Tuesday, July 05, 2005

 

Don't Bluff When You Will Be Called

There's an interesting situation that comes up more often in speed tournaments than in the conventional games, and I must admit that when I worked out the Maths I was surprised by the result. See what you think.

This is example is close to a real situation I was in a couple of days ago, altered slightly just to make the calculation simpler. Alterations don't change the basic idea. Playing 1000-2000 blinds, I have 8000 under the gun 7-handed and I pick up A6.

We're going to have to make some assumptions to work this out ; I'm going to keep them reasonable and not get carried away about how accurate the result is, and anyway the important point is that I make the same assumptions in the two cases.

Case 1) the other players all have enough chips to fold to the raise, including (most importantly) the big blind. If I move in, assume that I have a 50% chance of being called, that if called I will be a 2/1 underdog to win the hand, that I won't be called twice and that the caller will have enough chips to cover me. These assumptions are necessary to simplify the calculation, I don't believe that they undermine the central point. Anyway, my EV is now :

Everyone passes : Prob 50%, Result +3000, EV +1500
I get called and lose : Prob 33%, Result -8000, EV -2640
I get called and win : Prob 17%, Result +10000 *, EV +1700
...............................................................Total EV +560.

*10000 is an average value allowing for the fact that it's different according to whether the caller is in the blinds or not.

So we have a +EV play. Of the three outcomes, in one of them we're eliminated, in one we increase our stack sufficiently to withstand another round of blinds, and in one we double up and are in good shape. Whatever happens we're not going to be desperately short-stacked in the near future. This is almost as important as the +EV IMO.

Case 2) Everything is the same except the big blind is all in for the 2000. Now there are more outcomes, as follows :

Everyone passes and I win against the BB : Prob 30%, Result +3000, EV +900
Everyone passes and I lose to the BB : Prob 20%, Result -2000, EV -400
I get a caller and I lose the whole pot : Prob 33%, Result -8000, EV -2640
I get a caller and I win the whole pot : Prob 14%, Result +10000, EV +1400
I get a caller and I win the sidepot only : Prob 3%, Result +4000, EV +120
...............................................................................Total EV -620

The above probabilities assume that I'm a 60% favourite against a random BB hand with A6, and that it's quite unlikely for me to beat the better hand that calls me but lose to the random BB (thus winning the sidepot only)

The result surprised me because I thought it would be break-even. But it's much worse than that. The play is -EV. There is also a 53% chance that we will either be eliminated (33%) or very short-stacked (20%) as a result of this play. If we get heads up with the BB and lose this hand, we'll have to post 2000 in the BB next hand leaving us only 4000 left and an almost mandatory call of any raise.

Bottom line : when the Big Blind is all-in or so short-stacked that he will have to call your raise, the EV of a steal play goes way down. This also applies (although less so) when the SB is committed. Also look what happens if we don't have an Ace, if we have another hand like K3 or 98 or almost anything half-decent. In Case 1, our equity is much the same because we'll still be around 30% if we get called, unless we're unlucky enough to run into an overpair. In case 2, our weaker hand is less likely to win against the BB, probably not much more than 50%, which makes the figures even worse.

So, in any tournament (but especially in a speed tournament where this is more likely), watch out for the situation where a player in the blinds has to call you and up your raising requirements dramatically in this spot. Make sense really, first rule of poker - don't bluff when you're going to get called. And think ahead. In the situation where you're in second position with 8 SBs and the UTG player, who can just about cover the blind, passes - this is basically your last hand. You don't really have an extra hand before the BB like you usually would. So make your move now. Stealing is also less lucrative when a crippled stack has called in front of you, although it does depend how crippled, obviously if the all-in only has 200 with 1000-2000 blinds it doesn't matter much.

Perhaps the most interesting corollary of all is the situation where there is no small blind, the big blind is all in and it's passed to you on the button. What hand do you need to call, even if the big blind is only in for a small fraction of the blind ? Should be obvious if you think about it. But I wonder how many people would call with any hand.

Speed tournaments are great for honing your short-stack strategy. Situations that would be unusual in a normal tournament occur more frequently, allowing you to develop a strategy that might have much better EV when the situation does come up for big money. If I may quote the New Testament (or Harrington On Hold-Em Volume 2 as it is also known) "Be advised that playing correctly around inflection points [that is in short-stack situations] is the most important single skill in No-Limit tournaments". Quite so. And you might be surprised how many highly-regarded players don't play that well (at least compared to their reputation) with a short stack.

