Is following ICM perfectly always +EV

    • GunFlavoured
      GunFlavoured
      Bronze
      Joined: 19.10.2008 Posts: 626
      Hi,

      I've recently been playing super turbo sngs on Full Tilt and I want to know if I should be following the ICM model perfectly. From what I understand it assumes that every player has perfect calling and pushing ranges but what if players are way looser or tighter than the model suggests? Does this mean that the ICM hards are now -EV or are they still +EV but maybe not as much. For example, first hand of a 9 person sng, 300 chips each (15/30), ICM suggests to push JTs. From experience that surely can't be profitable if there are some loose callers on the table. Should I just push slightly tighter in some possitions or should I trust the ICM model?
  • 13 replies
    • Anssi
      Anssi
      Black
      Joined: 03.07.2008 Posts: 2,173
      ICM expects your opponents to play perfectly. If they do not, your play have to differiate from 'perfect' ICM play as well.

      Let's say opponents correct calling range according to ICM is 5% and you should push 100% because of this. Now, if he is actually calling with 30%, you have to push much tighter range.

      General rule: If opponent is tighter than he should be, push looser and if he is looser, push tighter.
    • PokerRoad
      PokerRoad
      Bronze
      Joined: 21.07.2009 Posts: 143
      If you are able to accuratley assume a Villians range , by using stats, reads ect. then of course thats going to be MORE +EV then using a chart which assumes all Villians ranges in a particular style -i.e TAG

      The problem is when you think ur assumptions are accurate, when in fact they are less accurate then a standerd chart,

      Being able to read ur (differant playing style opponents ) well is the single most crucial skill you can have when playing poker imo

      Better assumptions of your villians range = higher winrate
    • Meiffert
      Meiffert
      Bronze
      Joined: 13.10.2008 Posts: 151
      Originally posted by GunFlavoured
      From what I understand it assumes that every player has perfect calling and pushing ranges
      Originally posted by Anssi
      ICM expects your opponents to play perfectly.
      Not really. ICM doesn't assume or expect anything. You are mixing up several different things.

      ICM is a model which does the following: it takes the chipstacks and the payout structure and tells you how much every chipstack is worth in $ (if you have any math background, then the simpliest way to put it is that ICM is a function).
      In the following I will assume that ICM is actualy a correct way to calculate the value of your chipstack*.
      Now we have a Nash equilibrium which is a concept from game theory and it is a strategy (strategy for all players, not just 1 of them). Nash equilibrium is such a strategy, that no player can improve his expected value if he changes his individual strategy from the Nash equilibrium and others don't change theirs (he can hurt other players, he can give them better EV, he can hurt himself, but he will never get a better 'deal' for himself unless other players deviate from the Nash equilibrium as well).

      If you take these 2 concepts, you can calculate the 'Nash ranges', which is a Nash equilibrium strategy for all players. Tools like ICM Trainer use ICM to find the Nash ranges (the ICM is used to calculate $equity of your stack if you fold, if you call and win, if you call and loose).

      I hope I made it clear, it's a bit complicated stuff for people who don't study/work with game theory.
      Now I feel I can answer you other questions and possibly the answer is clear from what I written already. :)

      Originally posted by GunFlavoured
      I want to know if I should be following the ICM model
      Yes, you should be following the ICM under the assumtion that ICM calculates the $equity correctly (there are exception, see *).
      But no, you should not be following the Nash ranges charts, if your opponents don't do it. You can get better results by adjusting to them. Get ICM wizzard or similar tool, or just use simple rule: if they call too tight, you can push looser (because they won't call anyway), if they call loosely, you push tight (because they will call and you need to win the showdown). If they push tight, don't call (obvious).

      Originally posted by GunFlavoured
      Does this mean that the ICM hards are now -EV or are they still +EV but maybe not as much.
      The Nash ranges are 0 EV if everyone uses them.
      If your opponents don't play accordingly and you do, you will be +EV in the long run, but some hands might be -EV (you and your opponent both loose equity and the players not in the hand gain). If you have a strong read on some villains, you should follow it and adapt your play.



      * - In reality ICM is pretty good and it gives results which are close to the real value in most cases, but it's not perfect. Under some extreme conditions it can actually be way off.
      Take this example: We have a double or nothing SNG, where 5 players are paid $100. There are 6 players left and 1 players has all the chips in the game except 5. The other 5 players have 1 chip each and you are one of these players.
      Now ICM will tell you that your 1 chip is worth $80 (because the bigstack 'always' wins and 4 out of the other 5 people win and it's random who will be the unlucky guy).
      But if you are actually in the big blind and the bigstack is in the small blind the next hand, you are forced to go all-in vs. him and there is 50 % chance of you busting.
      Therefore the real value of your chip is a little less than $50.
      The are other situations where ICM fails, there are better ways to calculate your $equity, but they are pretty complicated, so they aren't used in practise.
    • viewer88
      viewer88
      Bronze
      Joined: 19.04.2008 Posts: 5,545
      Nice post meiffert!

