Schooling

#1 10th December 2007, 5:52 PM

From this thread:

Quote:

Originally Posted by Cheetah

Actually, excluding the natural bias that may exist, some forms of poker are subject to more luck than others, especially against not very good opponents.

All limit versions of poker are very succeptible to "fish schooling", where many bad players play incorrectly, but collectively, their bad plays are less bad because of the odds they create for each other.

A good player's edge is greatly diminished in such games since the odds of winning are much smaller than a table with say just a few fishies. That translates into much bigger variance(even for passive tables, let alone wild tables). And big variance reduces your profitability when you apply BR management.

If you play with very good players, then I think the various forms of poker are much closer with respect to skill level requirements.

Refutation of schooling here and here.

#2 10th December 2007, 6:26 PM

Interesting Article. I have never heard of schooling before. I can see why it would be an compelling argument for a person that keeps loosing to poor players. The articles do a good job of explaining why that isn't the case however, except in very rare instances. I mean really how many times are you going to get 9 callers on the turn.

Thanks AG, I always love the links you provide to articles. I may not agree with them but that are always informative.

#3 10th December 2007, 7:15 PM

Although I disagree with the idea of schooling being something that cuts into winrates, Cheetah doesn't seem to be primarily saying that, but rather that high variance reduces winnings for people with strict bankroll management. However, and I believe this is key, high variance only reduces profitability of it's high variance but same EV. I ran quite a lot of simulations for this at one point, and unless I made the difference in EV almost 0, it was always better to choose the +EV, high variance game for profitability.

#4 10th December 2007, 10:02 PM

Quote:

Originally Posted by aliengenius

From this thread:

Refutation of schooling here and here.

While I agree with the author, it's very different in NL hold 'em. In limit, the max you'll lose after being outdrawn is 4 big bets on the river. In NLHE you could lose several times the pot size. What happens if for example you have AA and there are 5 people with draws. Paying one off when he hits is obviously -EV because if they all miss they most likely will fold to your bet. What happens when the river comes and you face a bet? It's possible it's a top pair that thinks it's good, it's possible that it's a bluff, and it's possible that they hit their set or gut shot. It's also possible that they already had a made hand (flopped a set or something). Even the worst players still make hands (and bet them). I'm pretty sure having lots of bad players in the game would still be +ev and increase the variance, but this article doesn't prove that, it only addresses limit, which is far different because the bet your making on the turn to price out draws is the same as the bet size on the river. That's not the case in most NLHE games.

#5 11th December 2007, 11:36 AM

At last AG has found me the `Holy grail` or `The meaning of (poker) life`!!!

I`ve been trying to put a similar point across but not as well

#6 12th December 2007, 9:43 AM

Quote:

Originally Posted by aliengenius

From this thread:

Refutation of schooling here and here.

I only read the first article and it is anything but refutation.

In order to investigate this issue, one must consider all possible hands and a variety of players.

The author only considers maniacs who play any two hands and only two hands. On top of that, one of the hands is AA which is probably the worst hand to show fish schooling.

Despite these special circumstances, in the author's admission, he could observe schooling effect, but chose to label it small and disregard it.

He simply doesn't have the data to draw any conclusions. He has basically used 2 points: one type opponents against AJ or AA. One cannot extrapolate from this to all hands and all types of players. This would be like saying that since Cheetah was in his den on Sunday at 1pm and Friday at 1 pm, therefore he is always in his den.

As FP mentions, there are also the additional effects from increased variance that may not be compensated enough by the increase in profits.

I am too busy right now to investigate this myself and it is not a priority for me. However, a starting point would be to run all possible hands against several types of players, say 5 types.

Even that is not sufficient since post-flop play is very important in cash games and early in tourneys. When using showdown simulations, we are neglecting, among other things, the huge reverse-implied odds of big pairs against multiple callers.

FP, what kind of simulations have you done? I would be interested to see the results and any conclusions you have drawn from them.

#7 12th December 2007, 12:09 PM

Without going into too much detail, from my point of view it helps with bad beats.

When you constantly make the correct move and get sucked out it can make you mad. However, I am now of the opinion where the situation described is not a big deal to me now.

If I`m sitting with AA I will make a semi standard raise. If a lot of ppl want to call, so be it. What I do know, however, is that at that time I am the fav to win the hand, all be it if overall I`m less than 50% v 4 other callers. So I have made the correct move and I have made a lot of people made the wrong move.

Then you make your decision on the flop. The good news is, if your hand improves on the flop and u r sitting with the nuts, because the other ppl are looking at their holding and their holding only, u can then push and u will probably get a caller, or at worst win a bigger pot than u would normally have.

#8 12th December 2007, 1:47 PM

Quote:

Originally Posted by Cheetah

FP, what kind of simulations have you done? I would be interested to see the results and any conclusions you have drawn from them.

