Heads up in SnGs, Can you do anything?

[Womble]

If you are a even stacks at the end of a SnG with high blinds such as 5000 vs 5000 (300/600) isnt it basically luck to see who wins first?

When I get to this situation I dont know how t give my self a chance of winning. It seems to be the person with the best cxards. Is there anything you can do?

[UMTerp]

The only big mistake you can make heads up with the blinds that big is to fold too much, particularly if the stacks are unbalanced. In the example you gave, you can wait 4-5 hands to "pick a spot" if you get really bad cards, but soon you'll get to the unbalanced stack point, and yuo'll be correct to push from the SB with marginal or worse holdings. There is an inordinate amount of luck involved regardless. And you're obviously playing for your entire stack any time you enter a pot at this point.

[asofel]

[ QUOTE ]
If you are a even stacks at the end of a SnG with high blinds such as 5000 vs 5000 (300/600) isnt it basically luck to see who wins first?

When I get to this situation I dont know how t give my self a chance of winning. It seems to be the person with the best cxards. Is there anything you can do?

[/ QUOTE ]

Be aggressive. Bluff. Mix things up.
A quote from Andy Bloch "In heads-up play, when I realize someone's not going to bluff enough, I'm almost never going to lose against him. Because when both of us have nothing, I'm going to win all of those pots. I'm going to take well-informed stabs at pots. When someone checks, I'll be pretty certain I can bluff at the pot.

Obviously this is assuming you're not facing someone who's also playing aggressively. If you played a mirror image of yoruself, then yes, the cards would determine things. But there will always be some inequalities between you and your opponent. Take advantage of their flaws and try to minimize your own.

[junkmail3]

I don't know if this is short term luck or what ... but I find a relatively good SnG Heads Up Strategy (at least for the $10s and $20s) is the call/fold the SB and check/fold the BB (if you don't have a hand) for the first two or three orbits.

In my experience this will loosen up the other guy and he'll get more passive (or he already is). they I'll lay into him with pushes from the SB (unless it's an awful hand) and he'll start folding his SBs to me.

It makes me think we're playing nice ... and then I poke his eye out.

(I'm not recommending giving away half of your stack this way. It usually evens out for me, or puts me at a slight disadvantage at first, but it's very easy to make up after that.)

has anyone else noticed this general trend?

[lawfive]

Be aggressive. That should increase your EV over .5.

[citanul]

see daliman's "all in for the title" experiment, amongst other posts of that nature.

i believe that there was a post by someone that said something like "the only thing that you can do wrong is fold, and you can't be all that wrong going all in every hand," to which strassa replied "my heads up strategy changed immensely when i realized this" or something like that.

citanul

[The Yugoslavian]

You can push. And push frequently.

If your opponent allows you to see flops and is very aggressive then you can also trap by limping and then check/calling with any piece of the flop.

Yugoslav

[Big Limpin']

Pretend your checks are a hot potato.

[Unarmed]

[ QUOTE ]
to which strassa replied "my heads up strategy changed immensely when i realized this" or something like that.


[/ QUOTE ]

The gem from Strassa was the fact that the absolute best your opponent can do against a push every hand, call every push strategy is to beat you 52% of time. That is an absolutely massive observation, and suggests that effective HU play (with relatively large blinds) beings with getting your chips in the middle on every hand, and can be improved by stripping out certain hands/situations from your push everything strategy.

HU is completely opposite from early SNG play. Early on the default is fold, and I need a real good reason to play a hand. HU the default is push, and I need a pretty compelling reason to do anything else.

BTW, the first time I came upon an opponent who pushed everything, I absolutely HATED playing him. After that point I was pretty much converted.

[UMTerp]

[ QUOTE ]
The gem from Strassa was the fact that the absolute best your opponent can do against a push every hand, call every push strategy is to beat you 48% of time.

[/ QUOTE ]

Why wouldn't it be 50% if he called every time?

And the blinds obviously factor into this too. If the blinds are VERY large (400-800 at least), intuitively, I could see how pushing every time could be correct if you thought that there was a chance your opponent would fold. Obviously, if you know that he'd call any push, pushing a hand like 32o is clearly wrong. Also, if the blinds are still 50-100 or something, pushing every time is clearly wrong.

I'd like to see this post by strassa that you're reffering to. I think he's one of the more informative posters on here, but I don't see how he could come to the conclusion that the best your opponent could do against a "push every time" strategy is 48%. Maybe he had a stringent criteria for your opponent's calling standards or something, I dunno.

[Jman28]

The difference between heads up play and non heads up play is that going all in on a 50-50 shot in non heads up play hurts your $EV. Heads up it does not.

When heads up, be much more likely to call an all in if you think you are slight a favorite than on the bubble.

