Hi forum,

I've intalled the ICMIZE tool recently, and quickly realized that my intuitions for when to shove suck quite a bit. To remedy this, I thought it best to look at each variable separately. Thus, I've taken the following example:



In this example, ICMIZE tells me that shoving is positive with +0.15% EV. Okay, cool. Now, I've written down a bunch of ways to change this situation, predicted what that would to to the push-EV, and tested my intuition. The first two are extremely off, but I don't understand why they're off. If someone could explain to me why my reasoning is wrong and how to think about this properly, that would be super helpful.

Change 1: Change stack of BB from 23 to 13
predicted change and reasoning: this should make it much riskier for the BB to call, thus significantly improving push EV. I predict +0.5 change in EV
actual change: -0.04

Change 2: Change stack of BTN from 8.2 to 20
predicted change and reasoning: if the button is much richer, that decreases my chances to get into the money, thus making survival less important. Consequently, shoving should be better now. I predict +0.1 change in EV
actual change: -0.2

but whhhhhhyyyyyyyyyyy

I've also made three others (change Hero's hand to T8o, change hero stack to 5.3BB, move BTN by one place so that Hero is on the BTN), but there my predictions were much better, so they're less interesting.

If every beginner (or professional lol) took this approach poker would be cooked in a few weeks, so I commend you for your process and want to let you know it will serve you extremely well. Keep it up.

It looks like there could be a possibility of overthink or paralysis by analysis that may affect you in game, so try to be a bit more freed up when you play but maintain your analytical nature off the tables as much as possible.

In Change 1, did you see how much villain's call range was actually affected by the stack adjustment? Overall I think what you're missing here is that ALL the ranges will slightly change a bit for each little micro adjustment, so perhaps when you do make these small changes, take the blinders off for your own strategy and pay attention to how everyone's strategy has adjusted and within that you will find clearer answers.

What's your strategic background? I think poker will treat you well, GL!

(Btw, I notice that I forgot to say what kind of tournament we have. It's a 9-max with 5-3-2 payout.)

I hadn't. Here's how the range changes for Change #1:



But this just confuses me more. Why does the range grow when the BB has a smaller stack? Shouldn't it shrink because risk premium is higher?

Okay but... how does this work in practice? How do you decide whether to shove if not by applying heuristics like 'I can apply more pressure -> shoving is better' or 'other people have small stacks -> shoving is worse'? What kind of algorithm is your brain running when you're in an actual game?

I guess this seemed like a bigger deal to me than you're making it sound. It would be one thing if I got the relative importance of things wrong, but in this case, I didn't even get the direction right. I assumed that a smaller stack for the BB would make shoving significantly better, and it turns out (at least according to the tool), it makes it worse. That seemed to suggest I was getting something really fundamental wrong. (Unless ICMIZE isn't trustworthy.)

Mathematics, computer science, other games like chess and dominion.

But also a not very successful history of poker. I've picked it up a bunch of times, always failed to make it work and then gave up. In the last couple of months, I played quite a lot of ~1-5$ sit&go's and only came out about neutral in the end. I've decided to take a break and study more theory so this is me trying to do that

If I take the results from ICMIZE seriously (btw, is it a good idea to take them seriously for lower limits, given that other players don't play close to optimally?), it primarily suggests that that (a) I should all-in a lot more in general, and (b) I should consider the lower card a lot more (T9 is better than Q2, who'd have thought). Those two things were pretty easy to infer, and maybe they already make a sizeable difference, but then there are other results that still confuse me, like the one I posted here. It definitely bugs me if I don't understand the tool's results after I see them.

Shorter stack decreases his "at the moment ICM" and as such decreases his risk because he is closer to 0 at baseline.

For example, give him an infinitely small stack so that he has 1 chip more than the BB he posted.... he's calling literally everything.

Seeing how that plays a role in Change 2 is a little less clear to me (other than, as criss mentioned, a more complex ICM rearrangement). I am not sure, though. Like you, I imagine BB would "feel more pressure" when not the lone big stack. Furthermore would think a 2-way fight for last instead of a 3-way would incentivize more risks taken in your shove decisions. I wonder if it is as simple as BB's relative ICM is enough lowered when not the lone big stack that he also calls wider (as in Change 1). Would you have a similar range change for that scenario?

