Stealing in O8 Tourneys as compared to NL

[gergery]

There has been some discussion of stealing in O8 tourneys, so I thought I’d throw out a few ideas and see what y’all thought. I’m looking for feedback/other input here.

-- As always in poker, you make money when your opponents make mistakes. So what are the mistakes people make during the mid-late stages in tourneys (the steal-period)?

-- Mistake #1. In tourneys during the middle-stage “stealing zone” time, you are usually up vs. just 1 sometimes 2 other players, whereas in a cash game you are often multiway. So hand value should be based on “shorthanded value” not “multiway value”. So something like 3477 that is an easy fold multiway is good heads up, while A234 which is great multiway is merely average heads up.

-- Mistake #2. If everyone folds to your steal, obviously it doesn’t matter what your cards were. It’s only when you’re caught that it matters. So you want hands that play well vs. the range of hands someone will catch you with – which are usually the A2/-A4 type hands. So something like 34JTds might be a better stealing hand than A49Qr because it is more “live”.

-- Mistake #3. Hands run much closer together in value head-to-head in O8 than in holdem. So if you just look at your equity on that one hand, it becomes more correct to call with “marginal” hands as blinds get bigger. Essentially, you don’t face the problem in O8 of being up against a monster, because there are no monsters hands preflop in O8. So you should defend more liberally than you would in holdem. Don’t go crazy – you still need to play tight to raisers, -- but you should call more often.

Then there are some implications from these mistakes.

First, a thought on game theory and equilibriums. In holdem, Let’s say Player A steals with a range of hands, say any Ace, ay pair, any K. Player B figures this out, and looks at the weighted probabilities and says, those blinds are pretty big, my hand is not much of a dog, he’s stealing with a lot of hands, so I’ll call with any pair, any ace, any K, any suited connector. And maybe that range of hands is about 40% of all hands dealt. Then Player A figures out B’s strategy, and says, if he’ll call with 40%, that’s a lot, so I need a decent hand to be a favorite, so I’ll tighten up.

Then B sees that and adjusts by saying, if he now has a better hand, then I’m more of an underdog with my weaker hands, so I now need to tighten up. This iterative process continues until you get a Nash equilibrium, where against each player has an optimal strategy to use against an opponent who is playing optimally. In fact, some people like Eastbay have tried to calculate this on the SNG forum.

Using some back of the envelop calculations just like that, and some data that [url=http://home.earthlink.net/~craighowald/data/matchup2.html] shows how a given hand will matchup to a range of hands [url], you can work out that head to head in level 5 in a holdem SNG, you should push with about the best 68% of your hands, and call with about the best 28% of your hands. Against opponents who are playing optimally.

So what does that have to do with anything?

Implications from mistakes:
1) If hands run closer together in value in O8, then any given hand you have will be less of a favorite over a given range of opponents hands. So an opponent playing optimally should be more willing to call your steal. Which means you should tighten up more. Which means they’ll tighten up more. Which means you can loosen up some. Etc. This should reach equilibrium when you are stealing considerably less than in HE, and your opponent is calling more. The fundamental driver of the equilibrium on that hand is your equity for a given hand vs. a range of hands, and the size of the blinds.

2) But you don’t play against opponents who play optimally – they make mistakes. So if they will play too many hands, then you should call more. And if they play too few, you should steal more. Not too earth-shattering a conclusion, except that in O8, not calling is a relatively bigger mistake since your hand is rarely a dog. Which is why paying attention to who is too tight becomes relatively more important as compared to the cards in your hand – when you compare this to holdem.

3) You also don’t evaluate things solely on one hand. You should pass up small EV edges if bigger EV edges are just around the corner (at least once you are in the middle-later stages of the tourney, and don’t take this idea too far). So if have a weak hand that is getting good odds, you could still want to fold it knowing the next hand you’ll get will be significantly better. But the size of your edge is less, so as blinds get bigger this gets less and less correct.

4) As a result, folding equity, which is critical in any form of poker, becomes even more critical in O8. Because most hands run close in value you get relatively little value from hand equity, so if you can generate good fold equity by targeting tight players, then your hand picks up a lot of relative value vs. HE. And most players in O8 seem to think they need the nuts to win. So aggressiveness should be relatively more important, if you can get it to work.

Conclusion
1) Steal with hands that play better head to head.
2) Steal more frequently in general, until they start calling a lot.
3) Steal against the tight opponents, avoid looser ones
4) Be willing to defend more liberally.
5) Target opponents who don’t adjust hand value to multi-way

Not sure how earth-shattering this all is, but I was thinking about it after my tourney the other day so I thought I’d throw it up.

