MATH: 45 extra chips to start 800-chip tourney adds >5% to ROI

[Nicholasp27]

take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)
ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill
If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament
10.51*1.265=13.29515 = what you expect to bring in due to skill advantage

13.29515/11=1.20865=20.865% ROI

that's 5.865 points higher than his usual 15%!!!


45 chips early on DOES make a difference

[Nicholasp27]

Now, someone tell me where the flaw is in the following double-up first hand strategy....


1600 stack vs 8 800 stacks...

.1844 is your new ev


18.44*1.265=23.3266
23.3266/11=2.1206 or 212% roi!!

now the games you don't double up first hand, you have -100% roi...

and the games where they all fold to your first-hand push, you will be at 20%, so still better than 15% you started out with if you fold hand one...

and if u wield a big stick for more than 26.5% skill advantage, your roi increases even more...not to mention $/hour implications and the increase in RB...

[schwza]

50/50 bust vs double up: 50% = 0. 50% = .184 or so by icm. on average, you get .092. not good.

[Nicholasp27]

yeah, that was the flaw...u aren't 0% roi if u bust

[Nicholasp27]

so you need to win 54.34% of the time if u do the double-up or bust strategy first hand to average out to .1 (.184/.1) that you have if you just fold

UNLESS

you have more than a 26.5% skill advantage when you are big stack...

if you are 50/50 first hand doubling up:

12.65/9.20=1.375

so if your skill advantage grows to over 37.5% roi and you can double up 50% of the time first hand, then it's worth it to you, as a 15% roi player normally, to go for it

however, u aren't gonna be 50% doubling up because you will face a better-than-average hand when called, so your random hand will be a dog

[microbet]

[ QUOTE ]
take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)
ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill
If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament
10.51*1.265=13.29515 = what you expect to bring in due to skill advantage

13.29515/11=1.20865=20.865% ROI

that's 5.865 points higher than his usual 15%!!!


45 chips early on DOES make a difference

[/ QUOTE ]

Seems like a wierd way to calculate it. I did this in the other thread for a 1000 chip game, but I would put your max at:

$1.65 (you normally make) + $.51 (you are given) = $2.16 (what you now get)

$2.16/$11 = 19.6% for an increase of 4.6%.

[Nicholasp27]

why do you only get 51 cents?

that's what ICM says the AVERAGE player gains...you are 26.5% better than the average player, so you should get more than 51 cents...

what's the error in that logic?

[Nicholasp27]

let's try to get some criteria for when you should go for the doubling up first hand...

we know if everyone folds to your push, you gain in roi

if u just steal the blinds to be 825, then ICM says 10.28

10.28*1.265=13.0042/11=18.22% roi



so X% of the time, you steal blinds for 18.22% roi
Y% of the time, someone calls you and you win those hands Z% of the time for a 212 roi
the other times you get called and lose for ev=0


X(13.0042)+YZ(23.3266)>12.65, then you should do it



let's see what % of time your hand has to hold up for different calling %s Y:


10=12.65-11.70378=.94622/23.3266=.04056/.1=40.56%
50=12.65-6.5021=6.1479/23.3266=.263557/.5=52.7%


ok, so let's say that the average player will call with top 15% hands

if you are sb, Y=.15
if you are button, Y=.2775
if you are button-1, Y=.385875


so sb, you need to win 45.6% against top 15% of hands...which hands do that?
so button, you need to win 50% against top 15% of hands...which hands do that?
so button-1, you need to win 51.8% against top 15% of hands...which hands to that?

and so on...

[microbet]

Well, in the end the 8000 chips are worth the same to a winner as a loser, so each chip is certainly not worth some static non-zero percentage more for a good player than for a bad player.

I really think determining exactly what a chip is worth to various players based on their skill is just impossible. One could probably learn some stuff about this with massive computer simulations, but they would indeed have to be truly massive.

It does not seem obvious to me that giving a certain number of chips would give more money per tourney to a good player than a bad one (or vice versa), so I just assume it is the same.

I generally think of good players as building equity by making winning decisions, not as having more equity per chip.

[raptor517]

[ QUOTE ]
take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)
ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill
If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament
10.51*1.265=13.29515 = what you expect to bring in due to skill advantage

13.29515/11=1.20865=20.865% ROI

that's 5.865 points higher than his usual 15%!!!