Comments:
I suspect that part of the reason the hand falls away so badly is that Ax gets worse very quickly against multiple opponents. How does QJs or KQo perform?


The second (and I suspect more important) reason that the hand falls away so badly from Scenario One to Scenario Two is that in case one you are assuming that you get no more than one caller, whereas in case two you already HAVE a caller (the all-in big blind) and you are allowing for the fact that you might get another one.

Throw a 10% chance into case one that you will get one caller and that the big blind will call as well (althernatively, assume that no-one will call in Scenario 2, because the Big Blind counts as your "one" caller), and I suspect the EVs close up significantly.


Pete
 
"althernatively, assume that no-one will call in Scenario 2, because the Big Blind counts as your "one" caller"

No, that can't be right. The whole point is that the BB being all-in means you already have your caller, and you now have to fade almost the same odds of a second caller as you did for one caller in case 1. Similarly the probability of 2 callers in case 1) is not much less than (2 callers plus the BB) in case 2). I'm even more sure about this than I was before I think ! Although it would probably be fair to reduce the probability of "another" call in case 2) to 42% to allow for one less player.

As for KQ I would expect it to do better in both cases, but the difference would still be as pronounced.

The fact remains that in case 2) your chance of winning the blinds has dropped from 50% to 30%, and there is also an ugly scenario where you drop to 6K which does not exist in case 1) where the only three outcomes are double up, win the blinds or elimination *. I'm fairly sure this is a real, and significant, factor.

Andy.

* unless you are called by and lose to a smaller stack, which is also a factor in both cases.
 
We are arguing from different starting points. I agree that empirically in Scenario 2 you have to add in the chance of a second caller at the probabilities you supply.

My point was that it should have been no surprise that Scenario Two is far worse than Scenario One, because the BB "functions" as the first caller. Empirically, therefore, the raise in Scenario 2 is shit. In a sense, I was explaining why it was shit, because the two scenarios are not mathematically equivalent (one is a head-to-head if you get one caller, while the other is a multi-way if you get one caller).


In limit, Sklansky advises (correctly) that when there is a poster behind, you should be less likely to raise first-in rather than more likely, saving your raises for real premium hands (this is one of the major flaw of mid-limit players, raising in this situation when they shouldn't). This is because, with someone already single-bet committed, you are far more likely to get a multi-wayer.

This has a direct analogy in the tournament situation.

So, here's an interesting one. What kind of hands get a BETTER ev in this situation? The answer, of course, is the real killer hands like AA or KK or even QQ/AK.

Suppose you have 14,000 in MP2, BBs are 1K/2K and the BB is all in for 2K.

The player on your right raises to 6K and has you covered.

What range of hands do you want to reraise him (or flat-call, if you are so inclined)?

Second scenario: BB is all in for 2K and you have 9K. What rank of hands would you want to go all-in?

I would think that in Scenario 2 any pair 8s or above, plus AK and maybe AQs, would get a better ev from the fact that the BB is all-in than they would if the BB was not all-in.


Pete
 
Aha, OK. Perhaps the surprise is that I was so surprised :-)

I was though. About a week ago I would have thought "so what, I've got an Ace and he's got random crap anyway". Always nice to learn something even if it does mean I was being a bit thick (or thoughtless at least) before.

Andy.
 
'I would think that in Scenario 2 any pair 8s or above, plus AK and maybe AQs, would get a better ev from the fact that the BB is all-in than they would if the BB was not all-in.'

Quoi? Je ne le comprend pas. If I have AA (any pair 8s or above) with 9k stack, I clearly want the BB to have chips & not be all-in.
Typically with the BB chipped up I will either scoop the blinds or more preferably get my stack all-in against another stack (or two).

That's a whole lot better than the likely 85% chance of winning 1 the blinds - as in the all-in BB case.

maple leaf
 
I don't think that I am thinking straight in any way, shape or form at the moment, and my powers of expression seem to have gone the way of chaos's. Perhaps I should get into this adderall stuff.

Okay, what I assumed (but which is not necessarily true) was that because the value of some hands went down when the BB was all-in, then the value of some other hands must go up. However, this is not a zero-sum situation. It's quite possible that ALL hands go down in EV in a tournament situation when the BB is all-in.

