Pushing ranges differ between differtent tools?

    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      I guess ive finally found out why i run into so many cooler spots (and 5-7 handed bust outs in general) since i practice with the ICM trainer!!!

      the Trainer and the ICM Calculator (http://www.holdemresources.net/hr/sngs/icmcalculator.html) i use lately seem to differ a lot when it comes to pushing low pockets (but not only them), with the latter being much more conservative. I'll give u an example to make it easier to get what im saying.

      this is a hand i played yday and today i reviewed it using those tools:

      Hand converted with online PokerStrategy.com hand converter:

      Play hand

      $50/$100 No-Limit Hold'em (7 handed)

      Known players:
      CO:
      $2234.00
      BU:
      $955.00
      SB:
      $3881.00
      BB:
      $1100.00
      MP1:
      $1945.00
      MP2 (Hero):
      $1130.00
      MP3:
      $2255.00


      Preflop: Hero is MP2 with 3, 3.
      MP1 folds, Hero raises to $1,130.00(All-In), MP3 raises to $2,255.00(All-In), 4 folds.

      Flop: ($3535) J, 7, 6 (2 players)


      Turn: ($3535) A (2 players)


      River: ($3535) 6 (2 players)


      Final Pot: $3535.

      the problem is that while my push is good according to the Trainer (13,3%: 33+, A9s+, A5s, AJo+, K9s+, KQo, QTs+, JTs+, T9s) it seems to be horrible according to the Calculator (13.4%: 88+, A8s+, A5s-A3s, ATo+, K9s+, KJo+, QTs+, JTs)

      this finding confuses me so much i dont know what to do now, which one of them is 'the correct' one :f_confused:

      in fact i'm really surprised that the ranges differ at all since i thought that both tools use the same model (Nash Equilibrium) to calculate them. I guess i was totally wrong, maybe the fact they differ (and the explanation behind it) is well known, i just didn't happen to know that before this day.

      now i don't know wether i should still practice with the Trainer since it may not only be a waste of time but also it can make my play worse instead of better... (ofc i know that it may be the other way round as well - the Calculator is the 'wrong' one)

      i would appreciate so much if some more experienced players (but not only them ofc) or even ppl who established the Trainer will try to explain the situation and give some advice.

      Hope this post will start a nice discussion which hopefully leads to clarifying my (not only prolly) doubts :f_biggrin:

      e: the other interesting fact - how is that possible that almost exactly same range in terms of the percentage (13,4%) differ so much between the tools? :f_confused:

      e2: due to a typo (i misread the SB stack for 2881 :f_biggrin: ) i needed to do the calculations again. the fact that ranges differ between both tools didn't change (although the ranges itselves changed a bit obviously). i've edited the ranges from the original post to the ranges given by the tools with the correct stack sizes.
  • 26 replies
    • pinnryder
      pinnryder
      Bronze
      Joined: 10.11.2009 Posts: 597
      Wow. I would like to know that too. I hope I was not practising the wrong ranges with the trainer ?(
    • lessthanthreee
      lessthanthreee
      Bronze
      Joined: 30.06.2009 Posts: 16,300
      well im pushing wider than 88 here.

      i think i would push something like 55-66 into unknowns.

      the ICM trainer is based on an equilibrium strategy which usually assigns slightly wider ranges from early positions (because it assumes opponents will call an early position push tighter because they are positionally aware, when, in fact, they arent).

      In general the ICM trainer is a fantastic tool to practice your push/fold especially short handed and in late position. It becomes a bit less accurate from early positions in 7-9 player hands because the equilibrium strategy becomes less accurate because there are more players behind you to revert from the strategy themselves (which therefore causes you to adjust as well).

      In general you can trust the ICM trainer ranges from late position and short handed but it might pay to tighten up a bit more than it suggests from early positions, especially 7-9 handed.

      And remember, if you keep getting called by monsters this means your push/fold game is solid because your opponents are calling you too tight! In fact, if you are only getting called by 88+ AJ+ then you should actually be pushing MORE hands, because your opponents are calling you way too tight!

      This is a feature of the micro-games, which allows us to push ridiculously light against most opponents.

      In conclusion, your push isnt actually that bad, it might be slightly wider than standard. But, players on these limits will call you SO tight that you can profitably shove wider than normal. That is especially true for late position pushes. (for example if ICM trainer says 75% from the SB or BTN you can probably get away with ATC).

      Always always always look at the calling range the ICM trainer assigns your opponents after each hand. If you think those ranges are too loose, you can push wider than the trainer suggests, if those ranges are too tight for your opponents then you can push tighter than the trainer suggests.

