involved with them before they go broke. I realize now that there is no need to go out of my way to play pots with them. The are going to lose to someone, but that someone doesn't have to be me. Their money flowing into our game is all that actually matters. That flow of money is what makes the average players, that we can beat, small winners. And small winners keep playing and redistributing the windfalls they get from these truly aweful stop, drop, and leave players.

[bigred]

It makes sense. I always figured those guys who watch ESPN poker and decided to play with $50 are the main fuel for these poker sites. This is a cool post though.

[Halv]

Yeah it makes sense. But there is probably a sample size issue too (for excample that gambler that deposits once or twice a month and loses it all will not show up in the filtered stats for a month, but he might be in the yearly stats).

Here's the usefull idea I've taken away from this. We've all played with some of these aweful players. Everyone is trying to get involved with them before they go broke. I realize now that there is no need to go out of my way to play pots with them. The are going to lose to someone, but that someone doesn't have to be me. Their money flowing into our game is all that actually matters. That flow of money is what makes the average players, that we can beat, small winners. And small winners keep playing and redistributing the windfalls they get from these truly aweful stop, drop, and leave players.

This is imho a bad conclusion. The "average" players (that are actually probably breaking even/losing long term) will continue playing even if they're not winning off of the aweful players. If they're smart they might move down a level and rebuild their rolls there, but they will come back up and feed us. Bottom line is that I want every dollar that someone put onto a poker site to find is way into my bankroll. Punishing both the aweful and the average players will put me closer to that goal.

[biondino]

How many non-regulars have you spoken to who ask "how much do you spend on poker"? These are the people who are happy to spend $100 or whatever a month - they expect to lose, they expect to have to pay to enjoy their hobby. I have met quite a few - they are genuinely surprised when I explain that actually, you don't have to resign yourself to being a losing player.

[DubRod]

The "average" players (that are actually probably breaking even/losing long term) will continue playing even if they're not winning off of the aweful players. If they're smart they might move down a level and rebuild their rolls there, but they will come back up and feed us

So true i feel like crying!

[pokerroomace]

it all comes from the button and moves it's way to the top it seems.
you get new players at higher stakes but the vast majority of the money comes from the lowest stakes.

i've also wondered how anyone manages to profit from poker? apparently 85% of online players are losing players.

[biondino]

No, that's not true. Look at gdaviet's stats above - 40% are winning players (and though the great majority of players have tiny samples, they remain valid because of the sheer number of them.

On my old PC (45k hands), the percentage of winning players was like 39.5% - let me check what it is on this PC (74k hands)...

39.88%. Seems pretty conclusive.

[Halv]

i've also wondered how anyone manages to profit from poker?

Because:

apparently 85% of online players are losing players.

(Or was that a retorical question?)

[bigred]

In a recent university study they found 50% of all online players to be above average. This is why too many players for me to be profitable! We're doomed!

[givememyleg]

Very interesting post.

[zook]

Nice post gdaviet. Can you tell me how you did the filtering? I have a big database I want to do the same analysis on and I can't figure it out...

[martindcx1e]

Nice post gdaviet. Can you tell me how you did the filtering? I have a big database I want to do the same analysis on and I can't figure it out...

nOOb!

[dev]

There can be 40% winners at a given level, but those 40% tend to bring their money to the next level and lose. Therefore, less than 40% will be overall winners.

It's interesting to look at the statistics behind all of this. I play mostly B+M and home games, and we tend to look for specific marks rather than trying to play a more mathematically tight game. It's the 'awful players everyone is trying to get involved with' that gdaviet was talking about. I guess when you play the volume that you play online, and the game never really ends (rather than ending shortly after the fish go broke), it makes more sense to just outplay your opponents the sklansky way.

[zook]

Nice post gdaviet. Can you tell me how you did the filtering? I have a big database I want to do the same analysis on and I can't figure it out...

nOOb!

Thanks for the help

I can filter players who have played over a certain # of hands in the preferences tab, but can't figure out under. Also don't see where the total $ won numbers come from.

[biondino]

There can be 40% winners at a given level, but those 40% tend to bring their money to the next level and lose. Therefore, less than 40% will be overall winners.

I'm not convinced this is true - so the assertion is basically that the average poker player will continue to move up to the level at which he becomes a losing player (and where he manages to lose all his accumulated earnings and more)? I dunno, that sounds pretty unlikely to me. There are probably 100x as many players at 100NL than $10/$20NL and I'll wager the considerable majority have never played above that level.

It's common sense to either play at the highest level where you remain a winning player, or at the level where you have the most earning power, right? Let's say the 60% of losing players on my PT are donks who will never be winning players. That means more than 60% of the winning players will ultimately overstretch themselves and become losing players? It just doesn't make sense - which doesn't mean I don't want to be shown how my argument's wrong.

