Hey I know this isn't strictly a theory question but I was thinking just how big is the poker economy?

How many regs do you think there are playing in the population, and how many players overall (regs+recs). What's your guess?

I would define a reg simply as someone who spends an appreciable amount of time dedicated, focused time each week towards poker (say 15hrs per week).

Basically, a reg is anyone who takes the game somewhat seriously either as a genuine hobby or as a pursuit to make money.

And a recreational players is just anyone who isn't a reg.

I would guess there's something like 10k-20k regs out there and maybe like 100k people around the world who do at least some amount of poker activity. Do you think I'm way off?

I read somewhere that 100M people globally play poker but that doesn't sound right at all. I think that's way too high.

I was trying to find stuff like site-traffic data to do some guesswork but no avail.

Another thing I'm wondering is: what % of players are actually long-term winners? Has anyone ever done a study on that? I think I heard somewhere that only about 2% of players are actually turning a consistent profit, can anyone confirm that?

I wouldn’t be too surprised at the 100M number if it includes people who just get together for “penny ante”-type home games. The vast majority of people who play poker likely have never played at a casino, online, or in “serious” home games.

That's a good point. I guess I should rephrase:

"What's a good estimate for the number of players who take the game at least semi-serious in some formal manner (casino, online, serious home game, etc) i.e. a "Reg"?"
I think that's a pretty reasonable definition for a reg?

My personal guess would be like 100k-200k people who satisfy that definition. Do you think that's too high or too low?

In today's environment for MTTs (online especially), what do you think is the cumulative distribution for how many regs (players who take the game semi-seriously in some sort of formal way) achieve each of these ROIs? (My personal guesses included below).

1. <=0% ROI (I guess 98% or so of regs never do better than breakeven?)
2. 0%<ROI<=10% (I'd guess maybe 1% of the population of regs ever does better than 10%. Agree or no?)
3. >=20% ROI (my guess is maybe something like 1/500 regs can achieve ROIs that high. Too high or too low do you think?)

Here is some analysis for you regarding 5nl and cash game players.

credit to tombos21 who also posts on 2+2 btw.

https://www.reddit.com/r/poker/comme..._44m/?sort=new

Here are his bullet points basically:

Thanks for the credit Brokerstars

This graph is my favorite. It's a win rate distribution of 4.4M hands at full ring 5NL from a few years ago.



So, it's hard to define "reg" or "rec", but if you just look at overall results:

The top 10% of players took 92% of the profits (after rake).

The bottom 1% of players supply 20% of all money invested into the pool (before rake).

Poker is predatory like that.

The second conclusion seems pretty far fetched. The graph would smooth out a lot if you had true EV.

The graph would probably look pretty similar if you got the data from a fixed edge game like roullete. The players winning the most and losing the most are just luck outliers.

Many uncertain datapoints can still be useful to understand the overall shape.

I agree that it would flatten out. Here's the same graph, filtered for players with at least 2000 hands. Their win rates are more accurate, and therefore flatter.

However, this filter introduces bias, as filtering out low-volume recs makes it look like more players are winning.



It's a trade-off between accuracy and bias. Heisenberg's winrate, if you will.

lmfao, wp

Those graphs are cool, though I will say I was more interested in MTTs and also I would rather see raw winrates and not all-in adjusted winrates.

So if I'm reading this correctly, about 40% of ALL players in the 5nl pool (which has obscenely high rake, no?) are at least breaking even, and about 50% of all players with at least 2k hands played are at least breaking even?

I must say those results are quite surprising. I would have guessed something like 98% of players are long-term losers. An extremely skewed distribution of winrates with almost all the mass of the distribution below the breakeven winrate.

But I guess long-term results aren't necessarily going to be captured in your dataset? What's the average # of hand played by players in your dataset, and how many players in the dataset have more than 25k hands?

Also how many total players in the dataset, and what's the timeframe on it? I would say a rough OK definition of a reg is someone who plays like 2k hands/month. What proportion of players in your dataset achieved an average # of monthly hands played of at least 2k--I think that would be a decent proxy to one of my initial questions which was basically "how many regs are there?" and "what proportion of the population are regs?"

Wouldn't the distribution get more skewed, not less, if we were looking at true EV?