      I would like to add that sometimes you have to make -$EV calls or shoves to avoid even worse -$EV situations in the future OR to make a lot of +$EV situations possible (making one -$EV shove that gives you a big enough chipstack to own the bubble).
      For the same reason you can pass up on small edges in early game to exploit you bigger edge in late game.
    • davodka
      davodka
      Bronze
      Joined: 20.10.2008 Posts: 61
      So should Nash ranges vary between different SnG programs, e.g. between ICM trainer and SnG Wizard? Provided that the opponents p/f with the same range should hero's range be the same for both programs if they use the same ICM model and Hero's edge?
    • Atoks
      Atoks
      Bronze
      Joined: 01.06.2008 Posts: 1,354
      The key to being good at SnGs is the same as with any other type of poker ... if u will better adapt to other players and other player aren't able to adapt to u as well then u have an edge. ICM model and Nash equilibrium are guidelines. They are a mathematical approach to a problem that is far too complicated to be solved in the time one normally has to make a move in the SnG. This is why people often digress from these patterns and create their own.

      The problem with personal p/f stage patterns is the fact that they change to the degree that u can not accurately predict them (unless u have very extensive and reliable reads). Because of this I feel that the best approach to wining SnG play at 10$+ levels is to evaluate each situation separately depending on who u are facing and what the circumstance is. Don't always push in marginally +EV spots and don't always fold in marginally -EV spots. The true art though is realizing when it is best to stick to the ICM or the Nash or when it's best to just ignore either and follow a read u have.

      So bottom line: Don't live and die by the ICM or Nash ... if not for the reasons listed above consider this: u are very easy to read if u blindly follow either pattern.
    • Meiffert
      Meiffert
      Bronze
      Joined: 13.10.2008 Posts: 151
      Originally posted by davodka
      So should Nash ranges vary between different SnG programs, e.g. between ICM trainer and SnG Wizard?
      No, I believe both these programs use ICM and therefore with the same setting they should recommend you the same range.

      Originally posted by Atoks
      ICM model and Nash equilibrium are guidelines...
      stick to the ICM or the Nash or when it's best to just ignore either and follow a read u have...
      Don't live and die by the ICM
      Please, reread my previous post. You confuse ICM with something else. ICM doesn't tell you what to do, therefore it makes no sense to 'ignore' it. Well, you can ignore it, but it's the same as ignoring that flush beats a straight.
      ICM is just a function, if you know it and can use it, it's just a knowledge that can help you win. ICM is NOT a strategy, it never tells you what you should do.

      Originally posted by Atoks
      Don't always push in marginally +EV spots and don't always fold in marginally -EV spots.
      Why would you want to do that?
      Again, I have a feeling that you said something else than what you meant.
      Maybe Nash ranges say that you should do one thing, but another play has actually higher EV, because the opponent doesn't play optimally.
      Therefore: Do push +EV spots and do fold in -EV spots even if Nash says they aren't +EV or -EV respectively.

      Originally posted by Atoks
      So bottom line: Don't live and die by the ICM or Nash ... if not for the reasons listed above consider this: u are very easy to read if u blindly follow either pattern.
      You are easy to read, but you are unexploitable. Therefore there is nothing wrong with following Nash strategy. If you don't have a read on your opponents it's the best thing you can do.
      If you don't follow Nash, you are making yourself exploitable and if an opponent play Nash, you hurt yourself. Even worse if he can read you and actually adapt to you, you hurt yourself even more. Your only hope is that he will missread you and consider you a looser player when you are actually tighter or the other way around, but I can't see how you can make your opponent think that you are tight (and he should therefore fold to your push) when you are loose and pushing every hand. :)

      Of course if you have a read that your opponent does NOT follow Nash equilibrium himself, than you should adapt and exploit him.
    • Atoks
      Atoks
      Bronze
      Joined: 01.06.2008 Posts: 1,354
      No I meant what I wrote. I'm just not sure how to explain it better. From my experience (which doesn't include games above 22$ SnGs) people don't always do what u expect them to do if u perceive them to follow ICM or Nash p/f ranges. Very often I used to get called a lot lighter then what the correct calling range according to ICM or Nash would be, or noticed that people push a lot tighter. Depending on how often and against which this players this happened I adapted to that play.

      For me reads on other players are more important but reliable reads take time to collect. However when I do have reliable reads I will think how I should adjust my push/call ranges depending on what player I face in what situation. As soon as one person widens their ranges for pushing or calling that's how soon u have to rethink ur own ranges for pushing or calling against that person in the future. I hope I've made my views more clear now.
    • Meiffert
      Meiffert
      Bronze
      Joined: 13.10.2008 Posts: 151
      Originally posted by Atoks
      I'm just not sure how to explain it better. From my experience (which doesn't include games above 22$ SnGs) people don't always do what u expect them to do if u perceive them to follow ICM or Nash p/f ranges. Very often I used to get called a lot lighter then what the correct calling range according to ICM or Nash would be, or noticed that people push a lot tighter. Depending on how often and against which this players this happened I adapted to that play.

