Tuesday, October 09, 2007

Exploitability as it relates to tournaments

There's not a lot of interesting stuff to write about from this week so I figured I'd try writing about some strategy/theory stuff after a quick recap. I lost about $20K playing some big buy-in sit and go tournaments trying to earn some frequent player points on Stars to get to Supernova status in their VIP promotion for the year. I figured I was around a break-even player in these tournaments and I'd be quite happy to breakeven or even take a small loss to rack up a lot of points since there's a pretty decent amount of value in becoming a Supernova. Unfortunately since I've more or less stopped playing cash games on Stars I was no longer on pace to make it for the year. I ran pretty bad though and didn't cash anything so I dropped a bunch of money and I still need 24K VPPs on the year. Cash games and tournaments didn't go great either so it was a pretty crappy week. I got killed on Sunday cashing only one tournament despite playing the busiest schedule I ever have. Oh well. I made back a little tonight by having a good cash session and finishing a disappointing 12th in the $1K Monday on Full Tilt for $4k. OK on to some strategy stuff. A few people have asked me something about exploitable play recently so I figured why not write about that.

In a loose sense, when I talk about playing exploitably I mean playing in a predictable way to which there is a simple counter-strategy. If your opponents figured out what you were doing they could exploit your predictable play by implementing the counter-strategy. At first glance this may seem like a bad thing. You might think we should try to balance our play more. For example, instead of always having a big hand when we make a big bet we should also bluff sometimes too so we're not so easy to figure out. Of course in general this is good advice, but overall the situation is not so simple. It's been suggested by Sklansky and others that there is some unexploitable strategy, perhaps based on game theory in some way, to randomize your play such that you can not be beaten. The best an opponent could do is play the same strategy and breakeven. Fortunately, as I know next to nothing about game theory (and I probably know a lot more math than you), in the real world such a strategy is not ideal. The reason is simply that our opponents are not that good, and we can make a lot more money from them playing exploitably. There are two simple reasons:

1. Our opponents are not observant enough to figure out what our strategy is.

2. Even if they do figure it out they won't know how to counter it properly.

In particular, for mutli-table tournaments, realted to point 1 is the simple idea that due to the structure of these tournaments we don't play against the same people very often, therefore making it difficult for anyone to play enough hands with us to figure out what we're doing (of course if you're a high-volume regular player this rule doesn't apply as much).

Let's look at a very basic example. Suppose we're playing against a player who always calls our bets. How do we counter his exploitable play? Simply, we will never bluff this player (and we'll bet more of our marginal hands for value since he'll call with so much worse). But now we are playing exploitably ourselves of course. If he realized we were never bluffing he could just stop calling us without a good hand. But anyone who has played microstakes poker at all has run into one of these types of guys who just never folds if he has any kind of pair or draw, and never really changes his strategy even though everyone with half a clue never bluffs him. The fact that we are playing exploitably is irrelevant, and in fact clearly leads to maximal profit against this player, whereas someone mixing in bluffs to remain unpredictable is wasting money.

That was obviously an extreme example but it illustrates the point. Now let's focus on tournaments more specifically. It's not a big secret that there a lot of players in tournaments at every level who aren't really the best poker players. That combined with the fact about not playing against the same people frequently in tournaments means the general rule in tournaments is that we don't need to worry very much about playing exploitably. By the time people figure out how you play your table will break and you'll have 8 new people who have no idea how you play. For the most part we just worry about maximizing our advantage against the typical mistakes our opponents will be making. However, there are a lot of situations that happen frequently enough that it's important not to be too predictable. Here are a few common things people do:

1. Raising preflop. If you vary the amount you raise preflop based on the strength of your hand, the strong players at the table are going to catch on fairly quickly. Also if you always raise when it folds to you in late position everyone will notice, but a lot of people won't adjust properly.

2. Reraising preflop: Same as above about raise amount. There are also a lot of players who will never reraise in most situations without a huge hand. This is often not really such a bad thing, but it lets strong players off too easy.

3. Continuation betting: Again if you change the amount you bet based on whether you like the flop or not in some constant way, people will pick up on it since you're going to be c-betting a lot usually.

However, beyond these simple things it's going to be hard for anyone to pick up anything much more complicated. The other thing is that beacuse of the typically fast structure of online tournaments it is often very difficult to exploit anything but these most basic tendencies even if you pick up on them. Having some detailed information about how an opponent plays the river won't help too much when you're all-in after the flop every hand. OK I'm pretty sure this post sucks but I'm too tired to try to fix it and Im going to bed.



Anonymous said...

Hey Watts, did you get my email about friday in Waterloo? Gonna be ugly buddy...


Dave P said...

Nice post. I am teaching Game Theory this term and am running a prisoner's-dilemma type tournament, and once of the things that always comes out of it is, as you point out, in real life the "equilibrium" strats are not the ones that will make you do the best.

I guess forcing them to compete in a poker tournament would have the same effect... is it unethical to take money from your own students? Even so, maybe it would be less unethical for *you* to take their money instead...

Mike said...

Now that I'm rested I've decided this post does not in fact suck. And yeah I got the e-mail Tim, I can't wait.

Dave Churchill said...

For online play, probably true... unless everyone assumes this strategy.

Isn't this one of those "you can justify anything if you talk about it long enough" issues? Not to say the point isn't interesting, but in the end of comes down to surviving coinflips and just not sucking. By the time it 'matters' online, they are at the table long enough to be able to get a feel for you, so is it really an issue?

Toulouse said...

Great post mike. You should write more posts on game theory and hand analysis. Your blog is always a great read. Keep it up.

P.S You better be coming home for xmas