My simulations were made to try to answer the question "how bad is variance for someone with a small bankroll". Basically, I stipulated the following

a) I will beat my opponent for an amount that is proportional to the stakes I play, and
b) with higher variance I'm more likely to go on swings that will force me to move down in stakes and therefore make proportionately less money until I'm back up to where I started.

Since moving down effectively halves my income (stakes are usually like that) the idea is (was) that the safer route might generate more money since it will not be at all as likely to risk moving down.

So I decided to test how variance/EV effected money made. The simulator basically runs random all-in coin-toss situations where one guy will bet money on every toss that's 50% or better to win (pushing the edges, and the variance, to the max) and the other guy will be more careful, only taking, say, 55% edges or better before getting the money in.

I did not manage to run any simulation over any meaningful period of time where it wasn't better to take the +EV, high variance route in terms of earning money.

Of course, this is a very rough model. My code doesn't properly account for rake (although it does to some extent) which hits the high variance player harder. On the other hand, it doesn't consider pot odds at all either, which works in favor of the high variance player, so perhaps those two simplifications somewhat cancel each other out.

Another, very real, problem with the model is that people aren't machines, and running a high variance game may be psychologically detrimental. But the pure mechanics of variance vs. EV speaks strongly (very strongly) in favor of EV.

#9 12th December 2007, 5:58 PM

Quote:

Originally Posted by Cheetah

I only read the first article

You really should read the second one as well (it's really one big article, imo)...

#10 12th December 2007, 6:16 PM

That's one of the best pieces of information that I have seen. If you want to see it in action look at the first or second hand of almost any large free roll. You would see most of the table if not go all in for the first or second hands.
Then a hand like Q3o wins with quad 3's.

#11 12th December 2007, 6:24 PM

Quote:

Originally Posted by rufcut68

That's one of the best pieces of information that I have seen. If you want to see it in action look at the first or second hand of almost any large free roll. You would see most of the table if not go all in for the first or second hands.
Then a hand like Q3o wins with quad 3's.

Clearly you didn't read very far Mr. ADD interwebs user.

#12 13th December 2007, 1:05 AM

Quote:

Originally Posted by F Paulsson

My simulations were made to try to answer the question "how bad is variance for someone with a small bankroll". Basically, I stipulated the following

a) I will beat my opponent for an amount that is proportional to the stakes I play, and
b) with higher variance I'm more likely to go on swings that will force me to move down in stakes and therefore make proportionately less money until I'm back up to where I started.

Since moving down effectively halves my income (stakes are usually like that) the idea is (was) that the safer route might generate more money since it will not be at all as likely to risk moving down.

So I decided to test how variance/EV effected money made. The simulator basically runs random all-in coin-toss situations where one guy will bet money on every toss that's 50% or better to win (pushing the edges, and the variance, to the max) and the other guy will be more careful, only taking, say, 55% edges or better before getting the money in.

I did not manage to run any simulation over any meaningful period of time where it wasn't better to take the +EV, high variance route in terms of earning money.

Of course, this is a very rough model. My code doesn't properly account for rake (although it does to some extent) which hits the high variance player harder. On the other hand, it doesn't consider pot odds at all either, which works in favor of the high variance player, so perhaps those two simplifications somewhat cancel each other out.

Another, very real, problem with the model is that people aren't machines, and running a high variance game may be psychologically detrimental. But the pure mechanics of variance vs. EV speaks strongly (very strongly) in favor of EV.

This seems to be partially related to what I am currently doing on BR management. If you have your numbers from these simulations handy, I would be interested to see them. Perhaps in a different thread to avoid hijacking this one.

One more note. I think that the ratio EV/variance is more relevant than using either one separately, regardless of how we define variance. Of course, a ratio is very simplistic and the only way to truly account for their effects on overall profitability is to link them to BR management. This is what I am currently trying to do by avoiding any definition of variance.

I wished there was a sub-forum for math stuff where we could post formulas and numbers without fear of decreasing the CC membership due to premature death from excessive exposure to numbers.

#13 13th December 2007, 1:07 AM

Quote:

Originally Posted by aliengenius

You really should read the second one as well (it's really one big article, imo)...

I will AG. I just wanted to review the first one while it was still fresh in my head.

Similar Threads for: Texas Hold'em Poker > Schooling
Thread	Thread Starter	Forum	Replies	Last Post
How to combat the fish - schooling	Ronaldadio	Cash Games	9	16th August 2008 5:50 PM
The Table Selection Game!	c9h13no3	Cash Games	7	28th July 2008 8:12 PM
how to laydown KK	neil619	General Poker	13	27th January 2008 7:06 AM

Play Texas Hold'em Online Poker \| Schooling
BetOnline Poker is a top online poker site accepting all US players with an INSTANT cash bonus of up to $900 FREE. It also has a casino and sports betting available.	PokerStars is one of the best places to play poker online that accepts US players. Use PokerStars marketing code CCHAT for a $600 bonus. Note: No US Players Accepted.