45 chips early on DOES make a difference

[/ QUOTE ]

i dont really care what any of these things say.. if you give me 45 extra chips my roi wont go up 5%. thats all there is to it. holla

[Nicholasp27]

well i don't think this 26.5% skill advantage stays static throughout the tournament, but i think that adding 45 chips only won't greatly affect that skill advantage


do you think that your average donk losing money every tourney will really gain 51 cents just from 45 chips in a no limit tourney? i think he'll gain less than 51 cents and winning players will gain more than 51 cent

[ilya]

[ QUOTE ]
[ QUOTE ]
take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)
ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill
If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament
10.51*1.265=13.29515 = what you expect to bring in due to skill advantage

13.29515/11=1.20865=20.865% ROI

that's 5.865 points higher than his usual 15%!!!


45 chips early on DOES make a difference

[/ QUOTE ]

i dont really care what any of these things say.. if you give me 45 extra chips my roi wont go up 5%. thats all there is to it. holla

[/ QUOTE ]

hey dude just cos you suck don't mean he's wrong! /images/graemlins/blush.gif

[Nicholasp27]

well i think math shows it does

however, you could maybe make an argument that ICM is more inaccurate with 10 stacks than with 4 stacks and ICM is overly exaggerating the effect of 45 chips in an 8k chip tournament with 10 players left

[microbet]

For me, this discussion is just a fun math problem.

Pushing early with less than top hands is another issue. I don't really want to discourage people from doing it, but I'm not yet heartless enough to encourage it.

[Nicholasp27]

well, this would only give you a narrow range of hands...i don't have pokerstove on this comp, but what hands can be 50% against the range of top 15% of hands?

i suspect it's a small number of hands, so it would tell you that you can only push a small number of hands only in late position only with no limpers

[raptor517]

[ QUOTE ]
well i think math shows it does

however, you could maybe make an argument that ICM is more inaccurate with 10 stacks than with 4 stacks and ICM is overly exaggerating the effect of 45 chips in an 8k chip tournament with 10 players left

[/ QUOTE ]

math doesnt show it. none of this math done so far is good enough to take into account even half of the factors in sng poker. im still waiting to be proven wrong. you give me those extra 45 chips and take 5 from everyone, you say u get 5.5%, i say i get closer to 2 or 3. ill go into why later. that includes the fact that im a better player of course. holla

[curtains]

These arguments should assume that all players are equal to begin with. Once we solve for that, then we should seeing how skill level affects the results.

[raptor517]

[ QUOTE ]

These arguments should assume that all players are equal to begin with. Once we solve for that, then we should seeing how skill level affects the results.

[/ QUOTE ]

the original argument had the player beginning at a 15% roi, did it not? thus.. all the math was based upon that no? holla

[microbet]

At 3% you're practically agreeing. I think the 5.5% math is not good. I put the cap at 4.6% in that example and as I mentioned in the other thread, I think it HAS to be less and the higher your ROI is the less you gain from the chips.

I think the opposition in the other thread was thinking .4% was more like what the increase in ROI would be.

Intuition tells me the totally average player, who loses exactly the rake, is going to be who makes the most from 45 extra chips.

I think there are two factors that determine the value of those chips: The FREQUENCY they make a difference, and the EFFECTIVENESS of the difference they make when they do.

The chips will make a difference infrequently for an excellent player, but that player will make excellent use of them. The chips will make a difference frequently for a terrible player, but that player will make terrible use of them. But I don't think these cancel out or compound, I think this product shrinks both ways.

Here's a thought experiment, which is just to intuit, not meant as hard logic:

Take the perfect player. He's psychic, knows everyones' hole cards, and his mind is the freakin' pattern mapper. His ROI is maximal, he gets first every* time. The chips do nothing for him.

Take the perfectly terrible player. He actually doesn't have a mouse, so he just folds every single hand. His ROI is minimal, he loses his entire buyin every* time. The chips do nothing for him.

I'm of course willing to see math that suggests otherwise, but until I see some justification that the math is working as it does in reality I give my gut some credit.


*Yeah yeah, a tiny amount of the time it doesn't happen. Kinda interesting... there can exist only one player with maximal ROI but there can exist seven with minimal ROI...