-- Greg

[MarkGritter]

So what does a good shorthanded O/8 hand look like? I can't imagine I'm going to showdown with 3477 all that often.

Maybe it is easier to answer this way: What is A234 missing for shorthanded play? My guess would be that it has little high hand strength.

In Hold'em I know that suitedness and connectedness become less important headsup and shorthanded, and pairs become a lot stronger. Do similar considerations apply in O/8?

What about high-only hands?

[gergery]

[ QUOTE ]
I can't imagine I'm going to showdown with 3477 all that often.
Maybe it is easier to answer this way: What is A234 missing for shorthanded play? My guess would be that it has little high hand strength.

[/ QUOTE ]

3477 is a better hand heads up than A234 is, because A234 doesn’t have high card strength. It is precisely because most people can’t imagine this that I started this thread.

pokenum -mc 500000 -o8 as 2d 3h 4h - 3s 4d 7s 7c
Omaha Hi/Low 8-or-better: 500000 sampled boards
cards scoop HIwin HIlos HItie LOwin LOlos LOtie EV
As 2d 4h 3h 150242 171254 305336 23410 202946 1691 28405 0.477
7s 3s 7c 4d 184957 305336 171254 23410 6095 179526 28405 0.523

[Luckless57]

You hit a topic that interests me quite a bit. I play a decent amount of omaha/8 limit multi's on stars and the middle to the end portion of the tournament when it is mainly 2 - 3 players in each pot, can be a tough spot to play. I've made a decent amount of final tables, but I know there is plenty of room for improvement. The comparison you provide with 3477 v a234 is interesting. I know that 3477 is a favorite, but it's a hand where your going to be forced to make tough decisions postflop more often then not. Where as a hand like a234 is fairly easy to play postflop. I find myself thinking about playing more non-conventional hands like 3477, but to me the problem with them is playing them post flop. Enjoyed reading your post, it was nice food for thought.

[Phat Mack]

[ QUOTE ]
[ QUOTE ]
I can't imagine I'm going to showdown with 3477 all that often.
Maybe it is easier to answer this way: What is A234 missing for shorthanded play? My guess would be that it has little high hand strength.

[/ QUOTE ]

3477 is a better hand heads up than A234 is, because A234 doesn’t have high card strength. It is precisely because most people can’t imagine this that I started this thread.

pokenum -mc 500000 -o8 as 2d 3h 4h - 3s 4d 7s 7c
Omaha Hi/Low 8-or-better: 500000 sampled boards
cards scoop HIwin HIlos HItie LOwin LOlos LOtie EV
As 2d 4h 3h 150242 171254 305336 23410 202946 1691 28405 0.477
7s 3s 7c 4d 184957 305336 171254 23410 6095 179526 28405 0.523

[/ QUOTE ]

I want to suggest an idea, but most apologize for not providing an extensive analysis--I'm somewhat tied up right now.

It's true that 7743 looks better than A234 in the showdown sim, but in volatile O8 situations (especially in pot limit), many times a decision must be made on the flop as to whether or not one will commit their entire stack. So besides looking at what type hand will show down a winner, it's also a good idea to look at which hands can take and provide pressure on the flop. Off the top of my head, I think A234 could "play" better on more flops than the 7743 could; it has iron-clad escapes.

I'll try to provide more detailed reasoning, and some math, when I get a chance.

[gergery]

[ QUOTE ]
It's true that 7743 looks better than A234 in the showdown sim, but in volatile O8 situations (especially in pot limit), many times a decision must be made on the flop as to whether or not one will commit their entire stack.


[/ QUOTE ]

I completely agree with this. A234 can often have better "bet"-ability. It would have been more accurate to say that 3477 is a stronger hand, but not necessarily a better one to have.

My point was more that most people don't accurately shift the their assessment of the intrinsic value of the hand, and that can be exploited.

I would be very interested if you or Buzz have data on the "% time you get a flop you can continue with" for various hands, as that is one of the next topics I'm interested in.

--Greg

[Cleveland Guy]

Greg- Great post - you put a lot of thought into it, and there are some good points.

A couple things I have noticed and wanted to ad.

First - you see a lot less re-stealing in O/8 than you do in HE. If you raise, and the guy re-raises he has a hand. And he probably has a monster like AA2x - so be careful when facing re-raises

Second - Make sure you steal with hands that can play both ways. I hand like JT25 - DS or TT25 is a perfect example. While not a hand to raise with or play early - you now can steal with this late - and if you are caught will have many chances to at least keep half the pot. a 52 low isn't going to win the low often in a ring game, but heads up with a raise pre-flop you will keep half here a great majority of the time.

