Quote:
Originally Posted by MozartKills
How is this not a situation where the chips won are of greater value than that of chip lost (contradicting what icm dictates, namely that in a tournament chips won are of lesser value than chips lost)?
I think your main problem might be with this actual statement as for icm it is not correct!
Some may say "chips won are of lesser value than chips lost" but this is only true in some instances. Icm does not create or destroy any wealth so if by gaining or losing chips someone gets an extra $ boost then someone or sometimes many are losing exactly this same $ boost amount. So if there is a player where "chips won are of lesser value than chips lost" then at the table others have the opposite chips won are of greater value. The total $value available is always the total prize money available, this doesn't change if you use an icm value for chips.
If you are in first place gaining more chips may not benefit you to the equivalent of the chipev amount, but if you are in last place icm may ascribe to you an amout above chip ev value for them.
Icm is just a model of chip value at some point in a tournament but is is quite complicated and takes some effort to learn.
Quote:
All models are wrong but some are useful. George Box
In your post you are implying a chip value model where everybody gets the exact payout for their current position and this is quite often wrong and generally not too much use as a model
... although ...
this model could be pretty useful if this was the very last hand of a timed tournament, Hero is in 10th and by winning the hand comes first, losing comes 120th, folding comes 120th. After this hand everyone gets paid out according to current position. Using this model we could calculate how tight we should call.
Winning = $1500
Folding = $100
Losing = $20
So from this we could work out what hand equity you need to call (using your model) to see what is at least break even compared to the no risk fold option. (Rather than equity I'll just use win probability and ignore any blinds to make it read easier.)
To break even:
(w*1500) + ((1.0 - w) * 20) = the return if you take the bet.
Here w = win prob. and so (1.0-w) equals lose prob.
To break even you need to get back at least $100 you would have if you folded so:
(w*1500) + ((1.0 - w) * 20) = 100
1500w + 20 - 20w = 100
1480w = 80
w = 80/1480 = 5.4% (phew!, if I calculated this correctly you don't need much hand equity.)
So if you call with any hand that has 5.4% chance of winning you make money on this final hand. Your model is useful for this one strange situation.
If this was just a chipev situation you would only call with any hand that had a 50% win probability (nb, 50% if no blinds in pot). Folding hands where you have 20% 30% 40% is losing you lots of value here.
Icm works in an equivalent way, you get a decent estimation of how much stacks are worth and then plug these into an EV calc to come out with a break even hand range.
Sometimes people say "icm is too tight, you should always just go for the win" but I think they are wrong. Icm estimates stack size value well and sometimes it will actually tell you to push or call like a lunatic compared to the chipev, it only seems tight in some contexts and then nearly always correctly so. The context is how the stacks are arranged. If you are the chip leader at a very big bubble with a safe micro stack present you can push like a hungry monkey after bananas, but if you a middle stack with the chip leaders to follow you have to be very careful and might have to fold decent looking hands. This can make sense after you study icm and situations.
Icm is a little complicated but if you plug good quality ranges into icm tools they should generate good ranges. If you put crap ranges into the tools they come out with rubbish.
(Don't get confused by always using 'icm' tools set to icm. Early on icm is very small and chipev is close enough. To use icm you have to set up lots of information about other tables and all payouts - this would be better but very hard work when chipev gets about the same result. On final tables always use icm. You have to make adjustments for close to FT situations.)