Quote:
Originally Posted by DeathCabForTootie
There is another thread that discusses the folly of speaking in absolutes in poker.
Never playing 55 unless in late position is one of those examples. In tough games, you should consider folding 55 in EP. In soft, passive game where your opponents are fit-or-fold, you should consider raising pre and barreling favorable flops.
I agree. I think for GG's games, they're more often calling stations, though. This can still work out in hero's favor if they are often calling down with as worse as middle pair and lose to his set. But if it's often 6 to a flop and they're chasing their draws, hero has to be aggressive to give them incorrect odds to call.
Quote:
Originally Posted by gobbledygeek
Reviewing hands is obviously a good learning tool, and I of course beat myself up over key hands (both winners and losers) after each session.
However, I'm not convinced that tracking live hands / stats / etc. is going to build enough of a sample size to have any real confidence in what is going on.
An example. Skippy mentioned something on AboveEV one time that he sometimes (in some games) open folds hands like 55- in EP (something I must admit I don't believe I've ever done). So I thought, wow, being at ~2000 hours it sure would have been nice to have collected some data on hands just to see how profitable they actually are. For example, with 22 UTG, I would have been dealt it 2000 hours * 30 hands/hour * 0.5% (frequency of being dealt 22) * 1/9 (frequency of being UTG at 10 handed table that sometimes plays less) = ~33 times. I of course would typically only continue if flopping a set, so that means I would have flopped a set with it about 4 times (and of course all this is assuming I don't get raised out preflop). So my conclusions of playing 22 UTG over 2000 hours of live play would boil down to how profitable the 4 times I flopped a set with it were. As I've learned, it's pretty damn easy to lose with a set, and we'll typically lose our whole stack. Other times, we don't get paid off at all, or make a small profit postflop. And other times we stack someone. But obviously a sample size of 4 flopped sets of 22 UTG is lol and we'd never be able to come to any conclusion whatsoever as to its profitability based on these 4 specific results.
That's just one example, but that is sorta how I feel about the idea of tracking stats like this. The live game is such a long term one that I don't believe we'll ever track enough useful data. So basing conclusions on empirical results has always seemed a little meh to me. Which is, of course, a little disturbing to conclude.
GnolivedatabaseG
I mostly agree, but what about if we consider all the situations in which we call with 55- UTG and UTG+1? Those are slightly different situations, but not by that much. That would give us 4*2*4 = 32 samples of when you flopped a set. I suppose when you're UTG+1, you can get raised, but that doesn't happen often enough to change the number of samples that much.
I think 32 samples is not very representative but it's not totally lol either. I think if 28/32 of those times showed a large profit, and you assumed that the tables you played at the whole time were the same, that would give us a little bit of confidence that calling at the tables you play at is profitable. If only 3/32 showed a large profit and a significant were unprofitable, then you could get a little bit of confidence that calling at the tables you play at is unprofitable.
The most obvious problem is that your opponents and your opponents' cards and the board vary so much. Still, I don't think it's totally hopeless. With online data (of which I have none myself), I think you could gain just a bit more confidence with larger assessments.
In
this Wikipedia article, the "Statistics based on real online play" has some data. Obviously that's for a lot more than one player, which is good, but also bad because we don't know the quality of play of the average player in the sample (though I'm guessing terrible). Following the citation there, I found
this, which lists starting cards by position. Unfortunately, the # of hands is still too small to conclude anything! I'd love to see a Pokerstars database with this information, though.
Last edited by nimbleninja; 05-19-2015 at 03:58 PM.