      Hope that clears some things up
    • Lizocain
      Lizocain
      Bronze
      Joined: 03.06.2009 Posts: 173
      Originally posted by wiarygodny
      I guess ive finally found out why i run into so many cooler spots (and 5-7 handed bust outs in general) since i practice with the ICM trainer!!!

      the Trainer and the ICM Calculator (http://www.holdemresources.net/hr/sngs/icmcalculator.html) i use lately seem to differ a lot when it comes to pushing low pockets (but not only them), with the latter being much more conservative. I'll give u an example to make it easier to get what im saying.

      this is a hand i played yday and today i reviewed it using those tools:

      Hand converted with online PokerStrategy.com hand converter:

      Play hand

      $50/$100 No-Limit Hold'em (7 handed)

      Known players:
      CO:
      $2234.00
      BU:
      $955.00
      SB:
      $3881.00
      BB:
      $1100.00
      MP1:
      $1945.00
      MP2 (Hero):
      $1130.00
      MP3:
      $2255.00


      Preflop: Hero is MP2 with 3, 3.
      MP1 folds, Hero raises to $1,130.00(All-In), MP3 raises to $2,255.00(All-In), 4 folds.

      Flop: ($3535) J, 7, 6 (2 players)


      Turn: ($3535) A (2 players)


      River: ($3535) 6 (2 players)


      Final Pot: $3535.

      the problem is that while my push is good according to the Trainer (13,3%: 33+, A9s+, A5s, AJo+, K9s+, KQo, QTs+, JTs+, T9s) it seems to be horrible according to the Calculator (13.4%: 88+, A8s+, A5s-A3s, ATo+, K9s+, KJo+, QTs+, JTs)

      this finding confuses me so much i dont know what to do now, which one of them is 'the correct' one :f_confused:

      in fact i'm really surprised that the ranges differ at all since i thought that both tools use the same model (Nash Equilibrium) to calculate them. I guess i was totally wrong, maybe the fact they differ (and the explanation behind it) is well known, i just didn't happen to know that before this day.

      now i don't know wether i should still practice with the Trainer since it may not only be a waste of time but also it can make my play worse instead of better... (ofc i know that it may be the other way round as well - the Calculator is the 'wrong' one)

      i would appreciate so much if some more experienced players (but not only them ofc) or even ppl who established the Trainer will try to explain the situation and give some advice.

      Hope this post will start a nice discussion which hopefully leads to clarifying my (not only prolly) doubts :f_biggrin:

      e: the other interesting fact - how is that possible that almost exactly same range in terms of the percentage (13,4%) differ so much between the tools? :f_confused:

      e2: due to a typo (i misread the SB stack for 2881 :f_biggrin: ) i needed to do the calculations again. the fact that ranges differ between both tools didn't change (although the ranges itselves changed a bit obviously). i've edited the ranges from the original post to the ranges given by the tools with the correct stack sizes.
      that spot sucks for you, only good thing is that you cover the BB so he cant call anything lower then 99 or maybe TT+ and AQs+, so pushing 33 is good, problem is MP3 and SB, if you use stats, I would do the following, check the players after you, if they are really tight, and fold bb alot (lets say vpip 10-13 and fold bb sb 90%) then IMO best thing is to raise 2.5x bb and fold to their shove, I am more then 90% sure that if a tight player shoves you, they will never have a weaker hand then 77+ AJ+, so you're hand is pretty much dead, after that you still got 9bb which is a good stack still to shove and double up ur back to the game. I would do this only against very tight, if you have ppl with 18-25vpip after you, its a must push allin, you cant do anything sadly. btw what limit and type of sng is this?
    • goldchess
      goldchess
      Bronze
      Joined: 17.02.2010 Posts: 641
      Originally posted by Lizocain

      then IMO best thing is to raise 2.5x bb and fold to their shove,
      No, don't do this with 11bb
    • Hahaownedlolz
      Hahaownedlolz
      Bronze
      Joined: 24.04.2009 Posts: 1,755
      arg tripple post.. internet disconnect and it sent it 3 times..
    • Hahaownedlolz
      Hahaownedlolz
      Bronze
      Joined: 24.04.2009 Posts: 1,755
      Fold preflop imo. 5 players behind you and you have 33. You will never be really ahead or crushing them only against A2 or 22 or any 3X hand your good. your only slightly ahead against pretty much any hand and your crushed by any pair higher. You still have 11BB so you can still wait a little while.


