reverse ICM is not really the right term but i didnt know how else to describe this in short for the title, so here is what i have been thinking of for quite some time now:

Of course, the closer we get to the top price money, the more ICM plays a role until it is down to heads up. Meaning that for example at the final table with 6 players left we could be facing an all in against our open that would be marginally +cev but -$ev. And the further we go away from the final table, the less does ICM affect things.

Following that logic, this would mean that in the very early stages ICM has the least effect. I would find this to be true except for when it comes to bloated pots where we risk large portions of our stack. Because we are generally so deepstacked at this point that we have tons of stack utility plus the field at this point still is the weakest with all the bad players still in.

Now what i have been thinking is that there could be stages and situations in tournaments, where ICM has even less of an effect. Even to the degree that in rare situations we could be facing a -cev decision that will in fact be +$ev, because the risk reward is so great that it increases our stack utility and therefore future ev tremendously while we are still early enough in the tournament that we dont risk huge amounts of $equity. 

Im talking about a situation where we have a stack with not much stack utility, lets say 15-20bbs and we are at a level were antes just kicked in or one or two levels after (its about where i think this stage where this could be possible would be). Lets say we are facing three all ins in front of us and have a hand where (lets say we knew the math at the time) it would be slightly -cev to make the call. Now id be argueing that due to the great amount of possible gain in stack utility and future ev, this could actually turn out to be a +$ev decision to throw our chips in. Mainly i think my theory accounts for situations like this with multiple all ins, as with just one all in, stack utility gain will not be great enough. So situations where this principle would occur would be very rare I think.

Just something ive been thinking about and im neither convinced of it nor do i think its not true. Would just like to hear some other thoughts on it. Maybe others have been thinking about this too