Another thing to think about stealing that is different from HE is knowing where you are after the flop. If a person defends your steal and you hold true junk - it's easy to muck. Or if he is tight, and you have 2 high cards and he bets into you - you are behind. Or you are defending, and hit top pair on the board - you are good more often than not.

In 0/8 this isn't always the case. Even if you are defending your blind to a "steal" with a hand like KQ53 - DS (K and Q) and the flop comes Q76 - 2 suited, but non that you have, all you have is top pair with a gut shot, and a medicore low. this is a tough hand to to to war with - even if you might be ahead now on the high.

I like to steal near the bubble alot - but remember this is much more of a post flop game than HE. So while blind stealing at the 100/200 level can add a lot to your stack around mid tournament or bubble time - don't get carried away - you don't want to be putting too much of your stack in that you are pot committed to a medicore hand, as you can be way behind after the flop even with what looks like a good hand - A3356 - DS. If it comes all high cards, and non of your suit - you don't want to feel the need to put the rest of your chips in.

One final thought - is you need to look at the BBs stack a bit differently here. I think in HE - a guy will gladly, and should fold a stack often even if he is at the 5xBB level - say blinds are 100/200 and he has 950 before posting. He could and often should fold his last 750 when you push on a steal.

But now lets look at in in O8.

Blinds are 100/200 - you open steal for 500. He only has 750 left, but he has a flush draw, a couple low cards, and a connected high cards - lets say QJ62 - with 1 suit. It's not a great hand - but it's not really that big a dog vs. anything you can have - so he would be right to take a gamble and push back here - forcing you to call on pot odds.
So you need to look a bit more at the BB or a short stack when you are making this play, cause they should be playing back at a bit earlier point, so to speak than in HE.

[bodie]

I think this is a really interesting thread, and I'm glad gergery started it....I'm hoping Buzz checks in with his opinion too.
I have only played in live O8 tournaments, and have only played in seven so far. These are the first tournaments I've ever played in and I've finished in the money five times, finished fourth the last time I played. I can say that I've learned something in each tournament about strategy, what sorts of hands to play, and how different the tournaments are than ring games.
I've found that the best hands seem to be ones which have both high and low potential. I wouldn't really want to have to play 3477 - though like any shorthanded hand, it could work for you. In fact, oddly enough I lost a fairly big pot with A277 - the flop was 733 - a low was never enabled and I was beat by someone holding 37xx. Oh well. A234 is limited, but if a low comes you're going to be pretty secure for at least half of the pot, and the Ace could theoretically help too for part of a high.
Stealing seems to depend a lot on position - I've had many successful steals if you can bet when the board pairs or a flush becomes enabled, thus scaring off others who don't want to take a chance that you might not have it.
Some of the things I've learned (so far):
a) late in the tournament, be careful of completing the small blind unless you're prepared to have the BB raise you
b) it's important to create a strong table image early so if you bluff later other players believe you
c) as it gets closer to the bubble, don't bet indiscriminately
d) as gergery said, if you get stuck, don't be afraid to defend and hope that the board helps you - strange things can happen in O8

[Tom Bayes]

Great post!!! You did a great job (better than I would have done) in explaining how and when to blind-steal in omaha/8 and brought up a few points that I have missed.

The omaha/8 SNGs are a great microcosm of bigger tournaments-you just reach the point where you need to turn off your "tight-see some flops, play multi-way pots and wait for scoop situations" switch and turn on the "loose-aggressive raise raise raise isolate and steal" switch. How successful you will be in blind stealing is often up to the whims of what seat you have in relation to the weak-tights who can't adjust to shallow stacks and/or short-handed play.

[Phat Mack]

[ QUOTE ]
I would be very interested if you or Buzz have data on the "% time you get a flop you can continue with" for various hands, as that is one of the next topics I'm interested in.


[/ QUOTE ]

There have been some posts on "% of good flops" for specific hands over the years. Most recently, Buzz posted some work on high hands, but I don't think there have ever been any compilitations of large numbers of hands. I'll try to search my databases tonight.

Also, last summer, eleaticus was posting some interesting work on shorthanded O8 hands on RGP.

It might be interesting to have a thread on how to approach this problem. % of good flops is good to know, but so is % of bad flops. Good flops allow me to continue, bad flops allow me to fold, mediocre flops get me in trouble. So from my perspective, it is both the amount and the quality of information that a hand can generate on the flop that is important. Put another way, I believe that hands that can flop big, but that I can easily get away from on the flop are more valuable than some hands with higher EV's. A2TT, to cite an example from another thread, is a hand I deem extremely powerful for this reason.

Of course, as this information is so dependent on playing style, it would be difficult to present objective and universally-applicable data. I did some work on this five years ago when I was playing a lot of shorthanded NLO8. I'll look around and see if I have any useful data.