The paradox of proper play

[deucesevenoff]

Thanks everybody for the great responses...I definately recognize that there's junk and then *junk.* I guess the point I'm trying to get at is that as you move up in limits the cards become less important and reading people becomes more important. The only way to do this is to practice. Though profitable at the low limits, it is very difficult to improve if you're always (and only) going in with the nuts. If somebody tries to just play ABC poker at one of the high stakes games I think that the players up there would catch on pretty quickly and slaughter him/her. Guys like Barry Greenstein play with the same deck we do, get the same number of good and bad hands, and suffer the same beats yet can have stellar results. Yes, I know that play at 2/4 is much different from 2000/4000 but I would assume that any serious student of the game is always looking to improve their play, move up and make more $$. Correct me if I'm wrong?

So maybe paradox was the wrong word to use, but I feel as though I'm missing out on something by only playing premium starting hands. (This kind of discussion might be more appropriate for NL games rather than limit since if you sense weakness you can push somebody off a hand).

As an aside, there is a great discussion going on about a similar topic in the high stakes no limit forum titled "Don't players make moves? Is bluffing non-existant?"

Here is a link:

http://forumserver.twoplustwo.com/sh...14&fpart=1

[OrangeKing]

[ QUOTE ]
Though profitable at the low limits, it is very difficult to improve if you're always (and only) going in with the nuts.


[/ QUOTE ]

I think you might have a misunderstanding that a lot of people who start posting here do: nobody is advocating always and only going in with the nuts. Nobody here is suggesting weak-tight play, which seems to be what you're talking about. You should definitely be agressive, and definitely be willing to push people with hands when you don't hit the flop (in the right circumstances, of course) and occasionally, it's even correct to play with junk (blinds steals, for instance). But the vast majority of the time, especially in limit, you need to stick to a solid preflop strategy. If you're playing 20% of your hands (and probably seeing 25-30% of the flops, including the blinds), you won't be predictable. If you're playing 10%, then you're way too tight and turning into a nuts-peddler.

[deucesevenoff]

Thanks...I think (know) I have the problem of being a weak-tight nuts peddler. Its something I'm trying to work on, so please bear with me here (frequently I end up seeing 15-20 percent of flops including my blinds and that's an optimistic estimate).

So if I had to come away from this thread with anything, would it be appropriate to say that changing gears does not involve decreasing the quality of the starting hands you play, just how you play them?

[A_C_Slater]

Changing gears simply means occasionaly raising with something like 98s or 76s from LP after a few limpers. Hold em Poker For Advanced players even advocates occasionaly raising with these hands from early position first in, to throw off opponents. It should also be done if your preflop raises are getting too much respect and you are only winning the blinds after your UTG raise (nearly unthinkable in today's small stakes games, thus changing gears is largely uneccesary.)

This should only be done against aware players whom you play with regularly. This should not be executed against unknowns or bad players. They simply will not think anything at all of your 76s UTG raise.

Changing gears does not mean raising with J2o in a limit hold em game. Middle suited connectors or one gaps like T8s, J9s, etc. A J2o raise would not be as bad (though still terrible) in a NL or Tournament No limit game because those are games that are more based upon proper wagering skills. And only if you are open raising or executing what Dan Harrington would refer to as a "squeeze play" when you expect your opponents are routinely raising with non-standard hands preflop.

But since the wagering limits are fixed in limit then routinely raising with Q4o to "throw opponents off" in a limit hold em game is suicide and not even the best player in the world could overcome this preflop blunder.

My previous comments were directed at limit HE cash games.

[BluffTHIS!]

[ QUOTE ]
My previous comments were directed at limit HE cash games.

[/ QUOTE ]

Mine however were primarily intended for big bet games which is virtually all I play, and where getting paid off is more critical money wise than in limit.

[Wally Weeks]

[ QUOTE ]
Hey Wally: nice answer, but I'd have to quibble with your definition of the word "optimal". My understanding of this term is that it does indeed refer to a strategy or solution that scores the absolute maximum on some objectively verifiable criterion.

I think that in your discussion we need to subsitute a word like "adequate" or "reasonable". Herb Simon, a big fan of heuristics, came up with the term "satisficing" when reaching around for the right word in a similar situation.

[/ QUOTE ]

Hi jtr,

Your quibble is quite legitimate. I think I have a bias toward speaking in a more concrete way even if it is inaccurate as in this case. While the concept of satisficing is popular in concrete science disciplines like computer science artifical intelligence, it seems to be pretty popular amongst the social sciences such as economics where I have seen it referenced more often.

For fun, I decide to open up my old college AI book as a refresher since I remember really being introduced to the concept in AI:

[ QUOTE ]
Without heuristics, we would become hopelessly ensarled in a combinatorial explosion. This alone might be a sufficient argument in favor of their use. But there are other arguments as well:

Rarely do we actually need the optimum solution; a good approximation will usually serve very well. In fact, there is some evidence that people, when they solve problems, are not optimizers but rather are satisficers [Simon, 1981]. In other words, they seek any solution that satisfies some set of requirements, and as soon as they find one they quit. A good example of this is the search for a parking space. Most people stop as soon as they find a fairly good space, even if there might be a slightly better space up ahead.