Not that the situation is as simple as you (Maple Leaf) suggest. With the BB all-in, players behind me after I raise are distinctly more likely to call (which is why we have the "you will always have one opponent, and may well have two" scenario). I can see now that this very proviso could make everything less EV, although that's an uncomfortable conclusion -- that there is no way that you can turn the situation to your advantage.

So, let's suppose that player "X" behind me when I am raising the BB is likely to call me on a range of hands one Sklansky level lower than he normally would, and will reraise me on hands one Sklansky level lower than he normally would. And let's assume that I have AA.

Taking Andy's original hypothesis:

Everyone passes and I win against the BB : Prob 43%, Result +3000, EV +1290
Everyone passes and I lose to the BB : Prob 7%, Result -2000, EV -140

I get a caller and I lose the whole pot : Prob 9%, Result -8000, EV -720
I get a caller and I win the whole pot : Prob 36%, Result +10000, EV +3600
I get a caller and I win the sidepot only : Prob 5%, Result +4000, EV +200
...............................................................................Total EV +4250


Compare that with scenario 1


Everyone passes: Prob 50%, result = EV +1500
I get called and lose: 8%: result = ev -640
I get called and win 42%:
Result = ev +4200

Total ev = +5060.

And, you are correct; even with the best hand, the EV is less when the big blind is all in.

Glad that we've sorted that one out.

Pete
 
*my powers of expression seem to have gone the way of chaos's*

so it would seem - it's " chaos' ". Peter, you only have to ask if you need me to elucidate. Still let's there's plenty of upside in being less than lucid.

I've not looked into the math's of this(which will inevitably be rogue, but worth doing), but I'd sure be gutted if the BB was all-in when I held AA. Though, of course, the closer to utg the less unhappy I'd be.

Reading this from PB..

'In limit, Sklansky advises (correctly) that when there is a poster behind..................to get a multi-wayer.'

He may be correct, but I'm far from convinced & it sure aint wholesale.

If I'm utg (ish) with marginal hands like KQ,Kjs,A-Jo, A-9s, 66 then I'm more encouraged to raise and rightly so imo.

I don't like arguments that justify a view by only highlighting the increased downside of the position. Yes, I am less likely to win the blinds unchallenged, but when I do win them unchallenged, I win more. Yes, I'm less likely to win the pot HU against the BB, but once again, I win more when I do.

And of course the pot is more likely to be multiway than without the poster behind, but when it is multi-way I'm likely to be up against weaker hands than in a 'normal' multi-way pot. It's all about trade-off.

Consider there are two blind posters, are you folding A-Jo UTG. Has the value of your (marginally profitable) hand gone down or up because of their posts? i.e. because the likelihood of a multiwaypot has increased. Sure I'm out of position, but out of position against weaker hands, I'm not comfortable with it, but much more comfortable than O the EV passing.

I can see there are some hands I would be less likely to make a raise with, particularly small pairs, weakish As in latter positions, but a blanket < approach sure aint for me, and it is typical Sklansky.
 
If there are two blind posters, yes, I definitely pass AJo utg. In fact if there is one blind poster I would pass it utg. The value of my hand has gone down because of the posters.

But I wasn't writing about blind posters, I was writing about a Big Blind who was all-in, and using the Sklansky point as a corollary.
 
'But I wasn't writing about blind posters, I was writing about a Big Blind who was all-in, and using the Sklansky point as a corollary.'

I realise that, but it was nevertheless an opinion you expressed.

'If there are two blind posters, yes, I definitely pass AJo utg.'


Not even a limp? Well we're poles apart. Passing isn't in my handbook here, though I can see some merits of limping.

Of course this isn't something that can be proved with an argument, but I'll try and something anecdotal.

Ring game: utg AJo. EVERYONE has has posted blind. Do you still fold? If a couple of posters have done so much damage to your AJ, then what 8 posters to, say, AQ?

I'm sure you wouldn't recommend passing. If so then clearly there is no strict relationship of inverse-proportionality between a (strong)hand's value utg and the number of blind poster's acting after it (since you'd pass for two).

Now consider the two blind posters: have their random hands gained value from your utg raise with AJ? I'd contest that they haven't and, if I'm correct, where has it lost value to? Well I'd suggest it has some value to the AJ and to the hither to unknown random hands with position & no cash committed.


'The value of my hand has gone down because of the posters.'

It isn't apparent to me that this is true, it is apparent that it is less likely to win with additional posters. The pots, though, will be bigger.

chaos
 
A poker calc Monte Carlo simulation gives you 18% equity against seven opponents. Since I am out of position, I would say that, yes, AJ is a pass against a large number of blind posters behind.