[gdaviet]

On the filtering question:

You can't filter for 'fewer than x hands'. You have to filter the other way and work out the difference between those stats and the stats with all the players. You can do custom queries too, but I didn't get that adventurous.

Ex: All players: 4,000 Players with more than 100 hands: 3,000 .... so there are 1,000 players with fewer than 100 hands.

Total money: You take the average win rate and factor it against the total hands played (this is NOT the same as the total hands in your sample ... one hand history with 9 players is 9 hands ... but this number is on the summary tab).

Hope that helps, have fun.

[zook]

Thanks gdaviet. Didn't think of that... was hoping there was an even easier way I guess

Did the same analysis on my DB of 2.8 million hands, mostly from 100NL FR. The numbers come out strikingly similar:

Total unique players: 16,286
Total hands: 2,839,925
Avg. winrate: -3.65 BB/100
Total lost: $103,657
Winning players: 6,165 (37.5%)
Losing players: 10,190 (62.5%)

Players with more than 2,500 hands: 122
Avg. winrate: +0.74 BB/100
Total won: $3,742
Winning players: 74 (60.7%)
Losing players: 48 (39.3%)

Players with less than 100 hands: 11,106
Avg. winrate: -10.45 BB/100
Total lost: $78,265 (75% of all losses)

Players with less than 50 hands: 8,065
Avg. winrate: -17.17 BB/100
Total lost: $56,293 (54% of all losses)

Players with less than 25 hands: 5,099
Avg. winrate: -30 BB/100
Total lost: $34,086 (32% of all losses)

I've been thinking about this data, and while it's interesting, I think there's a simple explanation for it... very, very few poker players' final hand at a given site/level is going to be a win. Since all these hands are a snapshot, of course we're going to catch some players who just made their deposit, some who are in the middle of their poker "career" at this site/stakes, and some who are at the end. But since the end is almost always marked by a loss, the players with the fewest number of hands are going to be almost all losers. Unfortunately, because these databases don't include every hand each player has played, I don't think we can draw any meaningful conclusions (besides that the pond needs constant restocking with fish, because a lot of them leave after they lose). To sum up, I think there's a serious sampling bias here.

[dev]

I don't know if we can go with a sampling bias here. We have large sample sets, all with similar results, and no reason to believe that a larger sample would change anything.

[zook]

I don't know if we can go with a sampling bias here. We have large sample sets, all with similar results, and no reason to believe that a larger sample would change anything.

I'm not saying a larger database would change anything... let me try to restate... it shouldn't surprise us that the players with the fewest number of hands are the biggest losers, b/c most players only stop playing after a loss. These 5,099 players that played 25 hands or less and lost 32% of the total $ might have played hundreds or thousands of hands that aren't in this database. All these numbers show, imo, is that players tend to stop playing after a loss. To draw further conclusions, we would need a database that included ALL of every player's hands, starting from their first deposit.

[flyingPenguin]

All these numbers show, imo, is that players tend to stop playing after a loss. To draw further conclusions, we would need a database that included ALL of every player's hands, starting from their first deposit.

In terms of "where the money we win comes from" we only care about the money put into or taken out of the system while we are a part of the system. Those players may or may not be winning players overall, but the important fact is that they have been losers while we (individually) are in the system. Therefore, they are where our money comes from.

Basically, the rest of their careers is irrelevant. Since we're getting similar figures from different people, we can assume that if some people are winning players overall and go through a losing streak before quitting, there is a constant and consistent supply of them.

[gdaviet]

I like both of those last two points. I didn't think about the fact that nobody quits poker winner (except maybe Dewey Tomko ... but that brings back the idea that there are no FORMER poker players either).

I'm also glad to see that a larger sample yielded similar results. I had a feeling that 80k hands and 4k players would be enough for this type of distribution analysis, but hey, I'm no statistician so I don't know how to veryify that. 2.8 million though ... yeah, that should do it.

Side note ... how are you getting 2.8 million hands ... are you using PostGre with PT?

[zook]

In terms of "where the money we win comes from" we only care about the money put into or taken out of the system while we are a part of the system. Those players may or may not be winning players overall, but the important fact is that they have been losers while we (individually) are in the system. Therefore, they are where our money comes from.

Right, but what does that tell us? I'm arguing that this analysis doesn't tell us much about these people who give us their monies. Have they only played 5 hands or did we just happen to catch their last 5 in our sampling? I'm also arguing that the data is skewed because players tend to quit after losing.

I really think all we can conclude from this analysis is that a lot of players stop playing after they lose. But I'd love to hear what a statistician thinks...

