AK vs "all in" in early stages

It would be a clear fold for me.
AK isn't worth risking your tournament life on the first hand.
AA and Kk would be a different story.

tnx 4KYOU

it happened 2 time to me yesterday.. And I was thinking a lot about it ...

he can have AA,KK but if i have those hands I would try to make some callable raise.. so I extract more chips..

but maybe not all people think like that

thx for reply

You should not fear getting knocked out early. There is no benefit to busting out 5th instead of 9th. In fact, if you are a winning player, you should prefer busting out early if you must lose, since that saves time you can use to enter another tournament.

The question is whether AK has enough equity. With no dead money, the ICM says you want to have about 55% equity to get all-in on the first hand. There is always dead money, which usually reduces the percentage you need slightly. In a $1 tournament, I think AK clearly has enough equity against an unknown player because people do crazy things with trash just to see what happens, and even if they think they have a strong hand, they don't know hands like AJ aren't strong for 50 big blinds.

Originally posted by pzhon
You should not fear getting knocked out early. There is no benefit to busting out 5th instead of 9th. In fact, if you are a winning player, you should prefer busting out early if you must lose, since that saves time you can use to enter another tournament.

The question is whether AK has enough equity. With no dead money, the ICM says you want to have about 55% equity to get all-in on the first hand. There is always dead money, which usually reduces the percentage you need slightly. In a $1 tournament, I think AK clearly has enough equity against an unknown player because people do crazy things with trash just to see what happens, and even if they think they have a strong hand, they don't know hands like AJ aren't strong for 50 big blinds.

This. And also, you can discount the amunts of kings and Aces from their range, so I would just call.

You should not fear getting knocked out early. There is no benefit to busting out 5th instead of 9th. In fact, if you are a winning player, you should prefer busting out early if you must lose, since that saves time you can use to enter another tournament.

I dont wanna to fight, but I think its stupid.. .

the Truth is that I have no expirience in SnG.. .but it happened to me that I have 2 or 3 BB when there is 6+ players, and then to finish ITM ... so whenever I can saves 100 or 200 chips i take it.. I dont want to throw them away..

Maybe u r right, I dont know.. I want to hear reasons + or - ... So I decide what is better

thanks to all for helping me, DjolusVicenus

If I'd no knowledge of the other player I'd fold. I'd assume the all-in was a medium pocket pair which would beat A,K most of the time. If it was an aggressive player who tended to raise a lot I'd probably go all in. Although it's funny how often the guy who usually raises with rags tends to have the aces when I do that.

Originally posted by DjolusVicenus

You should not fear getting knocked out early. There is no benefit to busting out 5th instead of 9th. In fact, if you are a winning player, you should prefer busting out early if you must lose, since that saves time you can use to enter another tournament.

I dont wanna to fight, but I think its stupid.. .

the Truth is that I have no expirience in SnG.. .but it happened to me that I have 2 or 3 BB when there is 6+ players, and then to finish ITM ... so whenever I can saves 100 or 200 chips i take it.. I dont want to throw them away..

Maybe u r right, I dont know.. I want to hear reasons + or - ... So I decide what is better

thanks to all for helping me, DjolusVicenus

I think pzhon was referring to the "AK is not worth risking your tournament life on the first hand" statement.
Imo what he means is that you should not be afraid of being knocked out early, bearing in mind that you have to be a little risk-averse. But you shouldn't go to extremities (such a word exists?) and take this spot because it's +EV.

so.... how i understand... Its gray area...

If u are a little bit of gambler its ok to go all in... but, if u dont wanna to take risk its fold?

No, it's not a gray area. Some things are a matter of style, but this is not one of them.

If you aren't willing to get all-in with AK at level 1 (50 big blinds) against an unknown player in a $1 SNG, this is a clear mistake. It doesn't mean you automatically lose, but you are passing up a clearly profitable opportunity, one which may be worth the average profits of more than one tournament.

The way almost all serious player measure their progress in a SNG is using the Independent Chip Model (ICM). This lets you quantify how much equity you need to call all-in. Then you can estimate your opponent's range, and estimate how much equity your hand has against that range. See my video "Calling All-In" and my program ICM Explorer.

