Friday, September 18, 2009

Thoughts on why low limits are virtually unbeatable

I've been tutoring Riad the past six months or so and he's been asking me repeatedly why he can't beat the games he's in - he plays $2-4 and $3-6. He's also asked me whether I think I can beat the games he's in and my answer has been always been I'm not sure but probably not. I'm revising that answer to definitely not unless stringently game selecting or exclusively playing heads up and 3-handed.

We had a chance to look at his pre-rake numbers today and he was about a 3 big bet per hundred winner. His actual, post-rake numbers are about a -1 BB/100 loser. So rake at the limits he's playing is around 4 big bets per hundred hands which is obscene. If I'd been paying 4 big bets per hundred in rake at the limits I'm playing, I'd be down millions.

I can only imagine how many thousands of people have tried to become poker players and quit out of frustration convinced they didn't possess the skill. If they started out at lower limits with a limited bankroll, as most probably do, and practiced good bankroll management skills by not playing higher limits unless adequately rolled, they very likely busted. Or just quit out of frustration.

A lot of people who have survived as poker players probably took shots at higher limits when their bankrolls dictated they shouldn't have (myself included). Some of these players likely survived by unintentionally moving from unbeatable to beatable games. I find it interesting that rake, in a way, causes good players with bad bankroll management skills to progress through the limits.

My experience has been that the skill level difference from limit to limit isn't significant, especially in the current poker environment. There is a difference, but it's a small one. We're probably talking fractions of a big bet if it could be measured. But the difference in rake is significant. As much as a big bet or more depending on the limits and site.

I should mention that this all applies to limit hold-em. I'm not sure what the rake impact is at no-limit but I know much higher win rates are attainable.


  1. This was huge for me playing PLO. If you look at 50/1 and 1/2 the rake is enormous. When you get up to 2/4 it lk 2.9ptbb/100 and like 2.1ptbb at 3/6.

  2. Hi Tony,

    Dylan here again. I've decided to read your blog from start to finish and came across this post which hit a personal chord, since the majority of my nearing 1mm hand database has been 2/4 and 3/6 (remember all the propping I was doing). Here are some stats broken down for just those stakes, with my conclusions below.

    When I was propping on Merge years ago, the rake was insane:
    2/4: 4.4BB/100
    3/6: 3.1BB/100

    During those years, I indeed broke about even over 85k hands:
    Result: -.23BB/100
    Rake Paid: 4.3BB/100
    Pre-Rake Result: 3.8BB/100

    Late last year Merge reduced their FL rake, and I put in about 190k hands:
    Result: .79BB/100 (tougher games!)
    Rake Paid: 2.6BB/100
    Pre-Rake Result: 3.1BB/100

    My combined lifetime 2/4 and 3/6 stats over 375k hands:
    Result: .91BB/100
    Rake Paid: 3.1BB/100
    Pre-Rake Result: 3.7BB/100

    Further breaking down the combined lifetime hands by number of players:
    2-3 handed (82k hands):
    Result: .21BB/100 (ran pretty bad over 10k hands of 2/4 HU @ 4.6BB/100 rake paid!)
    Rake Paid: 3.5BB/100
    Pre-Rake Result: 3.8BB/100

    4-6 handed (240k hands)
    Result: .98BB/100
    Rake Paid: 3.2BB/100
    Pre-Rake Result: 3.7BB/100

    7-10 handed (53k hands)
    Result: 1.67BB/100
    Rake Paid: 2.3BB/100
    Pre-Rake Result: 3.6BB/100

    Conclusions: Yes, you might have a bigger edge over your weaker opponents when it gets down to 2-3 handed, but the rake gets extraordinarily expensive for the stakes, to the point where it almost doesn't matter how good you are. As the number of players increases, rake paid reduces, and thus winrate increases (plus, variance isn't as brutal). The downside is it takes much longer to get the volume in.

    Broken down the sample sizes are indeed pretty small, but it does give you a ballpark figure of how brutal the rake is, and how it gets a little easier as the number of players increases.

    In the high stakes world, you are paying the same rake regardless of how many players are at the table, since it's capped for pretty much every pot.

    Now, comparing all this data to 80k of mid stakes games I've played (10/20 to 20/40):
    Result: 2.27BB/100
    Rake Paid: 1.4BB/100
    Pre-Rake Result: 3.6BB/100

    That pretty much speaks for itself! Looking forward to the days of playing 15/30+ again, when my regular play isn't "catastrophically affected by the rake" as you put it.

    Thanks again for your amazing blog and all the perspective you've given me.


  3. No problem Dylan, glad you enjoyed the read. And thanks a lot for posting this information here, a lot of good info.

    One observation, your prerake win rate is nearly identical for all numbers of players. This is really unusual and surprising. It could be due to variance. Just in general the less number of players, the bigger your edge should be because you're making a lot more decisions on average.

    I didn't realize the games were raked that much higher though as the number of players lessens, I assume Stars is the same way. Great info, thanks again.

  4. Hmm, I didn't realize that would be rare. I just figured my "pure" winrate was consistant and that was a good thing. Maybe it's that I'm much less proficient at 2-3 handed play than I thought!

    Here then are the pre-rake figures for all of my FL play, about 670k hands, ranging from .5/1 to 20/40:

    2-3: 3.32 (167k hands, 2.76 raked, .50 net) SD=20 (62/45/2.4)
    4-6: 3.43 (427k hands, 2.97 raked, .87 net) SD=18.5 (31/22/2.4)
    7-10:3.98 (77k hands, 2.42 raked, 1.95 net) SD=15.6 (21/12/2.4)

    Total: 3.47, 2.86 raked, .91 net

    It's pretty consistant with the above numbers, albeit fanning out a bit more from the center.

    As you can see here and above, subtracting what I raked from the pre-raked winrate does not always equal my net winrate. It might have something to do with the raked stat coming from a custom stat for PT4 calculating it from MGR. Another odd thing is it seems I've raked more per hand in 4-6 play than 2-3 play.

    Regarding my low pre-raked winrate when it gets short, I suspect I'm naturally stronger against loose opponents when their biggest mistake is playing too loose. Once it gets short, either the overall quality of my decisions aren't as strong compared to my opponents', or it's variance as you so kindly suggest. :)

    Anyway, I totally forgot to mention why I wanted to respond to this post in the first place. I am living proof that these "unbeatable" games are beatable! As long as the rake doesn't surpass 3BB/100 by much, it's totally doable if you game select and play well.

    In fact, over my last 50k hands I'm doing exceptionally well, despite paying a bit more rake:
    4.69, 2.89 raked, 2.04 net