Side note ... how are you getting 2.8 million hands ... are you using PostGre with PT?

Yes.

Here is where people get their 40% figure: at the stakes they play, 40% are winners, 60% are losers and the winners make a certain amount of money like 2 BB/100 or something

At a lower stake maybe 80% of the people are losers and only 20% are winners. So there maybe the winners win 10 BB/100 or whatever the number can be (rake is a factor)

So unless you played at all stakes including penny stakes your statistics only represent a sample of ONLY THAT STAKE!

FYI I heard the figure 1 in 20 are profitable players in their lifetime. I don't know where that figure came from and I can't verify it.

[flyingPenguin]

In terms of "where the money we win comes from" we only care about the money put into or taken out of the system while we are a part of the system. Those players may or may not be winning players overall, but the important fact is that they have been losers while we (individually) are in the system. Therefore, they are where our money comes from.

Right, but what does that tell us? I'm arguing that this analysis doesn't tell us much about these people who give us their monies. Have they only played 5 hands or did we just happen to catch their last 5 in our sampling? I'm also arguing that the data is skewed because players tend to quit after losing.

I really think all we can conclude from this analysis is that a lot of players stop playing after they lose. But I'd love to hear what a statistician thinks...

I was going to go on another little rant basically repeating what I said before - that it doesn't matter. They may have made a huge net profit over their playing life and left after a short downswing, but they are still, for the purposes of working out where our money comes from, losing players. I mean if you want to know the life story and psychology of the player, it's relevant, but not if you're just interested in money.

However, having another look at the stats:

50 hands or less.
Unique players: 2,161
Winners / Losers: 765 / 1,396 (35% / 65%)
Average win rate: -27.64 PTBB / 100
Total $ Won: -$26,320.70 (56% of all the losses)

Players with less than 50 hands: 8,065
Avg. winrate: -17.17 BB/100
Total lost: $56,293 (54% of all losses)

The number of ex-winners dropping out after a losing streak is going to be a factor of the number of players at the site, and not change over time (they were winners before you got there, you see them losing over 50 hands, they drop out). The fact that there are two quite different sample sizes here, with similar percentages, shows that an insignificant number of them are in this 50 hands category.

That is not to say there aren't a significant number of them in the 1000 hands category or whatever though.

[dev]

It doesn't matter that players leave while down.

Hypothetically, let's say a player goes thru 2 stages, at stage 1 they win, and at 2 they lose. You'll get an even number of people from each stage when you play. When a player does his losing doesn't really matter, and it's not corrupting the data here.

When a player is in stage 1 another player is coming in in stage 2.

Personally, I think a lot of players leave while up. Take me, for instance. Lately I've been grinding out FR games for tourney fees. When I'm not in 2 or 3 tourneys, I play those games. I used to play until a certain monetary goal was met (when I get to $1000 I'll cash out half and go get an ipod).

I know I'm not exactly your average player here, but there are all kinds of reasons for people entering and leaving games.

Some of the worst players I've ever met have gamb00led their way to a few grand in a $300 buy-in game and left because they were scared to death that they'd lose it. Those are the players messing with the stats.

The truth is that our money comes from where we do better than our opponents. They call too much with TP (about good enough to wipe your ass with), we make money. They overcall with draws, we make money. They call preflop raises out of position and fold like origami to continuation bets, we make money. It's mostly the less experienced players making these mistakes, so I think it's accurate to say the people that are playing online for the first or second time are the easiest targets.

[zook]

All I'm trying to say is that when I first read this thread it was an ah-ha, but now it's a duh. Let's look again:

Players with less than 100 hands: 11,106
Avg. winrate: -10.45 BB/100
Total lost: $78,265 (75% of all losses)

That's 68% of the player pool (16,286 unique players) accounting for 75% of the losses.

Players with less than 50 hands: 8,065
Avg. winrate: -17.17 BB/100
Total lost: $56,293 (54% of all losses)

49.5% of players accounting for 54% of losses.

Players with less than 25 hands: 5,099
Avg. winrate: -30 BB/100
Total lost: $34,086 (32% of all losses)

31% of players accounting for 32% of losses.

So the big difference is winrate, which I definitely think is affected by the lose-and-go-home mentality. If an average joe makes a $100 deposit and wins big immediately, I think they're likely to keep gambling, gradually lowering their winrate. If they lose quickly, like most of them do, their winrate is astronomical.

What I've learned is from this is that 1) most players don't stick around the same site/stakes very long (b/c half the players in my database have played fewer than 100 hands), 2) a LOT of players lose a LOT of money in less than 25 hands, and 3) these losing players tend not to show up at the same site/stakes again.

What have you learned?