ICM Trainer

    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      hello,

      i just started using ICM Trainer, and i want to understand it a bit better, couse some things it does, really confuses me :) .

      now, when u take a quiz, it gives u a specific situation where it asks u wether to push or fold, and after u gave the answer and come back to that question u can also see down below, what was the pushing range for you in that specific situation, and this is the part that confuses me. becouse in similar situations it gives u complete different ranges and THAT is what i do not understand.

      for instance, i just had these 2 similar questions: i was in the SB (somewhere to 13 big blinds - 2000 chips/ 160 BB) and i'm holding T9o, and the opponent from BB shows a 17% calling range. the second situation is EXACTLY the same, only now i hold T8o.

      and the answers ICM Trainer gave me where that in the first situation my pushing range should be T7o+ and in the second situation 9x+.

      now for me, between T7o+ and 9x+ there's a BIG differnce. where did ICM Trainer get that? does someone know how this software works?.... 2 similar situations, same spot, same blind level, same stack, same opponents' calling range and 2 different answers... why?


      pls help :)
  • 22 replies
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      I think there were some significant differences between those situations which you overlooked.

      You only mentioned your stack of 13 bb and the blind level. The stack of the big blind is very important. The number of opponents you have is important. The stacks of the players who folded are important. This other information affects the reward:risk ratio on your push (and on the big blind's call), both in chips and in equity.

      For example, suppose there are 4 equal stacks of 3375 at 200/400, and the big blind will call 30% of the time.

      Give up: 3175 chips, 0.2403
      Successful steal: 3775 chips, 0.2685
      Failed bluff, lose: 0 chips, 0
      Failed bluff, win: 6750 chips, 0.3833

      Equity needed at 70% folds: 0.455

      So, you push the hands with at least 45.5% equity against the 30% range. This ignores card removal effects, such as that a hand like Ax might need less equity because it blocks your opponent from having a calling hand.

      Next, suppose the blinds have the same stacks, but the CO has 1000 chips and the BTN has 5750 chips, but for some reason the BB will call with the same 30% range.

      Give up: 3175 chips, 0.266
      Successful steal: 3775 chips, 0.2904
      Failed bluff, lose: 0 chips, 0
      Failed bluff, win: 6750 chips, 0.3911

      Equity needed at 70% folds: 0.5347

      Since you are so much more risk-averse, you can only profitably push the hands with at least 53.47% equity against that 30% calling range.

      These values were produced by my program ICM Explorer.

      ICM Trainer is estimating the Nash equilibrium, the ranges so that neither you nor the big blind have an incentive to change. Given the big blind's range, you push every hand which is profitable to push and fold the others. Given your range, the big blind makes every profitable call and folds the others.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      thanks for the reply, the quiz was set to 6 players with similar stack sizes, forgot to mention
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ...but still am not sure about this. you are talking about bubble situation and risk aversion, wich if i understood correctly, is affecting our pushing range only when there is at least one short stack at the table, which was not the case. as i sayd (after u replyed :) ) there where 6 players with similar stack sizes, somwhere to 12, 13 bb (in both examples i gave). what other factors could affect heroes' pushing range?
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      Your risk-aversion is not just affected by whether there is one very short stack. When chips are more concentrated in one or two opponents, it means you are closer to the money if you don't do anything, so you are more risk-averse.

      One very basic thing you didn't mention was the stack of the big blind. This affects the amount of chips you risk and your risk aversion.

      Please post the actual stacks for all players. If you no longer have the hands saved, then train more until similar situations occur. If you train for a few hundred hands and nothing similar occurs, then it could be that you are now noticing the important differences between the situations, or that the odd situations don't occur often (at least while training).
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ok, so i guess the easyer way to discuss this is to just post the examples. i don't know if the pictures are clear enough, or if they even uploaded so i also wrote the details.

      so, i set the quiz to 6 players with simillar stack sizes, hero is in SB, and came across this situations

      first example:

      1st situation

      bllind level 160/80

      MP2 (2380) fold
      MP3 (2540) fold
      CO (2330) fold
      BU (2150) fold
      Hero (2030) 94o
      BB (2070) 17.9% calling range

      heroes' pushing range 85.5% (9x+82s+, 84o+...)