The true EV, if it were even possible to obtain, would be the actually correct measure of skill and performance--these factors are distributed highly unevenly across each individual member of the population.

On the other hand, using observed winrates introduces short- and medium-term rungood or runbad into the mix which smooths out the observed winrates.

2% for long term winning regs seems about right, or might be even too high a guess IMO. Depends a bit on what constitutes as a poker player and in what time frame? If you take, say, every single person who's played even just one hand of poker during the last 20 years, naturally the amount of long term winning regs is very small in that field. But if you don't include randoms then the relative amount of winning regs is higher. Depends a bit on the POV.

I personally play NL5 to NL20 and looking at my database of about 500k hands, there isn't a huge amount of players I would consider consistent, long term winners. Most either lack the skills or the endurance. Being technically good at poker means nothing, if you don't put the hours in it to actually turn those skills into money. And likewise, being mediocre can make some decent money, if you grind like crazy. There's this crazy opponent in NL5 on the site I play. Been there forever, playing very tight and passively, having a winrate of +1,8bb/100 over 63k hands. But he plays 10+ hours a day, 10-20 tabling. If he's not a bot I raise my hat for him, he's like setmining NL5 for a living.

Maybe smooth was poor word choice but 100% of the big winners and losers in this graph are luck outliers. It's probably like <.1% of players with a winrate >20BB/100, but the graph shows 5% of players are winning at that rate. I would also say 90% of the "winners" in this graph are long term losers on a heater.

I feel as if some math tricks should be possible to get a more accurate representation of the true win-rate distribution, but not sure how without a lot of thinking about the problem. Someone strong in statistics could probably tell us right away. There is a lot of unused information/variables. For example, analysis of the data would likely show us that players with fewer hands have lower win-rates on average. Could you use some Bayesian statistics to incorporate that information into this graph or am I missing something? And do the same with all the other variables.

There are undoubtedly more sophisticated methods of analyzing this data. I've never thought about adjusting the winrate based on hands played- that's an interesting idea. You'd probably need some kind of correlation coefficient. Unfortunately, I only have a college dropout surface-level understanding of statistics. The best I could do was implement 1-sigma error bars. The data is publicly available here if anyone wants to analyze it.

I disagree that only 2% of players are long-term winners. The rake can be thought of as a force that pulls the entire graph down vertically. 11.7bb/100 is a pretty big downward shift, but more than 2% of players should have a "true" win rate higher than 11.7bb/100 at 5NL. Furthermore, players who win at 5NL tend to move up to higher stakes quickly, whereas a lot of losing players will stay at 5NL because it's so near the lowest available stake. Therefore, losing players should be somewhat overrepresented in this sample.

One experiment I've thought about doing was to track player's results across many different stakes on the same site and use that to create some kind of skill distribution for each stake. You could empirically compare the difficulty levels of each stake. For the moment it's just an abstract idea tho.

2% of all players, not 2% of those who play a lot. Most players play like a few dozen or hundred hands in their entire "career". Like my sister for example, who only played a few tournaments and of course luckboxed one of them and won 1500€. She's technically speaking a winning player since her lifetime earnings are positive, but she's not a reg. Neither is she a good player who could consistently make a profit so I wouldn't call her a winning player per se.

I'm fairly certain most players fall into that category. Try poker a bit, some lose, some win. Some might even win millions by luckboxing WSOP. A few of them might get the bug and actually study the game and become consistent winners. But very few do that or even have what it takes to play well for years and win consistently and also study the game constantly to improve their skills.

Also note that multitabling distorts the bias a bit. Winning regs usually multitable, while recreational players play just one or maybe two tables. So, many seats contain a winning player, but the amount of winning individuals is still quite small. The tables might look like they're full of winners, but actually they are just full of winning seats.

Also some sites allow you to change your table name. On the site I play at you can do that once a month. That can distort the bias even more, as it's quite likely many of the regs in your database are actually the same person. I know some regs change their name every month, as I can easily identify their playing styles (plus some of them don't bother to change their avatar when they change their name).

Yeah I guess maybe this topic is a bit more theory-oriented than I initially thought

oh i remember reading this awhile ago, didn't realize tombos wrote that. very cool