[JJKillian]

[ QUOTE ]
[ QUOTE ]
take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)
ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill
If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament
10.51*1.265=13.29515 = what you expect to bring in due to skill advantage

13.29515/11=1.20865=20.865% ROI

that's 5.865 points higher than his usual 15%!!!


45 chips early on DOES make a difference

[/ QUOTE ]

i dont really care what any of these things say.. if you give me 45 extra chips my roi wont go up 5%. thats all there is to it. holla

[/ QUOTE ]

nod and agree

JJ

[curtains]

Shouldn't 2+2 not be a forum where one person presents a mathematical reason for something and someone else just offhandedly says "I disagree, I don't care what you say?"

Do you all realize how absurd that is? Disagree if you want, but don't do so in such a juvenile fashion! Explain a few well thought out reasons why you disagree!

[ QUOTE ]

Shouldn't 2+2 not be a forum where one person presents a mathematical reason for something and someone else just offhandedly says "I disagree, I don't care what you say?"

Do you all realize how absurd that is? Disagree if you want, but don't do so in such a juvenile fashion! Explain a few well thought out reasons why you disagree!

[/ QUOTE ]

But what do you do when you agree?

[Nicholasp27]

ok, it's a small range:

45.6% (can push in sb if folded to you): ato+, 77+, kqo
50% (can push in button if folded to you): ajo+, 99+
51.8% (can push in button-1 if folded to you): ajs+,tt+

so the math shows it's +ev to push those hands first in pot in those seats first hand of the game...that is, if u think they only call with top 15% of hands (77+, a7s+, k9s+,qts+,jts+,ato+,kto+)

if you want to tighten or loosen their calling ranges, i can run the numbers and get new hands that are +ev

[curtains]

I don't know what you are saying, but whom should one be expected to believe when one provides well thought out mathematical and logical reasoning behind something and someone else dismisses it offhand saying that they "don't care" about their analysis.

[ QUOTE ]

I don't know what you are saying, but whom should one be expected to believe when one provides well thought out mathematical and logical reasoning behind something and someone else dismisses it offhand saying that they "don't care" about their analysis.

[/ QUOTE ]

Sorry, I misunderstood you when I wrote that. But I do want to say something about this anyway. I don't see any mathematical or logical reasoning that this is a valid approach to including skill in ICM equity modeling. I do see mathematical and logical reasonings to provide conclusions using this approach, but not such justifications for said approach. So when someone says "it isn't happening" they're just saying "sure, that math/logic is fine, but I don't believe it applies to reality and until you provide some argument that it does I have no reason to"

[Nicholasp27]

ok, even skill

10 is expected income per tourney
-1/11=-9% roi

10.51 is expected income if u get 45 chips extra
-.49/11=4.4%

so your roi went up 4.6 actual points

[Nicholasp27]

well take away the skill advantage (make hero an average player) and it's still >4 added to your actual roi number

that's the math of what ICM is saying

the question is: do u think that in practical applications, icm is correct in giving you 1% of the prize pool (50 cents) for gaining 5% of your stack with even stacks? if u think icm is inaccurate, then that is one argument...but i haven't heard an argument why the math, assuming that icm is correct, is wrong

icm is just a model, and maybe it is overstating how important 45 chips are in a sng...it's 5% of your stack now, but when blinds are 15%+ of your stack in future levels at the bubble, does 5% of your stack really matter that much?

[donny5k]

It cannot be proven mathematically, but doesn't everyone's intuition just scream that ICM is plain wrong in this case (mathematical intuition as well as from experience)?

This is like chip count chops at the end of MTT's, everyone "knows" they give the big stacks too much.

I believe ICM is perfect for identical players, and quite good enough, but definitely not right, for players of varying skill levels. I'm pretty sure you believe that too otherwise you wouldn't be doing this. I just think your approach of adding skill into ICM is incorrect, and I think that only because I see no obvious reason that it is correct and my intuition disagrees with the results

[Nicholasp27]

so if u fully believe icm, then you agree that if all players are equal skill, then this 845 stack gained 4.4% in roi over just having 800?

[curtains]

For what it's worth my intuition also tells me that ICM is wrong in this situation but not by very much. What I mean is that the amount these chips will add to your ROI is quite signifigant, but maybe not quite as much as math would say. However the number will be close, there is no reason why it shouldn't be.