-Jman28

[morgan180]

i agree - where's chief with the push cow??

seriously, be agressive, if you have a passive player any hand your are in the SB raise. if he completes in the SB push from the BB. You can filter out a few hands bands on stack sizes but it really doesn't matter. Open way up and get aggressive, you'll end up ahead more times than not.

[Unarmed]

[ QUOTE ]

I'd like to see this post by strassa that you're reffering to. I think he's one of the more informative posters on here, but I don't see how he could come to the conclusion that the best your opponent could do against a "push every time" strategy is 48%. Maybe he had a stringent criteria for your opponent's calling standards or something, I dunno.

[/ QUOTE ]

Oops, too quick on the draw, obviously its the other way around. Here's the post....how the heck do I shorten that?

http://forumserver.twoplustwo.com/showflat.php?Cat=&Board=singletable&Number=1245080 &Forum=f22&Words=%2B52%5C%25%20%2BhU&Searchpage=0& Limit=25&Main=1244652&Search=true&where=bodysub&Na me=&daterange=1&newerval=1&newertype=y&olderval=&o ldertype=&bodyprev=#Post1245080

[UMTerp]

Thaks for the link. You can shorten it if you play around with the "URL" button (below the tezt you enter on the "Reply" screen).

FWIW, since I really started keeping good stats for my SNGs (towards the end of November), I'm 153-108 heads up (58.6%), and I'm definitely more aggressive than the majority of my opponents. I'm not positive that's sustainable, but I would think that 55%+ would be a reasonable goal for a good SNG player at the $30 or $50 level. Of course, starting stack size matters too. I'd tend to think that on average, most of us would have a slightly smaller stack than our opponents due to the tight style most of us try to play in the early levels. I'm not basing the 55% on any analysis, just an inutitive guess.

[rachelwxm]

UMTerp,
I think you are talking about a different thing. What 52% is what best you can do against a maniac push and call w anything PF. Your normal opponent is not. So average opponent is doing worst than a robot. /images/graemlins/blush.gif Shocking but true.

[UMTerp]

Gotcha. I thought you were saying that 48% percent was the best you could do, which didn't make sense.

And the blinds matter too when computing this number. Theoretically, if your oppenent pushed and called any two with infinite-sized stacks apiece and infintessimily small blinds, your maximum win rate would be 85.2% (the amount of time AA wins vs. a random hand). In practice, this situation obviously isn't very helpful, but 52% still seems a little low to me. There's a big difference between pushing every time at the 100-200 level and pushing every time at the 300-600 level.

[rachelwxm]

I think they are using 300/600 blind with 5000 chips each. Good catch. /images/graemlins/cool.gif
I almost trap you.

[dethgrind]

Eastbay and some others did quite a bit of simulation work on this exact problem (heads up, 300/600 blinds).

Check out this graph to see the winrates for various strategies: http://www.pic.ucla.edu/~sharnett/maniac.jpg

I believe the equilibrium strategy at this blind level is .65/.58 push/call.

(the percentages here are based on a reasonable ranking of starting hands by Eastbay)

[rachelwxm]

that's very interesting. Can you or eastbay explain the meaning of this graph? I am all ears. /images/graemlins/smile.gif
Thanks in advance and I wonder if there is analytical way of deriving this.

On a different subject, where does 70/30 rule come from? What level of calling pushing stadard does it assume your opponent? Sorry to dig some old stuff since this must be discussed in great detail before I learn the word "SNG" /images/graemlins/smile.gif

[Unarmed]

The colors represent your win% given what range of hands (represented by the %s on the x and y axis) you will push and call Villain's pushes with. The little legend in the top left shows what win% the colors correspond to. So if you call Villain's push with the top 20% of hands, and push the top 20% of hands yourself, you'll win the HU match around 30% of the time.

You can verify deth's equilibrium point is the "reddest" part of the chart, and indicates a 53% win%. If you need a listing of what range of hands is considered top x% go to deth's homepage and poker tools.

[rachelwxm]

Thanks. One small point, if you fixed the blind, I guess you have to adjust your tightness/looseness of pushing calling too, right? Would't that be a better choice? For example, calling push w any two once he has 2bb left, etc. I wonder if that will increase your success against maniac.

Could you provide the link to original thread?

[dethgrind]

Actually, .65/.58 is the Nash equilibrium for the entire game, not the optimum strategy against the maniac. Effectively, this means that if you play that strategy, your opponent can do no better than 50%.

Note that this does not mean you should play that strategy in general. Usually your opponent will be making some mistakes which you should be looking to exploit by deviating from .65/.58.

This happens to also be around the optimum strategy against the maniac. Notice that there is a large range of strategies which all perform about as well as each other. I found this to be pretty encouraging, since that essentially means that the difference between playing exactly correctly and roughly correctly is negligible.

[The Yugoslavian]

[ QUOTE ]

I found this to be pretty encouraging, since that essentially means that the difference between playing exactly correctly and roughly correctly is negligible.