      For me reads on other players are more important but reliable reads take time to collect. However when I do have reliable reads I will think how I should adjust my push/call ranges depending on what player I face in what situation. As soon as one person widens their ranges for pushing or calling that's how soon u have to rethink ur own ranges for pushing or calling against that person in the future. I hope I've made my views more clear now.
      Well, you made yourself clear and I agree with you:
      Originally posted by Meiffert
      Of course if you have a read that your opponent does NOT follow Nash equilibrium himself, than you should adapt and exploit him.

      Originally posted by Atoks
      No I meant what I wrote.
      But you wrote something completely different in the first post.

      The problem is that you don't use the term ICM correctly. You seem to think that ICM is a strategy that tells you how to play. That it gives you ranges you should push and ranges you should fold. But ICM doesn't do this at all.
      You pretty much make ICM = Nash ranges, but these 2 things are like... petrol and F1 racing. Sure, both are somehow related to cars and one of them is dependant on the other, but you would never say "I ran out of F1 racing". That just makes no sense.

      Anyway, we agree on the content which is more important than the form and that's a good thing. ;)
    • Atoks
      Atoks
      Bronze
      Joined: 01.06.2008 Posts: 1,354
      OK ur right, I did express myself poorly. I'll try avoid that next time :)
    • davodka
      davodka
      Bronze
      Joined: 20.10.2008 Posts: 61
      I know these are silly questions but the only dumb questions are ones you don't ask.....

      Tools that use adjustable opponent ranges (e.g. SnG wizard) give us the complete hand range that is +EV for each case, but doesn't necessarily correspond to the Nash range?

      In the HU p/f stage is it generally best to stick to Nash unless you have a definate read, since at worst we will be 0 EV long term? (I guess this applies in any case?)

      This may be off topic but also in HU when the blinds are high compared to our stack is it more profitable to follow Nash, or use the Sit and Go Endgame System? I find remembering SAGE really easy... :)
    • Meiffert
      Meiffert
      Bronze
      Joined: 13.10.2008 Posts: 151
      Originally posted by davodka
      Tools that use adjustable opponent ranges (e.g. SnG wizard) give us the complete hand range that is +EV for each case, but doesn't necessarily correspond to the Nash range?
      Yes.
      With the slight problem that if my information is correct, you can't choose the exact range for opponent in SnG Wizzard, you can only choose the range in %. Therefore if you think that your opponent is pushing any pair, you need to give him a range of over 50 % because SnG Wizz "thinks" that 22 is not in top 50 % hands. But then you have to include in his range hands like Q2s, which is exactly on the 50 % mark. Hence it's impossible to set a range which includes all pairs, AX hands and broadways for example.
      But generaly if you have a tool that allows it, you will get the correct answer under the assumption that ICM is working (see my first post in this thread for possible exeptions).
      You can simulate this tool using Equilator and some ICM calculator (there are some easy to use online).

      Originally posted by davodka
      In the HU p/f stage is it generally best to stick to Nash unless you have a definate read, since at worst we will be 0 EV long term? (I guess this applies in any case?)
      Yes, you will be at 0 at worst in HU using Nash.
      With more people in the tournament this doesn't have to be true.
      As an example: you are on the bubble with a bigstack on the small blind. Another bigstack with the same chipcount is in the bigblind and there are 2 very shortstacked players (1 chip each) who already folded.
      Since you are both nearly guaranteed 2nd place, your $EV is half of 1st and half of 2nd place, which is (50 + 30)/2 = 40 % of the total prize money.
      If you would play agains each other, the winner would get 50 % and the looser 0 %.
      Therefore you are risking 40 to win 10 and you need 80 % equity. This means that if you push, your opponent should fold anything except QQ+. This means that Nash will say you to push any 2 cards, because you will pickup the blinds in 98.7 % cases. If your opponent is bad though and would call with a wide range, you should surelly fold unless you had a monster yourself.

      Originally posted by davodka
      This may be off topic but also in HU when the blinds are high compared to our stack is it more profitable to follow Nash, or use the Sit and Go Endgame System? I find remembering SAGE really easy... :)
      Nash is surely better, but not by much.
      SAGE is actually just an an attempt to calculate Nash in realtime in your head. The price for being easier (and usable for non-Rainmans) is the fact that it's not perfect and in close cases it sometimes gives you the wrong answer. But since this happens only in close cases, it's certainly profitable.
    • Flows
      Flows
      Bronze
      Joined: 09.03.2006 Posts: 543
      Originally posted by davodka
      [...]
      In the HU p/f stage is it generally best to stick to Nash unless you have a definate read, since at worst we will be 0 EV long term? (I guess this applies in any case?)
      [...]
      In my opinion it's not best to stick to Nash in HU! Well it makes you unexploitable but you loose a lot of value because many players are just too tight. It's good for beginners for the first HUs but for me HU is so much more than just calculating a few numbers! ;)
      example: first hand in HU, stacks are almost equal and around 10BB. SB pushes and you have for example K6o. I don't know what nash advises you to do but for me it's a fold because I have an edge on the random player on my limit.

      just my 2 cents :)