Re- How Do You Win? Part 3 (Long!)
Sgt. Rock
KENN...@webtv.net (CAMILLE Visconti) asks:
>
> {...how do you win?...]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Some things you might do (or even just think about) to help
win at poker might be:
Part:
1. Find (or create) a Delta Factor.
2. Emulate Treebark.
3. Beware (but learn from) the OTBS.
4. Be both intellectually AND emotionally prepared for "Luck."
----------------------------------------------------
3. Beware (but learn from) the OTBS
I briefly mentioned the OTBS (Opportunity to be stupid) concept
in another post a couple weeks ago. Some days your hands just
seem to play themselves, and you hardly ever get an OTBS. Some
days you have OTBS after OTBS, again and again. The trick, of
course, is to recognize each OTBS for what it is, and react
appropriately. Bad-beats, getting slow-rolled, encounters with
jerks, being "needled," and other things that upset many
players don't really bother me. Realizing after-the-fact that
I let an OTBS become an S-OTBS (Seized-OTBS) when I should have
passed on it does. It's the one thing that, if allowed to go
unchecked, can nudge me towards tilt.
You've probably seen many players taking notes at the table. I
do it sometimes, generally to record an S-OTBS. Here, let me
ignore my extreme embarrassment and share a notebook page with
you-
2/20/99 EQ 10-20
me AJ BB
him Q4o midpos yoyo limps
A4X,X,Q
BC
BC
BRC
Translation: 10-20 Hold 'Em, Emerald Queen (Tacoma). I have AJ
in the big blind, and get a free play head-up against a middle
position limper. Flop comes A,4,?- I bet, he calls. Turn is
a blank- bet and call. River is Q, I bet (OTBS #1), he raises,
and (OTBS #2) I call. He shows me Q4 offsuit.
I don't suppose it's necessary to discuss why my bet and
subsequent call on the river were moderately and very stupid,
respectively. Thanks. This is already humiliating enough.
The real point, however, is that I try real hard to avoid
"denial," and refuse to blame "bad luck" for a loss when in
fact I screwed-up. This page from my pocket spiral notebook,
along with some (too many!) others, are pinned, side-by-side,
on my bulletin board. I study, re-live, and scribble on
them. I take them down, look them over, carry them in my
pocket for a while, and pin them back up. I give them pet
names (like typhoons or hurricanes), and dream about them.
Bad dreams. Then I perforate, singe, fold, spindle, mutilate
and otherwise punish them in unspeakable devious ways.
But then, when the same (or similar) situations come up again
in live play, sometimes a wonderful thing can happen. I
sometimes recall a specific previous brain fart, and can
sometimes even "see" the notebook page before my eyes, and/or
recall the pet name I gave to that particular catastrophe. I
think I once exclaimed, "Mary!" and the poor dealer thought I
was asking the Holy Mother for help, when I was really just
remembering how badly I had handled the same situation once
before. Then, (finally, the payoff...) all the above-described
obsessive behavior might actually help me avoid making the
same mistake again. Imagine that.
----
See ya on the river...
Sergeant Rock
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
Abdul Jalib
> You've probably seen many players taking notes at the table. I
> do it sometimes, generally to record an S-OTBS. Here, let me
> ignore my extreme embarrassment and share a notebook page with
> you-
>
> 2/20/99 EQ 10-20
>
> me AJ BB
> him Q4o midpos yoyo limps
> A4X,X,Q
> BC
> BC
> BRC
>
> Translation: 10-20 Hold 'Em, Emerald Queen (Tacoma). I have AJ
> in the big blind, and get a free play head-up against a middle
> position limper. Flop comes A,4,?- I bet, he calls. Turn is
> a blank- bet and call. River is Q, I bet (OTBS #1), he raises,
> and (OTBS #2) I call. He shows me Q4 offsuit.
>
> I don't suppose it's necessary to discuss why my bet and
> subsequent call on the river were moderately and very stupid,
> respectively.
Let's discuss it.
> Thanks.
You're welcome.
> This is already humiliating enough.
Hey, really, no problem, I don't mind.
First off, let's suppose this player is totally unknown. Heads up
there is a game theoretic optimal way to play. Such a strategy would
be undefeatable. Any mistakes your opponents make then become money
in your pocket, though you won't exploit them to make even more money.
I assert, based on my experience but no proof, that the game theoretic
optimal play on the river would be to bet and then call a raise. If
your whole strategy is optimal in this sense, you cannot lose, no
matter how your opponent plays.
Next, let's suppose instead that this player is a passive fish.