[bones]

I think the initial math accounting for skill is off in the first post.

The problem with trying to practically apply this is even for break even players, and especially for good players (I'm not sure what we've decided to go with in this thread), is that by trying to pick up these 45 chips, you risk going broke and forfeiting the opportunity to find better advantages. The whole cash game/tourney thing.

If there was a way of adding the 45 chips with almost zero risk, I would certainly take it. But in very loose games full of sheriffs, you're just not gonna find that.

While obviously the 45 extra chips (and the 5 less from each opp.) will add a tiny bit to your ROI, it's wildly optimisitc to think that it can turn a 15% ROI to a 20% ROI. This is one of those things that "feels" wrong from experience.

[protoverus]

[ QUOTE ]
It cannot be proven mathematically, but doesn't everyone's intuition just scream that ICM is plain wrong in this case (mathematical intuition as well as from experience)?

This is like chip count chops at the end of MTT's, everyone "knows" they give the big stacks too much.

[/ QUOTE ]

Yes, I cannot possibly agree that 45 chips would add that much to anyone's ROI. The ICM model necessarily ignores many factors. It has to because the game is too complex to analyze exhaustively. I think that over the course of an entire SNG, those unaccounted for factors dilute the value of 45 extra chips.

It's like ignoring friction in a physics problem. Since calculating friction is hard and sucky, it's often ignored and a usable answer, though not precise, is arrived at.

That's fine if you are pushing a heavy box 5 feet, but if you are pushing it a mile and a half that friction is going to make itself known. I think its the same here. The ICM is overestimating the value of 45 chips at the start because it's ignoring the frictional elements for the entire duration of the SNG...which would dilute the value of those chips. Just what are all the frictional elements? I don't know, but my experience tells me they are there.

Be well.

[ QUOTE ]
so if u fully believe icm, then you agree that if all players are equal skill, then this 845 stack gained 4.4% in roi over just having 800?

[/ QUOTE ]

Yes I do, and I'm interested to see where you go from there

[curtains]

I didn't expect friction to come into this argument or the appearance of a "heavy box". /images/graemlins/grin.gif

[JJKillian]

[ QUOTE ]
so if u fully believe icm, then you agree that if all players are equal skill, then this 845 stack gained 4.4% in roi over just having 800?

[/ QUOTE ]

although I don't agree with this theory, I do agree with what you just posted.

My issue with it is it just seems this arguement could keep going. Wouldn't it also be the correct play next orbit in the bb if every one folds to you. Then the next so on and so on. Of course there would be exceptions, short stack in the bb/sb, lose player in the sb/bb, so on and so forth.

Eventually the whole game could be plugged into a computer with other icm/math models that we all come up with. And then this computer would play perfect poker.

I finally get to my point. Mr Kasperov (think it is spelled right), has proven a computer cannot beat him. Granted he is the best Chess player in the world and none of us here are the best poker player in the world. But you cannot get there if a computer (icm) is playing for you.

I just feel being a slave to the math of poker severly limits someones ability in the game. Knowing when someone is on a bluff is very valuable and math won't teach you that. So although this math is probably right, it is also probably wrong in the over all picture. Just my opinion.

One final thought I have for this thread that I would love to see some responces to is this. I remember a few months back when anything "Gigabet" was a super hot topic here, now it seems anything ICM is. The Gigabet posts usually revolved around some -ev play that he made, knowing that it was -ev when he made the play. ICM is all about the +ev no matter how small sometimes.

JJ

[protoverus]

Sorry for using an analogy... but it's the best I could do.

I guess the point was that, as a model, the ICM simplifies an SNG so that we can do some analysis. That simplification can lead to errors, especially when considered over the course of an entire SNG.

Or do you think the ICM is an almost entirely accurate model of the SNG? I don't, even though it's been the single most useful tool in improving my understanding of the game.

[Nicholasp27]

u are always a slave to math in poker, sorry

if u know someone is bluffing, that just changes the variables in ur equation

[curtains]

I think math is extremely important. Basically the best player will be the one that has a great intuition/pattern recognition ability and a fantastic mathematical understanding. However one of these two facets alone is usually not enough to make someone a great player.

I think I've always had relatively decent instincts, but as soon as I worked very hard on the math, my game, my confidence level and my results instantly shot through the roof.

