The question of when to call for all your chips has been under a ton of scrutiny. Everyone pounds the point home of the ICM saying somewhere between 51% and 56% should be a call hand one. But that’s an absurd model based on coinflipping with all players even those that put in their money with a bluff and bet small with a big hand, even those that put in 1/3rd of their stack before folding. Even those that only ever check and call and allow you to define bet size. I have a different mindset.
If you are just slightly better than average you can and probably still should follow ICM. This probably applies to 70% of players, and that’s a problem because 80% think they are better than average. If you can make the ante stages in a format like the below over half the time without ever being all in.
But where the ICM model really fails is where you have a significant skill advantage.
If you can manage (fairly) low risk chip accumulation at a fairly high rate you can hover above 20 big blinds perpetually in some slower structures when play is fast enough without being all in. This amount of skill level may be really aggressive but shows that theoretically a skill level may allow for the folding of really strong hands preflop and giving up really big edges if you can steal your way to a victory.
I will elaborate on whether or not the skill level is possible but the idea is to show that a substantially skilled player shouldn’t call off his tournament life according to ICM models if substantial skill can be consistently demonstrated.
A more likely scenario might be that for say the top 25% of the field they tend to last and face each other and their edge declines. I think the edge can go up substantially in the ante periods as the patient players who tend to survive tend to not adjust strategies to the antes. Another thing that happens is stack sizes decline relative to the blinds and the edge declines and/or risk goes up. But if you can model your chipstack over time from the point of where you are now to say 25 big blinds when skill level declines to where maintaining low risk accumulation begins to become difficult. It really doesn’t become a requirement for all in or fold until 7 big blinds in my opinion but if you get too short stacked it’s hard to gain any additional play/skill level without risk and there is a bennefit to doubling up earlier, so I feel by choosing below 20 big blinds or 15 it’s a better way to identify where you project to get to before it becomes jam or fold....
I could get to this number with more than 100 big blinds as if opponents poorly adjust to antes, play 10% away from equilibrium and you can find a few spots to exploit you could easily play close to equilibrium plus get the rest of the way with exploits of mistakes. That would get you to 10BB per 100 which in 5 rotations is about 2 big blinds, that doesn’t include the ante but WITH antes the edges might be bigger. Nevertheless, it’s a stretch to get to this at 20 big blinds without risk.
But, there is hope. In a post “How wrong is ICM”; we concluded that in the ante stages with a 2x pot bet we would win just over 4 big blinds per rotation in steals alone if everyone at your tabled adjusted to literally use ICM. I don’t think they actually will play that way, but you also would have equity when they try to slow play and let you see a cheap showdown. I would go one step further. The point I want to make now is that based on the current understanding the public has, it’s impossible for them to play anything like “perfect” poker in tournaments. While there may be something like equilibrium poker in cash games, all tournament poker is exploitable. If they open up really wide assuming their opponents adjust properly you can 3bet them really light and if they respect ICM they have to fold too much. And if you have enough chips that’s low risk chip accumulation. If they don’t, and opt to call too much you can value 3bet light and control the pot size later. If stack sizes get bigger you can flat call in position with a stronger hand than opponent or if you recognize flaws in his game. If they open up only slightly more than they call, you can autoprofit any two by stealing. If they play like a cashgame, other people will knock them out and the big pots will be an ICM disaster and if they don’t adjust to the ante there are huge profits to be accumulated over time. Ideally you can wait for better conditions and try to avoid them except in small pots where you have a fairly big edge. If they adjust to pot size you can jam light or play bigger pots and they can’t call you often enough and you steamroll, particularly if you have more chips and you can afford the volatility of bigger pots when you’re ahead and they can’t when they’re ahead or behind and they can’t afford to wait for good enough hands to take the volatility of raising a big pot. And if they bluff the pot will get way too big when you have a strong enough hand. There’s nothing like an ICM solution until it gets into Jam or fold where ICM sort of functions, there’s a lot of really bad ideas on how a tournament should be played. If they jam too often and you have them significantly out chipped you can call with a stronger hand and still expect to gain chips even accounting for volatility. If they have a bit more chips you have to wait for a much stronger hand but you can still gain chips at low varience.
How we play might look more like this. We then can see our tournament life hand one is worth about 12750. While we may be able to gain for a bit longer as we get to 14880, we need to buy ourselves a hand to play for a coinflip and play ICM poker from then on in. And we need to account for varience. And since according to ICM the chips we lose are worth more than the chips we gain, varience hurts our results. But we could take our survival hand one, incorporate our skill and factor in ICM from where we project to end up with that skill (12000 chips or about 3x our starting stack) and then run ICM to determine our value. Roughly speaking, if there are still far from the money bubble 12000 chips is worth about the percentage of the prize pool that we hold in chips.
FOr example, if there are 1000 entrants in a $10+1 tournament, there is $10000 in the prize pool and 3000*1000=3M chips and we have 12000/3M=~$42.00.
To be prudent we won't actually get to that spot 100% of the time but we will be comparing it to when we double up vs when we lose. The other challenge will be do multiple double ups help, do we stand to find better opportunities to double up by waiting?
Nevertheless if we are dealt AK first hand, someone jams and accidentally shows and we are 60% to win is it worth it? Lets reconstruct skill chart with 6000 chips.
In this case while hovering above 20BBs we have more low varience chip accumulation methods available to us The higher skill lasts until the competition thins and gets tougher and our stack dwindles in Big Blinds. Then it catches up to us and we will be forced to go into ICM mode. So a double up is worth maybe 35,000 to 45,000 chips where as surving is worth ~12,700.