      Not sure if you noticed. But 33 was at the bottom end of the hands you should
      have pushed with according to the ICM Trainer. Imo its a great tool. Just don't rely on it blindly. Adjust accordingly to the player and everything else. and pushing as loose as the ICM suggest is certainly not always the best play Then again pushing tighter then it isn't always the best play either..

      Just see it as more of a guideline then 100% how you should play.
    • Hahaownedlolz
      Hahaownedlolz
      Bronze
      Joined: 24.04.2009 Posts: 1,755
      internet was lagging so posted twice.. see post below.
    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      Originally posted by Lizocain
      btw what limit and type of sng is this?
      12$ Turbo FR STT SnG on FTP (so many abbreviations it looks ridicuolous :f_biggrin: )
    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      ive got another problem of similar sort. This time its abt a big difference in percentage of the pushing ranges between SnGWizard and Nash Calculator. And yes, i did adjust the Wiz ranges, all of them (exactly the same as those in Nash Calc).



      the rest of the range that you can't see on the pic goes: ...T9,T6s+,98,96s+,86s+,76s,65s

      nash ranges

      anyone has any idea why there's a 7% difference between those ranges?
      i kinda understand smaller differences and including/not including particular hands in same %ranges by different tools. but this one i don't get tbh. maybe there's simply some small detail that ive overlooked, if thats so sorry for the fuss :)
    • plexiq
      plexiq
      Diamond
      Joined: 28.12.2006 Posts: 1,024
      Did you also adjust all the pushing/calling ranges in Wiz after a FOLD by Hero?

      These matter a lot, and they will typically be too tight in Wiz (which results in a wider Hero push range).
    • petersveter
      petersveter
      Bronze
      Joined: 18.10.2011 Posts: 339
      I think also the overcalls differ in each program, and that also makes a difference.
      I was just actually just going to start a thread on this, but this topic relates. Where do you guys set the overcalls in wizard and where do you set the situation after hero folds?
    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      Originally posted by plexiq
      Did you also adjust all the pushing/calling ranges in Wiz after a FOLD by Hero?

      These matter a lot, and they will typically be too tight in Wiz (which results in a wider Hero push range).
      Yes, ive adjusted absolutelly all the ranges possible.

      Originally posted by petersveter
      I think also the overcalls differ in each program, and that also makes a difference.
      I was just actually just going to start a thread on this, but this topic relates. Where do you guys set the overcalls in wizard and where do you set the situation after hero folds?
      when you click on the call% number for any of the players at the table youre gonna get a popup with a hand chart, beneath it the % to type in manually or change on the slider, above the chart there's a list of all possible situations this plr may find himself in that particular hand with the pushing/calling range% next to them. By clicking on any of those you get to change the range using the chart/slider/typing it in. hope that helps :)
    • kurrkabin
      kurrkabin
      Bronze
      Joined: 12.10.2010 Posts: 5,976
      Hi, everyone!

      @ wiarygodny But that's impossible, man. Nash/wiz work using the same model, if you set everything exactly the way it is in Nash- Overcalling ranges, Hero fold scenarios, Edge set to zero, push/call range etc, Wiz should show exactly the same results.
    • plexiq
      plexiq
      Diamond
      Joined: 28.12.2006 Posts: 1,024
      Originally posted by wiarygodny
      Yes, ive adjusted absolutelly all the ranges possible.
      The "old" Nash Calculator and Wiz both use linear Ranges / Handrankings and the actual Rankings differ. So if you set a 10% call range in wiz, it is not going to be the same 10% in you see in the Nash Results. This will cause some difference, but typically not that big.

      Wiz also does not handle split pots correctly while holdemresources does, this will cause some minor differences as well.

      I really doubt that these make up for 33%->40% though.

      You can try to use the Beta at holdemresources and set the ranges *exactly* as they are in Wiz (ie, manually override the hand selection on holdemresources to match the Wiz handranking).

      (Disclaimer: holdemresources developer here, might be slightly biased)
    • kurrkabin
      kurrkabin
      Bronze
      Joined: 12.10.2010 Posts: 5,976
      Thanks for the e input, man. I forgot that ranges in wiz are not adjustable and 25% in nash is different than 25% in wiz as it includes different hands. I had that experiment with one of the hand judges here few months ago and it worked out to be exactly the same result as the one in nash. But it was 3-way situation on a bubble, so less ranges to consider.
    • petersveter
      petersveter
      Bronze
      Joined: 18.10.2011 Posts: 339
      @wiarygodny yeah thanks, I havent been using this so far :s_thumbsup:
      do you change these ranges often?
    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      Originally posted by kurrkabin
      @ wiarygodny But that's impossible, man.
      thats what i tought as well so i did the calcs once again just in case ive overlooked sth when i did it for the first time. Surprise, surprise... ive got exact same results again :D ive even changed the BBs range for a 4way allin but apparently whether its 0,5% or 100% my pushing range is still 40,3%. that is a massive difference compared to the Nash Calc and it does worry me slightly :( Maybe my Wiz was hacked by my opponents to make me play -EV ICM :f_biggrin: maybe, if youve got a minute and some good will you can analise it with your wiz, though i really dont think that would give a different result.