[microbet]

Without understanding the effect why do you assume the 45 chips would be diminished like friction? Why wouldn't it snowball? I would start with it doing neither.

When you say it wouldn't have much effect on ROI are you talking 2-3% or .2-.3%?

[protoverus]

[ QUOTE ]
Without understanding the effect why do you assume the 45 chips would be diminished like friction? Why wouldn't it snowball? I would start with it doing neither.

When you say it wouldn't have much effect on ROI are you talking 2-3% or .2-.3%?

[/ QUOTE ]

2-3% would be large IMO. .2-.3% sounds more resonable. However, i'll have to think again now...why wouldn't it snowball? That's a good point. I suppose it could. Maybe it could do either, but what factors would control it? Relative skill level most likely, but we ignore that in ICM...

I guess that in my mind, the ICM works great hand to hand because it's easier to ignore skill differences in a single hand situation. But over the course of an entire SNG? The differences between the assumption (equal skill) and reality must grow larger.

At any rate I appreciate the discussion which is making me think about the game in a fresh way again.

be well.

[JJKillian]

[ QUOTE ]
I think math is extremely important. Basically the best player will be the one that has a great intuition/pattern recognition ability and a fantastic mathematical understanding. However one of these two facets alone is usually not enough to make someone a great player.

I think I've always had relatively decent instincts, but as soon as I worked very hard on the math, my game, my confidence level and my results instantly shot through the roof.

[/ QUOTE ]

not trying to lower the value of the math in this reply. But this is going to sound nuts, and I have no way of totally proving it. But in my profession I am a sales trainer. Not like the guy at the local car dealership. I go in and train different types of sales people across many types of products. Many of which I am very limited in my knowledge about.

What I am getting at here is this. You said your confidence went up. When that happens something happens to your thought process and just makes things happen. You make better decisions, and just over all think straighter. I have taken a regular salesman that has the same product and same basic client base as the number 1 and actually made him compete with the number 1 guy by simply increasing his confidence.

Point is, I guarantee the math helped you, but gaining in your confidence lvl may have helped even more.

I am sure everyone that has been running bad or good knows it comes in streaks. Well why does variance run in streaks? That doesn't make a ton of sence. The word is very clear, it is variance. So why should it runs in streaks? I think it is because one or two bad beats starts a slight lose in confidence. Which then lowers the players ability which will eventually lead to a bad decision which lowers confidence, so on and so forth. It also runs the other way when your running good. It is amazing the things that happen when your confidence is up.

Just a random theory I have.

JJ

[curtains]

.2-.3% is totally unreasonable! You are gaining chips that have a theoretical value of like 4-5% ROI! To pretend that there is even a possibility that they are worth only about 1/20th of their actual value is absurd.

[curtains]

Why is it so hard for some poker players to just admit that math might be very important towards acheiving good results? You are not the only one who is so resistant to this concept.

I have done a huge amount of work on the math and I have many charts dealing with work I've done on particilar situations to back this up, things that almost no one in the world has, and these have aided me greatly. I constantly see winning players making egregious mathematical mistakes late in tournaments/sit and gos, because their instincts lead them astray.

While doing this I learned that a lot of the things I used to do were actually incorrect, and have been able to adjust them because of the math work I did. Suddenly I have unbeleivable results for an extended period of time, almost immediately after doing this work. This is not because I have "extra confidence", its because I did a lot of work and have more knowledge than my opponents and more knowledge than I had before.

Trust me I had plenty of confidence before I did this work and I see plenty of people on 2+2 who seem more confident than they should be based on how much they seem to know. Confidence doesn't win you pots or tournaments.

[Nicholasp27]

1070 vs 770 stacks = .1298 ev which is an18% roi player

so let's give him that to handicap the game...he's an 18% player and hte rest are all average donks


NOW give him 5 chips from everybody

ev=.1346


so 22.36% roi, or 4.36 extra roi points!


this may be more accurate as well, because you handicapped icm so icm knows he's an 18% player instead of assuming he's an average player if u give 800 to everyone...

Amen brother

Furthermore, I can't imagine why doing something counterintuitive would give you extra confidence until AFTER the explosive results start coming in, at least the first time you go for the Trust The Math methodology

[ QUOTE ]
1070 vs 770 stacks = .1298 ev which is an18% roi player

so let's give him that to handicap the game...he's an 18% player and hte rest are all average donks


NOW give him 5 chips from everybody

ev=.1346


so 22.36% roi, or 4.36 extra roi points!