It is common for people to say, "Don't get all-in early, you can lose tournaments at level 1 but you can't win them." In fact, accumulating chips early is extremely valuable and lets you use your skill advantage or survive bad luck later. It is common for people to say, "You must be up against a pocket pair. AK is an underdog against a pocket pair." In fact, this is a $1 SNG and people get all-in with ridiculous hands. It is common for people to play losing poker. If you want to win, and not just a little, you need to go beyond rationalizing whatever play feels right. You need to learn to figure out whether plays are really good or not, including ones which may seem uncomfortable at first.

This is a really interesting question. This specific instance you can probably stack off with AK, for the reasons above. Specifically that they stacking off range is generally wide enough and doesn't contain AA/KK as often as it should.

However, the broader question how much equity do you need versus their range (if you knew it) do you need? This definitely makes this question not only apply to AK but every hand for the early game. This is certainly a grey area, and I hope I can show that.

Assume there is no dead money to make our math easier. The question is, is it OPTIMAL to call with every hand that has at least 50% equity against their range? The answer is 'it depends', what does it depend on? A lot of things. Most importantly though, it depends on your skill edge in the tournament. If we do get it in with exactly 50% equity, we better hope we get at least double the skill edge if we win (unlikely)!

So how can we quantify skill level? Simple, we use it all the time. ROI. Suppose your ROI in SNGs is 10%, what do you think your ROI when you double up, and a player is knocked out is? 20%, 30%, 90%? If you are curious, you can probably look up this number in your HEM, if you play SNGs that start with 75bb, look for games where you played hands 140bb+ deep, and compare your ROIs.

I hope you can see what I'm getting at. We actually have enough information to calculate your EV of calling it off with 50% equity. We really should be taking ties into account, but we'll ignore that for now. We'll pretend the tourney buyin is $1, just for simplicity (so your ROI expressed below is your expected gross profit from the tourney).

Let's calculate your expected gross profit from each instance when you call ($EVc) and when you fold ($EVf)

Let your normal ROI be R (expressed as 1.10 for 10% ROI)
Let your ROI when you double up be D

$EVc = 0.5*D+0.5*0 = D/2
$EVf = R

So we need to call whenever $EVc > $EVf. Meaning your ROI when you double needs to be at least double your normal ROI. I'm not talking about 10% to 20%, I'm talking 10% to 120%! Do you really think your ROI goes all the way up to 120%?! Maybe it does if you play the big stack really well. Maybe it's even higher meaning you can call off with less than 50% equity here! But chances are it's not even close to this, and you can't call here.

So how much equity DO you need? Well theoretically,

If our equity is e (expressed as a number between 0 and 1, again we are ignoring ties)

$EVc = e*D+0.5*0 = e*D
$EVf = R

And we should call whenever

eD > R or
e > R/D

For most people, I would guess, their ROI is something like 10% and when they double, it's something like 80%. Meaning that in this spot you need,

e > 1.1/1.8 = 61.1% equity to call in this spot!

I hope this wasn't confusing, and you found it interesting. I hope I showed that this issue is actually much more complicated from a theoretical point of view than it looks. Calling it off with AK is actually much closer than it looks, it's not enough to just be the favourite, since we have to take skill and future game into account!

It's good to think of such things, but you are reinventing the wheel. There is already a well-established theory of tournament equities. See my video on the ICM. The ICM is not the end of the story, as I mention in my video, but please build on what is already done unless you have a clear reason to discard what most professional players use.

For most people, I would guess, their ROI is something like 10% and when they double, it's something like 80%.

I think your guess of 80% is wrong. So is your 10% for "most people." Together, they mean that you change from being a much stronger player than your opponents to a weak player. I don't think doubling up forces you to play badly later. If you have the skills to get a 10% ROI before doubling up, you should still be allowed to play skillfully after you double up. If you guess a different number, you would get a very different value.

