ICM Trainer seems off

    • Nabukadnezar
      Nabukadnezar
      Bronze
      Joined: 18.09.2008 Posts: 65
      I have a huge ROI without using the all in steal strategy in SNGs <75$ but I wanna start using it a bit more often to multitable easier. I'm trying to get at least 98% in ICM trainer but I can't seem to agree with all its decisions. For example even before hearing about this strategy I would have went all in with A5o in this situation 10 out of 10, especially if the BB was a loose player.



      ICM Trainer says at least A8o, probably because it doesn't want us all in in situations where we are likely to be dominated. I don't agree with that. I also don't agree with KJo+ QTo+. Why not KTo+ QJo+. This makes no sense. The king needs a better kicker than the queen? A2s instead of A5o? I know there's 1 in 15 chance to make a flush and beat any calling hand but still there's a bigger chance that when you'll get called you'll be dominated by a better ace and lose. You're probably hoping for a split pot in this case but the same applies with A5o. JTo instead of A5o? What if we get called by the KTo which is supposedly bad to shove with. Can someone explain this situations, especially the KJo+ QTo+ one or point to an article detailing this strategy with mathematical calculations?
  • 8 replies
    • Nabukadnezar
      Nabukadnezar
      Bronze
      Joined: 18.09.2008 Posts: 65
      I just calculated the exact same situation with SNG Wizard AND A LOOSE BB and it says push.



      So either ICM Trainer is crap or I'm wrong. Please help me understand. In SNG Wizard when I set the BB as very loose it also says fold, which is obviously wrong. Common logic dictates that when the opponent is very loose any ace is a +EV push.
    • cannell555
      cannell555
      Bronze
      Joined: 06.03.2008 Posts: 2,410
      Hi Nabukadnezar,

      Heres the link to the silver ICM article just for reference. http://www.pokerstrategy.com/strategy/sng/175/

      I will quote a little section which I think will help you:-

      EV and $EV:

      EV: (expected value) also called chip EV, is the number of chips you can expect to end up with, on average after a specific action. An action is +EV if it raises the EV.
      $EV: this is the money value of the EV. An action is +$EV whenever it raises the $EV.

      Note: since the EV and $EV are not linearly related (doubling the EV does not necessily double the $EV), there are actions which can be +EV but -$EV. These must be avoided since our aim is to maximize the $EV.

      Note: since we are only interested in $EV in tournaments, this is usually the one we mean when we speak of EV. In this article, however, we will differentiate between EV and $EV.

      That article is worth reading over several times, and will help you understand $EV.

      Hope this helps,
      Stiev
    • Nabukadnezar
      Nabukadnezar
      Bronze
      Joined: 18.09.2008 Posts: 65
      I don't believe folding there maximizes the $EV, unless you're playing for the third place and if so, you could as well sit out and hope for the best. I might be wrong though since the payout difference is not that big in SNG as in tournaments and I might have tournament play in mind.

      Anyway, does the ICM trainer factor $EV while the SNG Wizard is about EV? I believe this is what you were trying to make me understand and it would explain the discrepancy in results. Also, could someone explain the KJo+ QTo+ thingy?
    • cannell555
      cannell555
      Bronze
      Joined: 06.03.2008 Posts: 2,410
      Hi,

      I dont use SnG wizard, but you have give the guys ranges, which are not correct ranges. If you use the correct ranges, you will find both programs will say the same thing.

      If you read the example hands it explains exactly with the numbers to match.

      What do you mean "could someone explain the KJo+ QTo+ thingy?", what is it you would like explaining about it?

      Best regards,
      Stiev
    • SoyCD
      SoyCD
      Bronze
      Joined: 20.02.2008 Posts: 6,356
      Hey Nabukadnezar,

      The two programs use the same range-sets for their calculations (ICM model) and as cannell already stated - will give you identical results if you give opponents the same ranges.

      I think cannell quoted the EV and $EV part in response to "Common logic dictates that when the opponent is very loose any ace is a +EV push"

      Just because your EV (chip EV) is positive - doesn't mean the push is plus EV from a money perspective. Basically the looser your opponent is - the more he calls you - and the higher the chance of you being eliminated on the bubble with $0 winnings becomes. Even while you might be slightly ahead of whatever he calls you with - you don't really want to be called while holding A5o - you would prefer taking the blinds. As such the looser your opponent is - the tighter you need to push against him. Vice versa - the tighter your opponent is, the looser you can push against him.

      The different kicker situations KJ/QT are based on their connectedness and therefore the increased 'suckout value' against an opponents calling range. KTo would e.g. runs worse against a specified opponents calling range than QTo does (QT performs better against AT / AA and in other various dominated situations than KT does)

      Best regards
      SoyCD
    • Nabukadnezar
      Nabukadnezar
      Bronze
      Joined: 18.09.2008 Posts: 65
      Thx for the answers SoyCD
    • cristy007
      cristy007
      Bronze
      Joined: 31.07.2010 Posts: 26
      .
    • pzhon
      pzhon
      Bronze
      Joined: 17.06.2010 Posts: 1,151
      Originally posted by Nabukadnezar
      ICM Trainer says at least A8o, probably because it doesn't want us all in in situations where we are likely to be dominated. I don't agree with that. I also don't agree with KJo+ QTo+. Why not KTo+ QJo+. This makes no sense. The king needs a better kicker than the queen? A2s instead of A5o?
      ICM Trainer concentrates on the equities you have. Whether your hand is dominated or not isn't as important as the equities. Take a calling range of 5% suggested for the small blind. That's about AQ+ 99+. How much equity does each hand have against this range? Check this using PokerStove.

      A5o: 27.68%
      A2s: 30.37%
      KTo: 29.66%
      QTo: 28.36%

      So, I'm not sure why KTo was excluded, but QTo was not, but A2s has much more equity against that range than A5o does. There are some other 5% ranges, such as 88+ AJs+ AKo so that QTo has more equity than KTo. QTo makes more straights, although it is in worse shape against KK than KTo is against QQ.

      It is true that A3 "dominates" A2, but it's not much of a favorite at all. In fact, A3o can actually be an underdog to A2s, and I don't think A3o is in the calling range anyway.

      This hand is rather warped by the push/fold assumption. In practice, if you fold, the blinds will not collide all-in as often as ICM Trainer assumes, so folding is not as good as it assumes. Even if the blinds play a pot, it usually won't be all-in. So, it's ok to push a little wider than ICM Trainer says, even if you agree with the calling ranges.