[/ QUOTE ]

This also means that it's easier for the fishies to play close to optimal strategy, /images/graemlins/frown.gif.

Sweet graph though, I should make it into a big poster for my wall, /images/graemlins/wink.gif.

[rachelwxm]

[ QUOTE ]
Actually, .65/.58 is the Nash equilibrium for the entire game, not the optimum strategy against the maniac. Effectively, this means that if you play that strategy, your opponent can do no better than 50%.


[/ QUOTE ]
what does that mean? /images/graemlins/confused.gif

[dethgrind]

Sorry, I wasn't too clear with that. The game is heads-up, 5000 chips each, blinds 300/600. If you push with the top 65% of hands and call with your top 58%, your opponent can at best win 50% of the time.

From wikipedia.org "If there is a set of strategies for a game with the property that no player can benefit by changing his strategy while the other players keep their strategies unchanged, then that set of strategies and the corresponding payoffs constitute a Nash equilibrium."

So even if your opponent knows exactly what you are doing (playing .65/.58), it doesn't benefit him to play anything other than .65/.58, since nothing will get gim better than a 50% winrate.

In general, NE strategies need not be the same for all players; this game just happens to be symmetric.

As for playing against the maniac, I don't think .65/.58 is the best strategy, though it is close.

[rachelwxm]

[ QUOTE ]
Sorry, I wasn't too clear with that. The game is heads-up, 5000 chips each, blinds 300/600. If you push with the top 65% of hands and call with your top 58%, your opponent can at best win 50% of the time.


[/ QUOTE ]
Interesting, it is a little bit unintuitive to me though. Lets say if your opponent know you are doing 65/58, then he can calculate what is the optimal range of hands to call or push so that is +EV for him. And you say the best he can do is just break even? /images/graemlins/shocked.gif
That is totally shocking to me.

[zaphod]

[ QUOTE ]
Sorry, I wasn't too clear with that. The game is heads-up, 5000 chips each, blinds 300/600. If you push with the top 65% of hands and call with your top 58%, your opponent can at best win 50% of the time.

From wikipedia.org "If there is a set of strategies for a game with the property that no player can benefit by changing his strategy while the other players keep their strategies unchanged, then that set of strategies and the corresponding payoffs constitute a Nash equilibrium."

So even if your opponent knows exactly what you are doing (playing .65/.58), it doesn't benefit him to play anything other than .65/.58, since nothing will get gim better than a 50% winrate.

In general, NE strategies need not be the same for all players; this game just happens to be symmetric.

As for playing against the maniac, I don't think .65/.58 is the best strategy, though it is close.

[/ QUOTE ]

Interesting stuff!
So what is the best strategy against the average player(probably hard to say how the average player plays..)
I guess the average player will not push enough, and also not call enough pushes. Is this statement then correct:

You should thighten significantly up from the call with the 58% best hands, since your opponents hand rates to be far better than the best 65%, and since you would have many steal good steal opportunities later.

Does anybody have a link to a page where you can see what chances every hand has heads up, vs a random hand?

[citanul]

[ QUOTE ]
Does anybody have a link to a page where you can see what chances every hand has heads up, vs a random hand?

[/ QUOTE ]

here you go (http://gocee.com/poker/HE_Value.htm)

[dethgrind]

[ QUOTE ]
[ QUOTE ]
Sorry, I wasn't too clear with that. The game is heads-up, 5000 chips each, blinds 300/600. If you push with the top 65% of hands and call with your top 58%, your opponent can at best win 50% of the time.


[/ QUOTE ]
Interesting, it is a little bit unintuitive to me though. Lets say if your opponent know you are doing 65/58, then he can calculate what is the optimal range of hands to call or push so that is +EV for him. And you say the best he can do is just break even? /images/graemlins/shocked.gif
That is totally shocking to me.

[/ QUOTE ]

It's true. Check out this graph. (http://www.pic.ucla.edu/~sharnett/equilibrium.jpg) (the simulation has a small range of error, so some of the points are slightly over .5)

Since this game is symmetric, if there is an equilibrium strategy, it's payoff must be .5. And there must be an equilibrium strategy. (google: John Von Neumann minimax theorem)

[dethgrind]

I should point out that the .65/.58 is based on a list made by Eastbay, NOT a list by random hands. His list is something like "hands that do (relatively) well when called by a decent hand". If you think about it, this makes more sense, since you won't be called by a random hand when you push. You'll be called by a somewhat good hand.

It's a rainy weekend, so I think I'm going to gather up all my poker stuff and organize it on a website. I'll post Eastbay's list sometime soon.

[rachelwxm]

Looking forward to seeing your stuff! /images/graemlins/smile.gif

[dethgrind]

http://www.bol.ucla.edu/~sharnett/poker/

Please email me comments/questions, and I'll use the feedback to flesh out the page a bit more.