As I stated, the ICM suggests that to get all-in with no dead money, you want to have about 55% equity. With typical amounts of dead money you might only need 53%. AK has way more than that against the hands you will see pushed in the first level of a $1 SNG. You will see hands like AQ, A5, KQ, JT, 6♥9♥, etc. Of course you will sometimes see AA, but that is much less common. There are so many more combinations of bad or mediocre hands people can misplay, and do, and they usually play AA differently.

How do you get an ROI of 10%, which means outplaying your opponents not only enough to make back the rake, but 10% more? The way you do this is by recognizing good opportunities when they occur, not by turning them down wishing for even better opportunities.

Hi Guys

This is very interesting and well over my head, I've still got my L plates on! Will definetly check out your videos pzhon to gain some extra knowledge on ICM. For me, I don't think its a matter of feeling comfortable making this move or not, I've lost plenty of money in the past going all-in with AK early blinds. I now prefer to be patient and hope to take money from others who overvalue AK but I am here to learn and so I'm open to change.

I'm not ignoring ICM. The reality is that ICM is just ROI in disguise, they both calculate equity when you push and equity when you fold.

HOWEVER, ICM does not take large stacks OR skill OR future game into account! And ROI does. Especially in the early game, ignoring skill and future game is very costly.

For most people, I would guess, their ROI is something like 10% and when they double, it's something like 80%.

I've used these number just for the sole reason of having concrete numbers and being able to come to a conclusion. You can look these number up in your own HEM. I've provided a formula that does not require my bad guessing.

e > R/D

We can use HEM to estimate our true theoretical ROI, and theoretical ROI when we double up, and use that to see what equity we need.

I think your guess of 80% is wrong. So is your 10% for "most people." Together, they mean that you change from being a much stronger player than your opponents to a weak player. I don't think doubling up forces you to play badly later. If you have the skills to get a 10% ROI before doubling up, you should still be allowed to play skillfully after you double up. If you guess a different number, you would get a very different value.

I'm not sure If I was clear here with my ROI notation. By 80% mean +80% (really good ROI) and 10% means +10% (good ROI). This is net profit and includes the buyin. Both are good ROIs, and don't imply weakness at all. 10% means you make 1.1 buyins per game. and 80% means 1.8 buyins per game - on average.

Originally posted by Optroot
The reality is that ICM is just ROI in disguise, they both calculate equity when you push and equity when you fold.

No. The ICM is not "ROI in disguise." The ICM is a model which predicts a vector of finishing place probabilities. From this, you can calculate an equity, and your risk aversion (as my program ICM Explorer does), but the ICM is strictly more complicated than either of these. It gives you many predictions and assumptions you can test.

When you pull out an unsubstantiated guess about your equity in some situation, this is not easily testable. It does let you rationalize whatever play you want, which is not so great if you are trying to learn to play better. A better method is to start with a consistent framework which factors in many of the complexities, and then try to make sensible small modifications.

HOWEVER, ICM does not take large stacks OR skill OR future game into account! And ROI does. Especially in the early game, ignoring skill and future game is very costly.

If you watch my video on the ICM, you will see that I mention things like this. However, this question has come up many times in the past, and people consistently overestimate how much control they will have in the rest of the tournament after folding. Poker novices commonly talk about how they would not want to gamble their whole stacks as 2:1 or even 4:1 favorites on the first hand of the WSOP main event. This is a common error.

It is true that some of the best players in the world, while playing in soft games, can turn down some slight edges (gambles which gain ICM equity) which come with a lot of chip variance. (And yes, I can modify the ICM slightly to calculate which gambles are too big -- SNG Wizard and ICM Trainer do this incorrectly.) Far more common are people who rationalize avoiding gambles with large edges, like getting all-in with AK against an unknown player in a microstakes SNG for 50 BB, in the hope that this will make them great poker players. This is a mistake which needs to be corrected in order to become an expert.

I've used these number just for the sole reason of having concrete numbers and being able to come to a conclusion. You can look these number up in your own HEM. I've provided a formula that does not require my bad guessing.
...

I'm not sure If I was clear here with my ROI notation. By 80% mean +80% (really good ROI) and 10% means +10% (good ROI). This is net profit and includes the buyin. Both are good ROIs, and don't imply weakness at all.