But let's assume that not everyone calls (although, to be frank, it's hard to see how they can fold given the pot odds they are getting to a single bet).

With four callers in range 1 to 8 (Sklansky) and one random caller, your equity is 19% in a six-player hand.

Once again, given that you are out of position, the 2% edge you have over the "average" is not enough.

With no posters, and assuming that players behind you will call a raise cold with about 5% of hands (groups 1 to 3), there is 2/3 chance that it will be passed round to the blinds. If one of these calls with hands in range 1 to 8, you have 58% equity AND YOU HAVE POSITION.

If you get called (or reraised), you have about 38% equity and are out of position.

If the reraise gets the Blinds to fold (and they would have called otherwise) you would probably prefer a reraise to a cold call, because if the Blind comes in to a single raise with hands ranged 1 to 8, your equity plummets to 28%.

So: Lots of posters, equity about 19% with five opponents.

No posters: EV 58% (one opponent on whom you have position) 2/3 of time,

EV 38% 1/6 of time (one opponent who has position on you)

EV 27% 1/6 of time (two opponents and you are in the middle) .

So a lot of this depends on the value that you ascribe to position. But my (empirical) evidence, is that once AJoff becomes involved in a multi-way hand, it does very badly indeed. You hardly ever know where you are.

In this sense, being OOP with AJ is far worse than being OOP with, say 87s.

This is not like an all-in situation (the topic of which you appear to have hi-jacked). You will have decisions on all streets subsequently. I can't think of any worse scenario likely to lead to burng £10 notes than to have AJoff out of position with five opponents who could have virtually anything.

Pete
 
Aplogies for the delay, I've been away for a few days for my second visit to cardiff's millenium stadium in as many weeks. It's been hot!


I can't corroborate these numbers, and naturally, they simplify too much to draw out significant conclusions, but, granted they can give hints. That said, some additional comments:

Position doesn't concern me as much in large multi way pots partly because it is much harder for players 'in position' to manipulate pots, because players are harder to shift and partly because position has a greater tendency to float around in these pots.

Playing the aggressor still has value, albeit considerably reduced.

Add to that that I suspect that there is more of a tendency tendency to be paid off with AJ (w.r.t EV), than to pay off in these large multi way pots- than the smaller multi-pots or at least it is easier to get away from AJ.

Also consider that you have, hopefully, a skill advantage over the field too.

I would certainly have a significant expectancy of folds for many players particularly those who act immediately after the utg AJ and those much further away who may easily have to call 2 bets with many hands that wouldn't justify it. Suggesting they have pot-odds to call would be fine assuming the probability of a 3 or 4 bet was particulalry low - but Id suspect it to be far from low.

If there is a decent expectation of dead money your apparent target, for the 6 player scenario, of 16.67% is too high.

'With no posters, and assuming that players behind you will call a raise cold with about 5% of hands (groups 1 to 3), there is 2/3 chance that it will be passed round to the blinds.'

Two in three looks far too high, what about the BB?

'This is not like an all-in situation (the topic of which you appear to have hi-jacked).'

I thought Sklansky's argument that you delivered was directed towards limit. Besides, it appeared that you had concluded things before I entered the proceeding, so I'm struggling to see how I've hi-jacked the thread. Nevertheless, it's not my blog, so I'll bear it mind.

'You will have decisions on all streets subsequently. I can't think of any worse scenario likely to lead to burng £10 notes than to have AJoff out of position with five opponents who could have virtually anything.'

Opponents having 'virtually anything' encourages rather than repulses me. Yes they're harder to read, but they are playing with crap - it's a trade-off. As I said if you think this situation burns off tenners then we are poles apart.

It is likely to depend largely on how you feel playing out of position & your ability to maximise value in this situation. If you play the hand too cautiously because you fear what players may have then you will rapidly turn profit into loss.

You could probably try this out on some of the smart AI software out there and get some results closer to real play.

Also, even at the risk-averse 2+2 forums I'd be very surprised to see them rec mucking here. Naturally I can't see what players muck, but I have certainly noticed a lot of top players take more, not, less risks in these situations. And, I'd have little doubt they'd not muck the AJ utg with two blind posters acting after them. Not that that's any proof, it could be a common leak amongst these players, however, it suggests to me that it is much more likely to be + than - EV.

chaos
 
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?