      Originally posted by plexiq
      I really doubt that these make up for 33%->40% though.

      You can try to use the Beta at holdemresources and set the ranges *exactly* as they are in Wiz (ie, manually override the hand selection on holdemresources to match the Wiz handranking).

      (Disclaimer: holdemresources developer here, might be slightly biased)
      Thanks for the voice 'from the inside'! i knew that the same range% in both tools contains a bit different hands but as you said the differences shouldnt be that big. At least ive never seen that big of a difference before. I may try to run the Beta analysis when i have a spare moment. If I do ill definitely post the results here. Thanks again for the input!


      Originally posted by petersveter
      @wiarygodny yeah thanks, I havent been using this so far :s_thumbsup:
      do you change these ranges often?
      no worries, glad i could help. Whether i change them (all of them or just some) depends on the hand itself. I'll most likely do it if im analysing a bubble spot (since those are particulary tricky and at the same time can kinda cost/win you most money). The first change you should go for (or the only one if youre too lazy to do all of them, like me sometimes) is imo the SB open pushing range and BB range for calling it, since according to my exp Wiz does underestimate them big time, and obviously the chance that blinds collide can make a massive difference to your pushing range (especially if one of them is short on the bubble). I don't rather do too many adjustements if im pushing into 8 plrs since im simply too lazy to do that :f_biggrin:
    • wiarygodny
      wiarygodny
      Bronze
      Joined: 16.07.2009 Posts: 1,395
      Ive finally found some time to play a bit with holdemresources.net Beta calc.
      First just a quick reminder of what is this discussion abt (details in one of my posts above):
      - certain hand analisys in the Nash calc gave me a pushing range of 33,3%
      - after adjusting the ranges in SnGWiz accordingly to those from the calc (exact same percentages, not necessarily exact same hands!!!) gave me a range of 40,3%

      as suggested by plexiq ive input the Wiz ranges (exact hands) back into the calc, this time the Beta version. This time ive got a range of 43% (still a pretty big difference compared to Wiz - 2,7%, but i don't think we should worry abt it too much since its beta and it even says on the websites that there can be some major bugs). So thats not really important.

      What is important is that in the process ive realised that because both tools (Wiz and the original Nash calc) use different handrankings when you try to put the Calc's percentage into Wiz you not only get different hands (which is nothing new to most of us) but you may also get a different... percentage :)

      Example
      8,7% in Calc - 77+ A9s+ ATo+
      8,7% (manually typed in) in Wiz - 66+ ATs+ AJo+ which is in fact an 8% range

      Ive never really thought abt it, i guess ive just assumed that Wiz is gonna somehow round up/down the ranges to make them more accurate but apparently it doesn't which is imo the reason why we have this whole discussion. I guess in a vast majority of hands those little differences don't matter much so the results produced by both tools are very similar. But there are gonna be some hands where there's gonna be so many of those that when they add up they make the results differ by a lot, like in the given hand for example. Hopefully that is not gonna happen very often, cause it is a very confusing (at least for me - which range is the 'correct' one?).

      Hope that helps all of those that find different results as annoying as i do :)

      p.s. another hand with a big difference: nash 26,7%, Wiz gives BU 21,3% (Calc Beta with Wiz's hands 22,5%)
    • plexiq
      plexiq
      Diamond
      Joined: 28.12.2006 Posts: 1,024
      as suggested by plexiq ive input the Wiz ranges (exact hands) back into the calc, this time the Beta version. This time ive got a range of 43% (still a pretty big difference compared to Wiz - 2,7%, but i don't think we should worry abt it too much since its beta and it even says on the websites that there can be some major bugs). So thats not really important.
      I should change the beta message at some point, lol. Calculations have been online for 2 years and tested against other tools. (Icmizer author spent a lot of time to verify that his calculations match the beta calculator results). There are no known bugs in the calculation.

      It seems most of the difference comes from hand selection and wizards range rounding anyway, 43% vs 40.3% or 22.5% vs 21.3% does not seem bad at all and could very easily be caused by the simplified handling of split pots in wizard.
    • 1
    • 2