How do I which hands are profitable to open?

Let's say i'm on the button, if I fold, I lose nothing, how do I know which hands will make more than nothing?

The only way I can think of is if you had a database of hands, you could come to the conclusion that it's probable that "x" hand makes more than nothing.

Obviously you can assume that opening aces is always profitable, so if you start there and you you keep going down the line, KK, AK, A10, K9, Q9, etc...

How do you know where to draw the line?

People often assume it's profitable to open hands such as KQ suited from the button, etc... But how do we KNOW for certain that it's profitable to do so?

My question is, is there a way of KNOWING for certain which hands are profitable to make the risk of opening worth the reward? (The only reward to gain at that moment is the blinds, 1.5bb, and opening is a risk of at least 2bb.) And HOW do we know?

In your response, if you decide to respond, could you say for example... "We know it's profitable to open such and such hands, because..." Then showing how we know such and such to be true.

Thank you.

'Big Data' is pretty much the only proof, and it's how poker bots that are based on neural networks and "regret minimization" built their strategies. So basically you're right: Get a tracker, play millions of hands, draw conclusions from the results.
To save time, you can just read books and forum articles by people with a lot of experience. If they say T9s is an obvious open on the button, it's because their experience has shown it to be profitable.

The more hands you open, the more money you will make with the top of your range. The bottom of your range will always lose money.

No.

The profitability of a hand is based largely on mathematics but human psychology is also an important factor. In order to know absolutely the EV of a hand, you would have to solve both the math and have total access to your opponent's thought process. Solving hard math is one thing but solving a brain is another.

So yeah, basically all we have to go on are our hand samples and collective experience. Data should not be relied on too heavily since its just averaged out across average opponents. You might raise 89o on the button against average players but its a fold against aggressive ones.

Its not an exact science. Instead of thinking hand as profitable or unprofitable, its better to think of them as clearly profitable, clearly unprofitable, and marginal.

There's some logic to playing bad hands to entice your opponent to tangle with you more so you win more often with good ones, but in order for that to work your opponent has to be thinking and adjusting, and be doing so in a way that lets you capitalize it. I prefer to just fold everything I don't think will be profitable. Making -EV plays purposefully is a slippery slope.

It doesn't matter, the bottom of your range will always lose money. Go through your database. You will see, given you have played enough hands that it's true. Opponents make decisions based on your %s, so it's natural that if you raise a tight UTG range out of AK, AQ and AJ, AJ will lose money as people won't call you with AT.

You're assuming opponents are adjusting (as he said).
You're also assuming that opponents are not being outplayed in general.

If you have a tight range and you don't pay off speculative opponents (because you read them better than they read you) and/or they do pay off your speculative hands then it is possible not to lose money at the bottom of your range (which may be those speculative hands).

..which is not to say I could do that or that it wouldn't be better to widen the range from that point.


OP: 'big data' is the best certainly you're going to get but paying attention to specific opponents will make a huge difference.
Either that or you can solve GTO [which I'm pretty certain would lose money on the bottom end of its range]

You can't really just filter "bottom of range" through a database because its different for different opponents. You also might not be taking position into account. Bad hands are certain losers overall due to being unable to defend from the blinds. From outside the blinds I really don't think the lower part of my range is -EV. If it was significantly -EV I would consider that a leak.

If you filter out the blind positions, is every hand positive ev?

I think it should be. Obviously not every hand is going to show a profit in pokertracker unless you have some inhuman sample size.

If you're saying "the bottom of your range is -EV, so you should fold it", then I completely agree. (32o UTG will never be a profitable hand in the long run).
If you're saying "Play some -EV hands in order to increase your action with big hands", I disagree.

You should only put money in the pot if your expectation is to have more money in your stack at the end of the hand than you started it with.

Since the majority of playable hands are only marginally profitable, variance plays a huge role in results. I'm absolutely convinced that 76s is a (slightly) +EV open on the button, but it might take 2 million total hands or more before I have a sample size that is statistically significant enough to be a compelling 'proof'.

Bottom of your range should always be very close to 0ev but never -ev.

I decided to mess around in pokertracker a bit. Its really hard to get a significant sample. I have around a total 100k sample on this computer for 25NL - 200NL, but that only leaves ~30k for the bottom of my range outside the blinds. I actually am a small 2BB loser pre-rakeback (my current site has gradually eliminated rakeback in the past couple years though). I was up after 20k hands though, and on the button I'm a 4BB winner (8500 hands).

Again, small, almost meaningless sample size but it still makes me think a little about tightening up outside the blinds and button.

Seems to me like some hands at the bottom of your range will be slightly minus, because they balance that loss with positive shania. In theory, you maximize your EV by opening a range of profitable hands, summing the total EV of those hands vs the SB and BB's counter strategy. Then, after noting that EV, you add another marginal combo causing a cascade of tiny changes in the SB and BB strategy. Sum the EV again, if it increased, continue to add a new combo and reiterate the steps until you reach an equilibrium where adding a new marginal combo doesn't precipitate an EV gain. The vacuum profit of your hands are irrelevant. Only the sum total of the EV of your range matters.