So is it setled? do we call? At first glance it seems that if we are even 32% chance to get to 40k we should take the gamble that we can parlay it into steals in later levels... but without the ability to run a lot of hypotheticals and pair it with probability of each hypothetical, this is very much an educated guess and somewhat subjective also. Is our skill level really this high the first few levels into the ante? Even if you monitor your hands it will take 300,000 hands before you are within 1.5big blinds of your actual bb/100 win rate*. and in a tournament it will be very different also.
*source https://www.pokerstarsschool.com/strategies/what-is-bb-100-poker-explained/754/
This is tricky though... What if we can fold and we can double up and hover above 20BBs anyways? What if you using a bit higher varience can gain a larger skill edge? If a gamble the first hand even with only a 32% chance to get to What other methods are there?
Well we can look at max patience to determine probability of getting a better hand, but there is no guarentee we get called. We could also build a more sophisticated spreadsheet to run a model that features probability of us winning the blinds, winning on CBet, and other variables for every hand based on some assumptions about how our opponents play and what our range is and playing that way determine our results by approximate finish.
For the time being, we will construct a model based on hands approximately dealt during each time frame. Without the double up, we have 280 hands left while we apply that edge. With the double up we have about 380 hands. They’re both huge. And we still need to know what blind level the final table takes place. I usually guess average stack is about 15BBs. With 1000 players starting at 3000 and thus 3million in chips divided by 9 at final table is 333,333 big blinds average stack divided by 15 is 22,222. So round down to 10,000/20,000 blind levels at the final table. A single double up is not enough if done early. But what if we double up close to when we have 12,000 chips and do so with a superior hand to ace king? Can we perhaps parlay that into a much deeper run?
This is the challenge with a model that incorporates skill but it is much more useful than ICM in earlier stages. How about we say we double up halfway in. With 280 hands or 14 levels total we say 7 levels in. 6750 in chips. That’s great because it would put us at the 100/200 level just before antes get involved and that means we jump up to about 13500 in chips with some momentum and enough chips to play at a high rate of skill. In fact. Even when skill levels off a bit to 1.5 (which it might not since we might have close to 30 big blinds instead of 2)—we still have the ability to hover above 15 big blinds nearly all the way to the final table. And if we take some well timed aggression we may have a skill higher at a few levels. But ultimately a double up from 6750 leads to us grinding up to about 120,000 in chips before we have to get close to going into ICM mode. I don’t know how accurate this is because when you hover around 15 big blinds varience will cause you to dip below. And even though you can still open up pretty wide and fold it isn’t long before a few minraises forces you into jam or fold with less chips. But certainly it seems parlaying the double up at the right time is worth a lot more. The early double up we concluded was worth 40,000 in chips. If we win 60% of the time, that’s worth 24,000 in chips. Even if you took the late double up with 120,000 only 20.01% to win or more it would be worth more than the early double uo if you could only double up once. But we still didn’t make the final table yet and we have the chance to consider a second double up. Let’s add another double up for both and convert it into final table percentage and top 5 percentage...
After the first double up we have 380 hands rather than 280. While there’s a chance our opponent won’t have us covered early there’s a lot more hands for the second double up. A late first double up 7 rounds (140 hands)in still will require one more double up to final table and we will have about 340 hands to find the next double up. But if we double up from 50,000 chips or so to 100k in the 19th round after the late double up we will for sure final table and probably get to 5 left with 7 big blinds.
And if we only need 2 double ups in 28 levels—lets say 560 hands—even if we only get called on half our big hands, in 560 hands how many times do we get dealt QQ? In a Monte Carlo of 560 hands drawn 10,000 times, we got QQ+ an average of 7.57 times per 560 hands. We got it 4 or more times 86.8% of the simulations. We got it 8 or more times 48.4% of the tournaments. We got KK+ an average of 5.02 times per 560 hands. We got it 4 or more times 74.5% of simulated tournaments. We got KK 8 or more times 13.3% of tournaments. So if we get called half the time we jam, we can jam with KK and get two calls roughly 74.5% of the tournaments where we play 560 hands. If we get called 25% of the time we jam we can jam and get called roughly half the tournaments we play at least 560 hands.
See also— http://nutballpoker.blogspot.com/2018/07/tournament-strategy-design.html
In any case, being all in with QQ+ vs calling range or KK+ vs calling range even if overbet jamming loses some skill by not protecting our other raising hands as much should yield a very high final table percentage depending on the timing of when we get these hands and the calling range. But if we are say 75% to win on average and need to survive twice, we are looking at a 56.25% chance of surviving. So players with substantial skill would be insane to risk a 40% chance of elimination on the first hand and the ICM model is absolutely horrible for the very very elite players or even those who participate in the very very slow structures against a very large field against very bad players like the WSOP main event.
Having more chips is a huge asset for risk management, but how do you get there while managing risk?
At any rate if there are about 2.2big blinds in the pot the any two vs ICM players is really close to an M of 2. If they open their hand ranges but fold too much at any stage of the hand and you manage the bet size accordingly, you can bennefit by gaining at a high rate with any two also. If they call too often when in the hand you benefit by not bluffing and extracting the full equity of the hands behind. I think a net gain of 2M per level is more than attainable and per rotation is difficult but doable. But we don’t need a NET gain of 2M, we need a net gain of 1M per rotation. That is about 2.2big blinds gained net per rotation. This is also for 12 minute blind structure and I had some difficulty finding the exact file I was looking for but even with this blind structure there are some online 15minutes increases and slower structure so such a high rate may be unnecessary as well.
Does this mean you should avoid all ins entirely? No, but it means you can substantially raise your standards for all in and find success with more passive styles for pot control and more bluff-catching rather than thin value betting, and bluffing very tactically*
Bluffing very tactically may mean picking the opponents who have too low of continue frequencies. It may mean less often, smaller bets.