      2nd situation

      blind level 240/120

      MP2 (1840) fold
      MP3 (2340) fold
      CO (2320) fold
      BU (2290) fold
      Hero (2410) 94o
      BB (2300) 21.1% calling range

      heroes' pushing range 72.5% (94s+, 96o+, 84s+, 86o+...)



      so in the 1st sit i had 12.6 bb and the BB had 12.9 bb with a calling range of 17.9%, and the program suggested a heroes' pushing range of 85.5% (9x+, 82s+, 84o+...). right now i'm not that familliar with the whole mathematics behind the ICM nor risk-aversion, and i only estimate my pushing range in terms of no. of the players left, stack sizes and position. and according to the 1st sit when me and the BB have somewhere to 12 bb left i can push 9x. so i can only asume, in a simillar situation (like 2nd sit) when me and BB have even less then 12 bb (like 10 bb) that i can push even wider, i don't know, like 8x... and here ICM Trainner confuses me... in the 2nd sit it gives me a pushing range of only 72.5% (95s+, 98o+...) wich is kinda counterintuitive for me, as i expected a wider pushing range and i get an even tighter pushing range and THAT's what i don't understand....

      and here's another example:

      1st situation

      blind level 160/80

      MP2 (2410) fold
      MP3 (2200) fold
      CO (2160) fold
      BU (2310) fold
      Hero (2200) 64o
      BB (2220) 16.4%

      heroes' pushing range 89.4% (62s+, 64o+...)

      2nd situation

      blind level 240/120

      MP2 (2340) fold
      MP3 (2210) fold
      CO (2070) fold
      BU (2200) fold
      Hero (2420) 65o
      BB (2260) 22.9%

      heroes' pushing range 74.7% (76o, 64s, 54s)

    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ...and no, the pictures didn't upload... don't know why...
    • ghaleon
      ghaleon
      Black
      Joined: 17.10.2007 Posts: 5,877
      Did you notice that there is differences of BB's calling ranges in both examples? Obviously, if villain is expected to call looser we have to tighten our shoving range.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      i was thinking that due to blind increase all players pushing/calling range increases... including heores' , otherwise one could asume the bigger the blind gets the tighter u push... wich is weired...
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      There are two important differences between these situations.

      First, the blinds are not the same sizes. The blinds affect your reward for stealing the pot, how much you lose when you fold, and what odds your opponent gets. The net effect is not clear. Sometimes you can push a wider range when the blinds are smaller, and sometimes you have to push a tighter range when the blinds are smaller. If it is the bubble with 4 equal stacks of 2000, the Nash pushing range is 93.7% at 200/400, 100% at 100/200, and 87.3% at 50/100.

      Second, your stack changes size. Your risk aversion (and your opponent's risk aversion) gets larger in pots of 4000 chips or more, and the difference between risking a 4000 chip pot and a 5000 chip pot matters. Usually larger equal stacks mean you can push a wider range from the big blind.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ok, so i think that was one of the thing i didn't get, with different push ranges that depend on blind level and are not necesarly increasing as the blind does... those are details that will understand the more i practice probably, couse right now they aren't to clear for me :) , thanks for the answers pzhon
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ...and than, i decided to try to see what SitNGo Wizard haz to say, i don't know, trying to play with the BB's calling rang and stuff like that and was surprised to see that the wizard gave me a totaly different heroes' pushing range than the ICM Trainner :s_o:

      for the 1st sit:

      1st situation

      bllind level 160/80

      MP2 (2380) fold
      MP3 (2540) fold
      CO (2330) fold
      BU (2150) fold
      Hero (2030) 94o
      BB (2070) 17.9% calling range

      heroes' pushing range 85.5% (9x+82s+, 84o+...) ICM Trainner
      heroes' pushing range 55.4% SitNGo Wizard

      why that big of a difference between those 2 softwares?
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      You have to adjust SNG Wizard's ranges. If you enter similar ranges for your opponents to what you see from the Nash calculator, then you should get similar pushing ranges.