Such a player generally calls with any pair and any hand with
a potential to make something, and he bets and raises with only
very strong hands, like two pair. You can exploit such suboptimal
play to make a lot of money. Since he will call with any pair, and
won't be inclined to bluff bet, your value betting the river is a
necessity, not a mild mistake, right? The only question is: should
you call a raise? Most fish will exhibit some backbone and
occasionally make a move, I've found, especially if they are getting
frustrated. If you are very sure he wouldn't raise you with less than
two pair, then you can lay it down and save a whole big bet. But
you're getting over 7:1 to call here. I have to admit that I almost
never lay down top pair good kicker heads up, particularly not on the
river.
> The real point, however, is that I try real hard to avoid
> "denial," and refuse to blame "bad luck" for a loss when in
> fact I screwed-up. This page from my pocket spiral notebook,
> along with some (too many!) others, are pinned, side-by-side,
> on my bulletin board. I study, re-live, and scribble on
> them. I take them down, look them over, carry them in my
> pocket for a while, and pin them back up. I give them pet
> names (like typhoons or hurricanes), and dream about them.
> Bad dreams. Then I perforate, singe, fold, spindle, mutilate
> and otherwise punish them in unspeakable devious ways.
>
> But then, when the same (or similar) situations come up again
> in live play, sometimes a wonderful thing can happen. I
> sometimes recall a specific previous brain fart, and can
> sometimes even "see" the notebook page before my eyes, and/or
> recall the pet name I gave to that particular catastrophe. I
> think I once exclaimed, "Mary!" and the poor dealer thought I
> was asking the Holy Mother for help, when I was really just
> remembering how badly I had handled the same situation once
> before. Then, (finally, the payoff...) all the above-described
> obsessive behavior might actually help me avoid making the
> same mistake again. Imagine that.
Great advice and I enjoyed your posts.
--
Abdul
David desJardins
>> Translation: 10-20 Hold 'Em, Emerald Queen (Tacoma). I have AJ
>> in the big blind, and get a free play head-up against a middle
>> position limper.
>
> there is a game theoretic optimal way to play. Such a strategy would
> be undefeatable.
Unfortunately, this isn't so. The problem is that constructing such an
optimal strategy depends on knowing which hands the opponent might have.
And that in turn depends on the player's strategy at a time when there
are more than two players in the game. If you don't know for sure which
hands the limper would have played, then you can't compute an optimal
strategy to apply once the game goes heads-up.
David desJardins
Abdul Jalib
I know. I was hoping to just gloss over that slight glitch. I
think there is still an approximately optimal way to defend against
an opponent heads up after the pot started out multiway. If need be,
assign a probability distribution over the possible hand distributions
that an unknown player might play, and then compute your strategy
based on that.
--
Abdul
WenMax
fold against that?
I don't actually know the answers. I'm not sure I know the questions.
Wenatchee Max
David desJardins
> I know. I was hoping to just gloss over that slight glitch. I
> think there is still an approximately optimal way to defend against
> an opponent heads up after the pot started out multiway.
I think so too. I also think this is the most promising approach to a
competent computer poker player. But there are a lot of details to be
considered.
Not everyone has even accepted the principle that limit poker with
infinite stacks is equivalent to a finite game; if this isn't true, the
analysis is going to be a lot more work.
> If need be, assign a probability distribution over the possible hand
> distributions that an unknown player might play, and then compute your
> strategy based on that.
I think this is more or less the right way to proceed. But I would
combine this with some sort of sensitivity analysis. In most cases, the
decisions will be relatively insensitive to the prior distribution, and
then the decisions that the algorithm makes will be relatively robust.
But there will be some cases where the decisions are highly sensitive to
the specific probability model. In those cases, the "optimal" algorithm
is basically generating noise: if the decision depends strongly on how
likely the opponent is to have played a certain hand, and the model for
whether the opponent will play that hand is just a crude one, then
there's no particular reason to trust the conclusions. And one can
clearly see that many of the most intense discussions in this group
(and, one might reasonably believe, many of the most intense decisions
at the table) are indeed highly sensitive to one's assumptions. Look at
the threads about people who raise and then fold because they become
convinced the reraising opponent has the nuts.
In cases where the optimality is in very serious question because of
strong dependence on the probablity model assumptions, a more robust
("play safe") approach is called for. This would be especially true if
one were to risk money on the computer player, or if one were competing
for an "important" prize. One failing of computer backgammon players
for some time was that they would get confused in some positions, and
double when behind. Even though this happened very infrequently, it
could make their EV highly negative, because they would lose a million
points from doubling and redoubling on those rare occasions.
Computerized algorithms for poker can fall into similar traps.
Of course, most people here probably don't think a computer is really
going to be able to compete at poker any time soon.
David desJardins