[/ QUOTE ]

So he gets less than the 4.4% the average player got, this is a valid comparison right?

So this would mean I was correct that the dead-on average player is the one who would gain the most from 45 extra chips? Or did nobody disagree with that anyway...

[JJKillian]

[ QUOTE ]

Why is it so hard for some poker players to just admit that math might be very important towards acheiving good results? You are not the only one who is so resistant to this concept.

I have done a huge amount of work on the math and I have many charts dealing with work I've done on particilar situations to back this up, things that almost no one in the world has, and these have aided me greatly. I constantly see winning players making egregious mathematical mistakes late in tournaments/sit and gos, because their instincts lead them astray.

While doing this I learned that a lot of the things I used to do were actually incorrect, and have been able to adjust them because of the math work I did. Suddenly I have unbeleivable results for an extended period of time, almost immediately after doing this work. This is not because I have "extra confidence", its because I did a lot of work and have more knowledge than my opponents and more knowledge than I had before.

Trust me I had plenty of confidence before I did this work and I see plenty of people on 2+2 who seem more confident than they should be based on how much they seem to know. Confidence doesn't win you pots or tournaments.

[/ QUOTE ]

sorry wasn't trying to insult you, just throwing in a theory I cannot prove. Actually said that in the beginning of the post. And also put in I am sure the math helped a ton also. You have to have the math, no doubt about it. But a machine is better than a human at math. But to this day no machine can beat the best at their respective game. So if math is the end all answer then why hasen't one of the elite poker players simply programmed a machine to crush us all?

JJ

[Nicholasp27]

because of what i said before...it's all about what you set the variables to


if u pick up that someone is bluffing from a tell, then you change the variables...a machine may not pick up on that...

u can think of other examples like that

the fact is, everything in life is mathematical...but very very complex equations in which we constantly and intuitively change the value of the variables

[Nicholasp27]

well we can do the numbers with a 1% player's handicap, 5%, 10%, etc and see if it trends downwards...

but from 4.4 to 4.36 isn't much considering it's 18% vs -9% roi players

not to mention, this isn't perfect, as we are manipulating the stacks...but it may be the most accurate to do within icm's scope, as it is handicapping the players based on skill level

[raptor517]

[ QUOTE ]

I don't know what you are saying, but whom should one be expected to believe when one provides well thought out mathematical and logical reasoning behind something and someone else dismisses it offhand saying that they "don't care" about their analysis.

[/ QUOTE ]

curtains. you cant honestly tell me, that these numbers 'prove' that with 45 extra chips, yer roi will go up 5.5% or whatever. thats just ludicrous. i mean.. say you start with.. 900 chips in a 800 chip game, and everyone is at 790 to start. what is yer roi then? do the returns diminish? assuming equal play, they shouldnt. however, i can guarantee, no matter who the player is if he is a winner, his roi will not be increased anywhere near 10%. holla

[curtains]

I'm not telling you anything other than "I think I'm right and I don't care about your analysis" isn't an effective debating tecnique.

[raptor517]

[ QUOTE ]

I'm not telling you anything other than "I think I'm right and I don't care about your analysis" isn't an effective debating tecnique.

[/ QUOTE ]

this debate is only useful assuming EQUAL skill. and even then.. i duno, i dont have much of an effective argument, but my intuition and feel has done me pretty well so far, and i think that the 'math' used here is off. ill find a more useful reason for my beliefs later. holla

[curtains]

[ QUOTE ]
Basically the best player will be the one that has a great intuition/pattern recognition ability and a fantastic mathematical understanding

[/ QUOTE ]

Please note that I said in my previous post that the combination of math and instinct is the key. I also stated that being good at one and not at the other will usually stop someone from being a complete/great player.

So your comments about why hasn't someone programmed a program to crush us all are not really directed towards what I was saying.

[ QUOTE ]
well we can do the numbers with a 1% player's handicap, 5%, 10%, etc and see if it trends downwards...

