« Why | Home | High Ate Us »
10.19.06
You Gotta Do What You Gotta Do: Predictability And The Roshambo Defense
posted by Absinthe | 3:30 PM
We’ve all been in a lot of situations when we felt we were dead at the deal. Shortstacked in early position, pick up pocket queens, there’s no way you lay the hand down and you can’t open anything less than all-in; someone behind you picks up kings or aces and you’re gone. Extra credit if it’s kings and you run into aces.
But what about all the other times when we let the cards play us rather than the other way around? Conventional wisdom is a powerful force, but it’s not always the same thing as common sense; what’s more, the prevailing forces of conventional wisdom can often create situations where common sense dictates the opposite play1 - if everyone’s jumping off a bridge, there’s a strong possibility that you should find a way not to.
This is one of the curiosities of poker theory. If everyone plays optimally, the game devolves into a game of chance. But the common understanding of what constitutes optimal play can actually change what the optimal play is. In other words - and this is a noodle-baker for sure - in poker, the more people believe in something, the less true it may become.
Every time you make a decision that you felt was “necessary” - especially but not only when it costs you a tournament - you should take the time to examine your reason for making it. You can’t and shouldn’t avoid using your experience to color your perceptions of your current situation, but if you can you should take the time to consider why and how you were led astray, or what it was that nudged you toward those happy accidents, bad decisions that paid off. Oftentimes you may have lazily - or accidentally! - made a predictable move, either because you were being predictable, or because your actions coincided with what your opponent believes to be indicative of the hand you just happened to hold.
If you have certain quirks that cause you to play a category of hands a certain way (checkraise-overbetting with overcards plus a flush draw, let’s say, to name one of my more glaring might-as-well-turn-my-cards-face-up weaknesses), and you don’t play other hands in that way, you have an information leak. Loose lips spew chips - if your actions allow an observant opponent to narrow your range to a specific set of holdings, you make it very easy for them to determine the proper course of action.
You may be playing the hand this way for very sound reasons, and there are circumstances under which narrowing your range does your opponent no good2. But sometimes you may be playing the hand in this fashion for no reason other than the understanding that “that’s the way the hand should be played,” which might be incorrect either because of a common misconception (”you gotta raise more preflop with a pair of queens than with a pair of kings or aces!”) or a prevailing belief in the value of a certain kind of play (”straight draw and a backdoor flush draw equals call the flop, checkraise the turn if my suit comes in”) - beliefs that may well have been true at their inception but which have become less correct as they are more commonly known.
There are, however, circumstances under which predictability can be turned to your advantage, if it’s taken to an extreme. If you narrow your actions to a specific range, you may never playing optimally, but you’ve made it impossible for your opponent to do likewise.
I’ve thought about this a great deal, and as a result I’ve tried to build a strategy around making betting patterns as Roshambo-like as possible. This means that bets should almost without fail be determined by the texture of the flop (and occasionally the number of opponents and their relative positions), whether you’re making a continuation bet, a value bet, a bet with a monster or a bet to ward off draws. Sometimes you may check, but if you bet you’re going to bet the same percentage of the pot for that situation every time, absent other considerations like the relative stack sizes which may force a check or an overbet. Yes, this means you’re costing yourself some chances to pick up a pot cheaply or costing yourself a little more when you’re beat, but in return you’re giving up no new information about my hand with either a bet or a check, especially considering you may also checkraise or check-call with a reasonably broad selection of hands and draws if you’re out of position and you know your opponent is likely to bet.
Your actions become predictable within a certain range but also inscrutable; they do nothing to narrow the range of your holdings, only show that you’ve thrown scissors (lead) or paper (check), with the possibility of rock (checkraise!). What’s more, since the bets are usually sizable3, they do allow you to get information from your opponents’ actions, as they are less likely to make a smooth-call with an extremely marginal hand as they might against a probe bet or other weak bet.
This is an incredibly easy strategy to adopt, since it requires little more than the ability to calculate the proper bet size and an understanding of your table image (if people are folding to your leads, keep makin’ ‘em!), and allows you to think more about your opponents’ holdings while knowing that they will only discern yours by accident.
I’ve had good results with this strategy, and often use it to autopilot through the first hour of a tournament when I have little information on my opponents anyway. Anyone who wants to give it a try (maybe even with a suitable random-action generator to see how well it would fare as a dumb-bot strategy) and report results back to me is welcome to help me hammer out Roshambo/System.
- See this RGP post, originally referenced eons ago by hdouble, for a better explanation of the Delta Theory, which also provides an instinctive rational for the great players’ ability to change gears. [back]
- See Phil Gordon’s description of Prahlad Friedman’s strategy in one of the appendices to Gordon’s Little Green Book. [back]
- I usually adopt a range between 60% and 80% of the pot for a single online tournament and stay as close to those percentages as well as round numbers permit, modifying particular cases for flop texture to bet more on draw-heavy boards [back]
Popularity: 16% [?]
Topics: Poker |
October 20, 2006 at 11:11 AM
Great piece of theory. Thanks for posting.
October 22, 2006 at 2:38 PM
October 24, 2006 at 8:07 PM
November 6, 2006 at 1:02 PM
In a lot of ways you’ve described the way that Gus Hansen plays. All of his play is determined post flop, with really very few exceptions. I just watched a video that he put together that really nails that thought down. And you’re right. It’s impossible to narrow to a range of hands if any hand is now playable before the flop.
Ideally, I think a person needs to combine a solid understanding of hand rankings preflop with a grasp on how relative hand strength changes post flop. This combined with good observation of basic opponent tendencies is going to be the most solid possible strategy, at least as far as I can envision. If one can combine a lack of emotional swinginess with this idea consistently, it seems like winning poker would be an almost automatic result, discounting bad beats.