I'm not confused by your notation, although I suggest that you not reuse "ROI" because you are talking about equity, not a return on a repeatable investment.

I would like to know how you query HEM to get your stats after reaching a particular stack size. I think I see how to write a Postgresql query for this, but is there a built-in method? If not, what query did you use?

Do you lose chips on average after doubling up? I'm pretty sure I don't, and that I continue to gain ICM equity (which is harder to read from a database), too. This is part of why it raises warning flags when you guess equities which suggest that doubling up sharply cuts or reverses your skill advantage. In a similar past discussion, Jerrod Ankenman said, "... everyone realizes that you get to apply your super-duper skill to the second $10,000, too, right?"

For simplicity, ignore the rake. With 10 equal stacks, the ICM says you average 10% of the prize pool. To have an ROI of 10%, you have to outplay your opponents to average 11% of the prize pool. The ICM says that after doubling up with a little dead money, you deserve about 18.9% of the prize pool. Your guess is that you will now be outplayed by your opponents by an average of 0.9% of the prize pool. You are assuming that your skill advantage is reversed by doubling up. This doesn't make sense to me, and it is how you are coming to a conclusion that I don't believe. The ICM suggests that you need 10/18.9 ~ 53% equity. This rises only slightly (by 1-3%) if you assume that you have a big skill advantage both before and after doubling up. 61% equity is way off.

My opinion is if it's +EV to call an all-in during the early stages against your opponent then I would definitely call.

Remember; poker is a game of high variance with small edges, therefore you want to seize all the +EV spots you can

I regret using those numbers in my example. I only wanted to illustrate that a number could be calculated, not that the conclusion is that you need 61.1% equity. I'm not proposing that you necessarily lose skill edge when you double up. It could be the case that ROI when you double up is exactly as ICM predicts, I'm not trying to argue that it shouldn't be. But there are certainly some cases when ICM simply isn't strong enough to calculate the correct response, specifically in cases with skill edges (or leaks) for particular players with different stack sizes, that they may only have with a particular stack size but not another.

There are players who are just average but potentially play the big stack really well, in that case, they should make a slightly bad call if their upside is that they get to play the big stack and increase their edge dramatically, enough to compensate for their bad call. A player could also get too passive with a big stack in which case taking a break even (according to ICM) gamble is actually -EV for them.

The ROI quantifies this skill edge (or leak), in a way that cannot be expressed by ICM. I encourage everyone to lookup their ROI when they double up early to know how well they play a big stack. If their ROI is lower than ICM predicts (+edge), then you have a leak in big stack play. But while you still have that leak, you need to compare your ROIs in order to determine what equity you need to call. If your ROI is higher than ICM predicts this implies that the field collectively has a leak that you exploit, which means you should be more eager to take this gamble even if slightly -EV according to ICM.

I would like to know how you query HEM to get your stats after reaching a particular stack size. I think I see how to write a Postgresql query for this, but is there a built-in method? If not, what query did you use?

I'm not at my 'poker machine' right now. IIRC, it's Advance Filters > Stack size, then enter an appropriate size in bb. It should let you see results of all tournaments where you were that deep at some point.

Is that in HM2? I don't see an "Advanced Filters" tab in HM1.

If you simply look at your tournament results for the tournaments where you had over 100 big blinds, this will bias the results. It will not give you an unbiased estimate of the historical value of doubling up (to 100 bb). Sometimes you reach 100 bb by overshooting, say by tripling up. In addition, reaching 100 bb at level 2 should be much more valuable than reaching 100 bb at level 1 because it is a lot more chips.

What I was suggesting was to add together the results of all hands where the tournament ID is equal to and the time stamp is at least as large as a hand where you have at least 100 bb at level 1. This will tell you whether you gain chips on average after doubling up, without overvaluing the times you overshoot 100 bb.

Honestly, if a person showed 32o the first hand of a 9 man and shoved, I'd still fold AK. I don't fold because if ICM, I just fold because I don't like calling.

If you are okay with losing the first hand half the time (on average probably), then go ahead and call.

If you are like me and are not a fan of this scenario, I'd recommend fold.

Cheers to 1st day on PStrat:)