      Another adjustment to make is to change the minimum edge to 0. If you use one of the default settings, the minimum edge is 0.14%, and that might tell you to push only a 55% range, but setting the edge to 0 changes that to 88%. The minimum edge system in SNG Wizard is vestigial and misguided, and you need a clear reason to use something other than 0.

      Another issue is that SNG Wizard does not evaluate ties correctly. At times ties are important, particularly if you push with Ax and expect to get called by many Ax hands, and you are quite risk-averse. SNG Wizard will undervalue pushing or calling all-in when ties are common and important. I believe the Nash calculator handles ties correctly.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      ...amd that would mean your program ICM Explorer would estimate more accurate the pushing ranges, especialy in tie situations?
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      My program doesn't (yet) tell you any ranges. It says how much equity you need, and how much ties are worth. It has to be combined with an equity calculator into which you put your opponent's range, such as PokerStove or Equilator.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      i took a break for a while, but as soon as i got back into those programs, i kept running into weired statistics again.

      i tryed to upload picture, but if doesn't work again, i just get the details below.



      so i'm in this situation, 6 handed, similar stacks, hero first in (from SB) and holds A6s

      MP2 (2100) folds
      MP3 (2320) folds
      CO (1990) folds
      BU (2430) folds
      Hero (2450) A6s
      BB (2210) 14.3%

      so it's Hero in SB with 20.4 bb against the BB with 18.4 bb. in this spot ICM Trainer suggest that a nash equilibrium calling range of the BB would be 14.3% while the Heroes' pushing range should be 64.4%

      in the very same spot and under very same circumstances (same opponents, same stacks, same blind level, same heroes hand, same BB's calling range) and the Edge set to 0%, SitNGo Wizard suggests a Heroes' pushing range of only 40%....

      it's a 24% gap bewtween ICM Trainer suggested range and SitNGo Wizards suggested range.... don't get where this comes from....


      p.s. can't seam to get those pictures uploaded, imageshack doesn't work anymore or what, any ideas guys?

      ty
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      What prize structure are you using on both? I can't match those figures using the Nash calculator at Holdemresources.net.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      am not sure what you mean by that, both ICM Trainer and SitNGo Wizard where set to Fulltilt NLH FR so i suppose they get the prize structure from that setting ??? IF not, then how can i get the prize structure?

      what do you mean you can't match the figures?
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      Be careful that you aren't using a prize structure of, say, 50-30-20 with one, and 65-35 in the other. If they are both set to full ring even though you have 6 players, then they should both be using 50-30-20.

      I entered that situation into the Nash calculator, and I got very different pushing and calling percentages. Usually the Nash calculator is not far different from ICM Trainer. I don't know if there is a real discrepancy here, or whether there is some important difference in settings or a data entry error. Perhaps you could post a screenshot from each.
    • altkiki
      altkiki
      Bronze
      Joined: 03.09.2009 Posts: 82
      a screenshot would help a lot, i totaly agree, i've been trying to that all along but i seamd to have a problem which now i managed to fix. so now i can finaly upload images. i lost the pictures with the examples i gave, so i'm gonna post some screenshots with the settings i had for the situations i explained.

      i normaly use this settings for ICM Trainer, sometimes i checked/unchecked the 'Show player ranges'. still there are settings with which i am not familiar, i only now see the heroe edge thing wich was set to 3%, should i make it 0%?



      and the SitNGo Wizard settings



      so basicly, i had those settings in the example i gave earlier, next time i encounter a 'weired situation' i'll be able to post the exact screenshot of the situations ...
    • 1
    • 2