[/ QUOTE ]

Satisficing is a better concept than optimizing since you cannot really measure to any reasonable accuracy what the actual best solution is. Oftentimes in a computer algorithm you can tweak or iterate (loop) more to get asymptotically closer to a more accurate solution. With the chaos involved poker this hardly seems to be the case (i.e. math, psychology, logic, etc.).

But from what I understand in poker in academia, (almost?) unbeatable heads up bots have been developed. I believe that linear optimization plays a large part in the computation of a strong bot's decision making process. But I could be wrong as I have not closely looked at the academic literature. Perhaps all of the variables are not too much to overwhelm current computer technology.

Regards,
Wally

[jtr]

Cheers for the reply, Wally.

My quibble was probably a bit anal: I completely accept that most people use the word "optimal" in a non-technical way to actually mean "pretty damn good strategy", and there's nothing wrong with that.

I like the car-park space analogy. In playing low-limit poker against human beings I think we're in a similar position. If we've hit on a way of playing that allows us to make 3 BB/100 against a certain set of opponents, in practice it may not be worth all the extra mathematical and statistical work needed to up that winrate to 3.5.

Good question about the academic work on optimal heads-up bots. Certainly I guess this is an arena where a truly optimal strategy could be worked towards, although it would be subject to the game theoretic limitation that one assumes fully rational play on the part of the opponent and then tries to make sure that your own strategy has no leaks. If, like most real people, the opponent has some non-rational preferences, like slightly over-valuing pocket pairs just because they like the look of them, then we need to use something other than classical game theory to help us maximize our performance against this guy.

I took a look at some of the academic poker literature (Uni of Alberta group mostly) a while back, and yes, it seems their heads-up bots were becoming very good, but I can't recall whether the authors made claims for optimal performance in the strict sense.

[ACW]

If you want to experience what changing gears is really all about, play a 3-handed limit game. Changing gears is vital against decent opponents in that game structure.

[Wally Weeks]

[ QUOTE ]
Cheers for the reply, Wally.

My quibble was probably a bit anal: I completely accept that most people use the word "optimal" in a non-technical way to actually mean "pretty damn good strategy", and there's nothing wrong with that.


[/ QUOTE ]

The word optimal also sounds better than satisficing and "mo bedda." [img]/images/graemlins/smile.gif[/img] For most intensive purposes the term works and is less technical than satisficing. Most poker literature is written for individuals with limited knowledge in such advanced mathematical concepts. Besides, for most people understanding these sorts of things don't really increase expectation not to mention the what-the-hell-I-just-wanna-learn-to-win factor. I can only imagine the complaints and criticism that Mason got with his Gambling Theory and Other Topics text with it's sigma notation.

However, I for one believe that the more solid the foundation in one's understanding the better a player can become in the long term. Being a good player at one game such as hold 'em is boring anyway.

[ QUOTE ]
I like the car-park space analogy.

[/ QUOTE ]

Interestingly enough, I looked up some history of the concept of satisficing. The reason why I found so many references of satisficing with economics is partially due to the fact that Herb Simon was awarded a Nobel Prize in economics. (Uh, derrr...cough.)

Simon, who died at 84 in 2001, was interested in decision making processes which naturally led him to the fields of computer science, psychology, and political science. ( http://www.umsl.edu/~sauter/DSS/10SIMON.html )

[ QUOTE ]
In playing low-limit poker against human beings I think we're in a similar position. If we've hit on a way of playing that allows us to make 3 BB/100 against a certain set of opponents, in practice it may not be worth all the extra mathematical and statistical work needed to up that winrate to 3.5.


[/ QUOTE ]

Exactly, the marginal increase in EV might not be worth the effort. This may also be the case when your bankroll might be more subject to fluxuations. More conservative players might shy away from situations/plays that causes increased monetary swings but also increases overall EV. These players may not like to play out of their comfort zone. (This could be also be considered a weakness in a player's game where he/she doesn't like to switch gears when deemed appropriate.)

However, as David Sklansky has written about, a relatively small difference in performance between two players goes a long way to increasing your hourly expectation. For instance, a player that earns 0.5 BB/hr more than someone else will make much more in the long run since this player can move up limits faster. Of course, this means that the player is also getting progressively better as he/she moves up. I suppose it depends on a player's goals, motivation/laziness, and abilitiles.

[ QUOTE ]
Good question about the academic work on optimal heads-up bots. Certainly I guess this is an arena where a truly optimal strategy could be worked towards, although it would be subject to the game theoretic limitation that one assumes fully rational play on the part of the opponent and then tries to make sure that your own strategy has no leaks. If, like most real people, the opponent has some non-rational preferences, like slightly over-valuing pocket pairs just because they like the look of them, then we need to use something other than classical game theory to help us maximize our performance against this guy.

I took a look at some of the academic poker literature (Uni of Alberta group mostly) a while back, and yes, it seems their heads-up bots were becoming very good, but I can't recall whether the authors made claims for optimal performance in the strict sense.

[/ QUOTE ]

Apparently, the Univ. of Alberta reseachers developed a couple of strong bots with slightly different strategies. One bot tries to maximize winnings whereas the other tries to minimize losses. On the surface, I would think that there would be a way to design a bot to also model an opponent's gear changing and personal style of play.

Regards,
Wally