What percentage lose at online poker?

[Tommy R]

I recently posted an article "Are you an Expert or Novice Player", and quoted Mason as saying 95% of online players lose. Mason has pointed out that he never said this and I apologise. The number is not essential to the article but I can't remember where I saw this 95% figure.

My question is does anyone have figures on the percentage of online players who lose in the long run?

[PokerPaul]

I have no data to back this up, just going on a hunch here, but 95% sounds too high.

I figure its closer to 75%.

I have 4 friends who play online, and i have talked to numerous people at BM cardrooms about their online poker play.

Based on everyone's responses, I'd say that 50% say they break even or make money. Like it or not, i apply a fudge factor to that, and figure its closer to maybe only 25%. Hence 75% online loser rate.

I myself am about break even (honestly, no fudge). Although the last 3 weeks have been a series of brutal river beatings.

[Jimbo]

Tommy,

I believe it is safe to say unequivocally that all of the people lose some of the time, some of the people lose all of the time and many of the people lie about whether they are winners or losers most of the time.

[beernutz]

Just a datapoint: My Pokerstat database has about 10,000 hands most of which were in some form of Omaha/8 at limits from $.5/$1 up to $10/$20 but mostly at $2/4 (Pokerstat is used to store and analyze Paradise Poker hand histories). It shows 35% of players I have played are net winners, 63% are net losers, and 2% breakeven. That rake is hard to beat.

[Tommy R]

That surprises me. You have a resonably large sample as well. Anyone else with this type of data?

[Jimbo]

Tommy,

It surprises me that you of all people would first of all consider 10,000 hands a reasonably large sample and second not querry how many different players were in his sample and how long each player played. I believe if you take a snapshot of online games at time 1 and another snapshot at time 2 you could reasonably have 100% winners. Since due to the rake we all know this is impossible so the variable of players entering and leaving drastically affects the data.

In order to accurately compute the percentage of losers, winners and break even players I believe a close estimate of the total number of players would be first required.

I think the 95% is too high. There is an extremely high percentage of players who really try to be winners online. Much higher, I think, than in B&M casinos and cardclubs. It's in the nature of online poker that money will always be of utmost importance, whereas live poker can serve much better as a recreational activity or a hobby.

As a result, I think online poker are made up by several small winners and break-even players. I think the losing player figure can be as low as 60-70% percent.

lars

[PokerPaul]

U must have this kind of data and can run an easy query to compare all players cashin vs. cashout + BR.

However, i also understand this might be something online cardrooms want to keep confidential.

After all, its not really great advertising to confirm findings that a large percentage of players will lose.

However, if the online loss rate is high, i'll bet its even higher in BM cardrooms, just beacuse rakes are even higher, plus tips for dealers, tips for cocktails etc., plus bonus jackpot dollar.

Scary....

[beernutz]

Jimbo, I don't think this sample represents conclusive data but it does agree closely with what some of the other people using Pokerstat have posted about their databases.

Factors to weigh when considering this evidence:

Only samples from Paradise

Only has Omaha/8 games (you have to use a separate verion of Pokerstat to analyze HE)

10,000 hands is not a huge sample but it is not insignifigant either. Do you think the win/loss %'s are going to change dramatically by doubling the sample size. I can tell you that the numbers I found when my database was half that size were within 1% of what they are now.

This sample includes over a thousand different players (assuming no nick changes) IIRC. This is not a snapshot but rather a longitudinal capture of data from various players around a single player.

Obviously since I am in every hand of the database my playing ability would affect how you would view the sample--if I was really good, then the % of losing players would be increased and vice versa.

Since there is no way to accurately measure the winning/losing %'s for online poker players save every single poker site suddenly opening their books (or databases) to the world, player statistics such as the ones I've posted are the best estimate we're going to get, IMO. If you've got better numbers I'd love to hear them and, of course, how you came by them.

i tend to think of my place the in pool of players in percentile terms. i know i'm a 'winning' player cuz i have the records to back it up. but at nearly any table i sit, there's at least one other player whom i consider a peer or better than me.

the accepted measure of success is BB/hr [for limit play] and using the B&M measure of 25 hands/hr i would estimate [from the seat of my pants] :


99.6 percentile = 2 bb/hr + = true pros who probably prefer large limit live play

96th percentile = 1-2 bb/hr = very good players, good enough to make a living, you'll find larger % online than in B&M simply cuz of convenience.

85th percentile = .5-1 bb/hr = regular who can provide part time income for him/herself.

50th percentile = (-.5) - .5 bb/hr = where most regulars fall but probably think of themselves as winners cuz they don't keep good records. most regulars in B&M fit here

20th percentile = (-3) - (-.5) bb/hr = true novices and the clueless. IMO they don't last long.

3rd percentile = > (-3) bb/hr. = true fish, probably won't see more than once.


anyway, interesting discussion

[Jimbo]

Beernutz,

I was not disputing the accuracy of your numbers at all. I was simply stating that they are not relevant to the total percentage correlation of poker losers vs winners. If you have 10,000 hands and 1000 players that proves to be even less valuable than I had assumed. An average of 10 hands per person is certainly too small of a sample to determine whether someone is a winner or a loser. Don't you agree? As far as a reasonable number if it takes 2000 hours in a B&M room to be 95% sure you are a winning player then it seems fair to divide that by 2 for online. Then we multiply the result by 35 hands/hour and then again by 1000 different players and it looks like you need 35 million hands. This is a far cry from a mere 10,000 hand sample that you provided. There may be a flaw in my logic, feel free to correct any estimates that you feel I may have exaggerated.

[MS Sunshine]

Without anything to back it up, I feel, this post, has the best handle on the breakdown of players.

MS Sunshine

[Tommy R]

I take you point Jimbo, 10,000 datapoints and 1000 players loses plenty of degrees of freedom. I did drop the ball a bit on that one. But this sort of data gives a bit of an idea.

Another interesting factor that has not really been discussed is if the % figure would be higher or lower for Hold'em and 7-card stud. My initial reaction is that the rate would be much higher for 7-card stud becasue of the complexities of the game. I guess the Hold'em rate might be about the same as the Omaha rate, without any concrete fiqures.

[Tommy R]

In defence of David, I think it would be a bit of a confict of his interests to post this fiqure :-).

It would be good to have this number though.

[BruceZ]

I agree this data is meaningless. Over a small number of hands, a person will have close to a 50% chance of being a winner regardless of how big a winner or loser he actually is in the long run. So you would expect about 50% winners, though it may be even less (which it is) since over 10 hands you are more likely to lose money than win. You need enough hands per player to have a small chance of misclassifying him.