Either way, I like this approach to analysis. I have done the same for some spots to generate my own heuristics.

Change 1:
What exactly did you change? Did you just subtract the chips from BB and therefore decrease the total chipamount? Or did you redistribute it?

For example, if you simply subtract, the Equity of everyone else increases. Hence, the loss when you get called and lose means a lot more -EV.
While on the other hand winning the pot through his fold, adds a lot less +EV.
The difference in BBs calling range however shouldnt change much, as his situation doesnt change, he is still the chip leader and still enjoys his chipleader EV when he folds.

The result, your shoving EV decreases.

Change 2:
Most notably to me: BBs is now threatend by BTN, his chipleaderEV goes down significantly (less owning), his calling range should therefore increase.

Also, without changes, you have 3 close stacks. In that situation, winning the blinds is very +EV as it increases your chance to get 3rd vs 2 aswell as getting 2nd vs 2.

With changes, you only have 2 close stacks, with 2 stacks far away. You only increases your chances to get 3rd vs 1, thus way less +EV.

I can only speak for myself, but here is what goes through my mind:
Current stack-setup, future game stack-setups (I.e. how are stacks distributed if I fold, if I shove and win the pot, if I shove and win all-in, if I shove and lose all-in (also vs all players)).
Calling ranges and shoving ranges and how me shoving now might (if at all) affect future ranges of my opponents.

This is nothing you can know or calculate like that. All you can do is look at as many situations as possible and develop a feeling how ICM works, what it affects and how. Some stuff is counter-intuitive at first, but eventually youll understand more and more of it.

Btw, for me it helped a great deal to look at the actual ICM-algorithm (which is pretty basic).

Dont worry, ICM is complex, as anything changes everything.
Youll get the hang of it, and even high-stakes pros still have room for improvement.

Its not a good idea to use Equilibrium ranges in low stake games. But ICMizer should let you adjust ranges. Generally speaking, regs tend to call to tight at low stakes, while recs call to wide. But thats just generally. For some it is reverse, in some spots an otherwise tight caller, might call to wide and vice-versa.

When considering what cards to shove, it is a good idea to consider what cards you are going to get called by.

For example:
Shoving A2 has the bonus that it has removal effect as your opponent now has less Aces he can call you with, your folding equity increases. However, if he has a tight calling-range mostly consisting of Aces and Pairs, you equity when called sucks as you are mostly dominated.

Thanks @everyone for so patiently entertaining my approach.

Subtracted.

That makes a lot of sense. So the real story here is that Hero's relative stack increases. This is the kind of insight I was hoping to get, so thanks!

I've tested an alternative of Change 1, where instead of subtracting 10BB from the Big Blind, they instead give both other players (but not hero) 5BB. In that case, the EV of Pushing with T7o goes from +0.15% to +0.24%, which is actually almost exactly what I predicted. Although the calling range also goes up, which is not what I predicted. Here are all three results:



It also makes me wonder why it's not more common to have your current % of all chips as a quantity. According to a google search, PokerTracker doesn't even have this as a statistic. I guess it's not too hard to compute it manually.


I do get that very small stacks have the largest range. I was envisioning range-as-a-function-of-stack-size to start at 100% for 1 Chip, then go down as commitment decreases, but eventually go up again as the stack gets so large that it can comfortably survive a call. Like, suppose the payout structure is 5-2-2-0-0-0, we're at the bubble, and someone shoves 8BB. My range is largest if I have 1 Chip, but intuitively, it's larger if I have 40BB than if I have 8BB (because with 40, I don't risk elimination). Thus (according to my model) it's a parabola-shaped function that goes down, hits a minimum somewhere, and then goes up again. However, I assumed that, surely, the commitment effect is less important when we're already at 13BB, so I put the minimum to the left of 13BB, hence expecting 23BB to have a larger range than 13BB.

Am I wrong about the function being parabola-shaped or just about where the minimum is?


However, you don't actually get that practice by playing poker (or at least I don't) because making a decision in-game doesn't tell you whether it was the correct decision. I think that's like the principled reason why I struggle more with poker than other games. For almost any other game, you can learn just by playing.

So, when you say 'look at many situations', do you do this with tools like ICMIZE, by analyzing games?


Got it. So, look what ICMIZE says for manually set ranges.

As a reg, I can attest to calling too tightly and seeing many others calling too widely.

Exactly.