Poker Strategy With David Sklansky: Manipulating For A Bracelet (Or Watch)Theory Of Poker Author Continues His GT-NO Series On Exploitative Play |
|
When I won the 1982 World Series of Poker limit five-card draw event at Binion’s Horseshoe in downtown Las Vegas it was basically a done deal when it got down to two players.
I say that not to sound overconfident, although I was, but because the rules had the player in first position either raising or folding in such a way that opening rather than folding showed an immediate profit if my opponent called less than half of the time. It was quickly obvious to me that my opponent needed a pair of sevens or better to call and was therefore folding about two-thirds of the time.
Thus, theoretically, I should have opened 100% of my hands and collected the profit.
But I didn’t. I will explain shortly.
John Nash proved that in any fair, heads-up poker game with two players and no rake, there is a strategy that cannot lose in the long run. A strategy that can actually be divulged to the other player and still not lose.
It includes playing certain hands in more than one way in a specifically prescribed random fashion and has you often playing more than one hand in the same way. So the opponent who knows your Nash/Game Theory Optimal(GTO) strategy won’t be able to deduce your exact hand.
He would know that you will, for instance, open for three blinds with 9-4 suited 35% of the time. But knowing those kinds of things won’t mean he can devise a winning counter strategy.
The bottom line is that GTO play cannot lose in the long run if the game is heads-up. And that is close to being true if the game is multiway. It doesn’t matter how the opponents play.
In fact, pure GTO strategy does not even take into account how that opponent has been playing. Yet in spite of this and in spite of telling him your strategy, it will not just break even, but rather will beat almost any opposing non-GTO strategy.
Still, in most games you shouldn’t try to use it even if you somehow had a perfect GTO playing computer in your pocket. Because if an opponent is legitimately playing a non-GTO strategy, you can beat him worse if you find a good counter strategy that “exploits” his errors.
(I say “legitimately” because it is conceivable that he is faking it in order to entice you to deviate from pure GTO into a different strategy that he can exploit.)
If your opponent is playing non-GTO, how much your own pure GTO strategy will win will depend on how far he strays from GTO. The same is true if you elect to also stray from pure GTO to take even greater advantage of his errors.
But the important thing to understand is this: The further he strays from GTO, the greater the gap between the increased GTO profits and the increased exploitive counterstrategy profits.
In other words, it becomes dumber and dumber to stick with pure GTO as the opponents play worse and worse. Anyone who would have used pure GTO when it got heads-up in that five-card draw tournament would be very wrong.
Almost all players who love their solvers are aware of what I just wrote. And they have a way to use that computer to come up with a better strategy than pure GTO. Its called “node locking.”
They essentially tell the computer to come up with a new GTO strategy under the assumption that the opponent plays certain hands in a non-GTO way. (Without node locking, the computer is essentially assuming that the opponent is playing GTO).
That computer would be telling them to open every hand in my 1982 match.
Furthermore, they now have a tool that generically guides them as to how to node lock. Its called MDA for Mass Data Analysis and it shows how real people play hands in real life. Thus a player armed with these tools can go into a small game with bad players, follow the advice of the computer, and have a winning strategy that ought to do better than pure GTO.
But even with all those tools those players are probably not going to do as well as the best players in small games with lots of amateurs.
One reason is that different amateurs make different kinds of bad mistakes. If you assign to all of them the same generic mistakes that your data reveals, rather than making note of the specific mistakes you see and extrapolating from there, you will win less.
The second reason is that the computers strategies don’t include manipulation. There are things you can do to drive opponents further away from GTO than they are already playing. Or, if they are already far from it, you can try to ensure that it will stay that way.
I don’t think computers/solvers have been programmed to think along those lines.
In that 1982 tournament I was rather sure that my opponent would wake up to the fact that he was playing way too tight if I opened every hand. So I occasionally folded a positive EV opening to keep that from happening (just one of several ways to manipulate).
Side note: 1982 was the lone year that the WSOP offered gold watches instead of gold bracelets, and I happened to win two of the 13 given away that year, along with Billy Baxter. You can see these watches and myself in an upcoming episode of Pawn Stars, which will be broadcast later this year. ♠
David Sklansky is the author of The Theory of Poker, as well as nearly two dozen other guides on gambling, poker, and other games. The three-time WSOP bracelet winner’s latest book, Small Stakes No-Limit Hold’em: Help Them Give You Their Money, is now available on Amazon. You can contact Sklansky at [email protected].
Copyright © 2001-2024
CardPlayer.com. All rights reserved.