I think this applies to both non-optimal and optimal opponents, as even donks pay some amount of attention to what you're doing and adjust in their own way against what they think you are doing.

Yes, but you don't vpip -ev hands.

Anything you can get away with.

You do if they cause a net increase to your range EV, which seems theoretically possible/likely.

Adding -EV hands to a range reduces the total EV by definition, unless you've learned to break the rules of simple mathematics.

There's also an argument that adding -EV hands actually decreases the EV of the top/middle of your range (but not the very top). If you're playing against some loose passive nit you want them to stay in check/call mode so you can just valuetown them all day. If they see you're playing some suited garbage they might adjust and start 3betting you more often in which case you're going to have pitch a lot of marginally good hands that you would normally have seen a flop with.

I was going to post something similar to this. I don't think it would quite work like you say, though. More likely is that adding a new marginal weak hand could precipitate a net negative EV to your range by reducing the EV of other hands at the BOTTOM of your range not necessarily at the top and middle.

As a thought experiment, let's say in 100bb NLHE you open the button with a 50% range, and you know the precise EV of each hand in that range against the given SB and BB opponents. You decide to open the next most playable 12 combos, whatever that hand may be, to gauge the effects on the rest of your range's EV. It may even be a smattering of mixed frequency offsuit hands and not just one, but that doesn't particularly matter for this.

I think you will find with the new adjustments to your adjusted range, your previous "Top" hands will make significantly more money. In addition to that, what was considered "Top" before will expand slightly. As you move down your range and compare the EV of each hand to the previous EV, it will generally be a positive number down to some point. Eventually the positive differences will start to really diminish. I suspect that this will be something like the 70% point in your range, where it starts to become clear that you do not want to face a 3-bet, even if you will be able to call it.

The next threshold will be the point where you are raising with a hand that must fold to a 3-bet. I think the point where the differences become negative lies somewhere in this region. From this point the differences will be slightly negative down to a point where it starts to become clear that you do not want to even be flatted by the BB. Hands below this threshold will be much weaker hands, maybe the bottom 30% of your range. Below that point, any increase in the 3-betting or flatting frequency of the blinds really hurts the EV of your hand. Hands at the bottom of your opening range really really want to get the blinds to fold.

The big mystery here is whether the positive range benefits would exceed the negative benefits for the threshold -EV hand. I think it is possible, but it seems just as possible that the opposite is true. Maybe the threshold hand is +EV. Maybe we should be deliberately folding slightly +EV hands so that the lower and middle regions of our range perform better.

You have no way of knowing, simply because the answer depends on your opponent's range and you will never know your opponent's range.

If you know how he plays his range however, then yes, you can easily solve it.

I have a few million hands in my db and am well aware of how chaotic variance can be for single hands close to the 0 EV threshhold. I understand the rift between us is you coming from a more theoretical standpoint and me just looking at practical results. I get your point too, I shouldn't be opening hands that I'm losing money with.

The thing is that if we are going to use your knowledge in practical play it becomes close to impossible to use as a strategy because your strategy roughly requires two parts of a range; a stem, which are +EV hands and leaves, which are hands that are around the 0 EV threshhold. The leaves are a part of a mixed strategy so that you raise them some of the time and some of the time you fold them. For a computer player this will be of no issue, it isn't hard for a computer to randomize T9s so that it opens it 22% of the time, but for a human player it's a lot more difficult to use that opening strategy, so human players use more stem and less leaves.

The problem with using a lot of stem is that your opposition can adjust. And there you have the situation I'm talking about where you will have the bottom of your range losing money. If you're going to have a strategy that mostly is static, such as a opening hand chart with very little mixed strategy at the bottom of the range, then there simply is no way out against adjusting players than going tighter, they go tighter, you go tighter etc if you don't want to lose money with the bottom of your range.

And going tighter and tighter and tighter is not an option so you'll just have to accept losing money with the bottom of your range.

For exploitative play, mixed does not mean randomized though. I'm mixing up my play, with say 89o on the button, by raising it against normal players but folding it aggros. I'm not rolling dice in my mind, my strategy just becomes mixed when I try to make the best play against the opponents I'm up against.

A GTO strategy on the other hand, will be mixed in a randomized sense. Although it still seems pretty unclear if that it means its more or less mixed than an exploitative one. From what I've seen of preflop charts of toy games that were solved, the GTO strat is almost all stem and maybe one or two leaves. Its tough to tell from the Cepheus bot since a lot of hands are "mixed" in the sense that you might do something like limp pre or 4bet with a hand a fraction of 1% of the time.

Why is that not an option?

I've actually seen some elite level players talk about this and do it in videos. The main example is HU preflop against a nit. Even though a hand like 27o might be slightly +EV simply by stealing pre, cbetting flops, bluffing turns and rivers and making the occasional 2pair+, its important to show a fold preflop here and there to keep your opponent from realizing that they need to adjust. You don't want to let your opponent start thinking outside their box.

In my opinion you haven't even posted the beginning of any sort of logical argument. All I've heard you say is you need to lose money so other people can see you do it. That's like saying X is true because X is true.

You haven't provided a shred of evidence that that would be a profitable thing, much less any evidence that it's true.