[/ QUOTE ]

Sure but I don't think it's necessary, we have some symmetry here too

[ QUOTE ]
but from 4.4 to 4.36 isn't much considering it's 18% vs -9% roi players

[/ QUOTE ]

Which is accurate on average

[ QUOTE ]
not to mention, this isn't perfect, as we are manipulating the stacks...but it may be the most accurate to do within icm's scope, as it is handicapping the players based on skill level

[/ QUOTE ]

Totally agree

[donny5k]

[ QUOTE ]
why hasen't one of the elite poker players simply programmed a machine to crush us all?


[/ QUOTE ]
simply? haha.

[raptor517]

[ QUOTE ]
[ QUOTE ]
why hasen't one of the elite poker players simply programmed a machine to crush us all?


[/ QUOTE ]
simply? haha.

[/ QUOTE ]

prolly not as hard as you would think. holla

[donny5k]

You probably realize this but I thought it was worth adding that the reason math isn't enough to maximize expectation is because your opponents don't play anywhere near optimally. Instinct/experience is helpful at exploiting these players while a purely mathematical strategy would still beat them, just not for as much.

[donny5k]

It's probably harder than you think. /images/graemlins/cool.gif

[raptor517]

[ QUOTE ]
It's probably harder than you think. /images/graemlins/cool.gif

[/ QUOTE ]

nope, its exactly as difficult as i think it is /images/graemlins/wink.gif holla

[lorinda]

Most of our edge comes from bubble play.

45 chips doesn't help with Folding Equity to the tune of 5% ROI

This is the major flaw. There are others but this subject is stupid.

Edit: That is to say the thread is stupid, the original post was fine, just flawed.

Lori

[raptor517]

[ QUOTE ]
Most of our edge comes from bubble play.

45 chips doesn't help with Folding Equity to the tune of 5% ROI

This is the major flaw. There are others but this subject is stupid.

Edit: That is to say the thread is stupid, the original post was fine, just flawed.

Lori

[/ QUOTE ]

thankyou. holla

[Nicholasp27]

that's the key point i was thinking about on my drive home

the way 2+2ers play makes it so that 45 chips don't matter as much as to a looser player....if we are tight until bubble and then pushbot with 150 blinds, 45 doens't matter

if u are gonna use that 45 to limp in 3 pots early on, then it'll add 'some' ev because sometimes u'll win pots in those 3, which will add to ur ev

if u are gonna play your standard 2+2 game even with the 45 extra chips, then the extra ev wouldn't be 5%


that's the problem with handicapping the stacks..u also gotta handicap the 45 chips...to us it may only be worth 5 chips ev to icm, whereas to a lag it may be worth more...

[Nicholasp27]

btw, i disagree that the thread is stupid

even if we are all wrong in the thread, it is sparking good discussion and getting people to think about icm more critically, so it's a good thread regardless of the 'right answer'

[Irieguy]

Seriously, it is not worth the time to prove that this whole idea is completely off.

But, Curtains is correct when he says that it is not sufficient to rebuff somebody's mathematical argument by just saying "this is wrong and I don't care what anybody says."

So, in an effort to put this nonsense to bed while still presenting a mathematical reason for why this is nonsense... I will suggest the following demonstration:

-Use Nicholas' model to calculate how much your ROI should increase if you start with 7991 chips and the other 9 players start with one chip.

Irieguy

[microbet]

Nic is not the only one to propose any math in here. I had stated before that the model should have points where the extra chips make no difference (ROI is perfect - Give someone all the chips). Perhaps you could start with ROI = 0 player and 1 extra chip not being adjusted and then figure some function that reasonably fits those points.

I don't think it is worth all that, and I probably couldn't do it, but it really isn't necessary to decide whether we are talking about a 3% difference in ROI or a .3% difference.

[microbet]

I think the extra chips are worth more to the average player than to the +ROI player. (that's based on ICM, player's styles could affect whether or not this is true in a specific case) I'm not sure what happens to the -ROI players.

[pergesu]

[ QUOTE ]
Most of our edge comes from bubble play.

45 chips doesn't help with Folding Equity to the tune of 5% ROI

This is the major flaw. There are others but this subject is stupid.

[/ QUOTE ]
What if the initial disparity allows you to push one less time each tourney? Even if you're making +EV pushes, every time you push you expose yourself to busting out. If the stack differences allow you to pass on even just one round before pushing, that gives you an extra round to let someone else bust without the chance of busting yourself.

This kinda makes sense to me, kinda doesn't. I'm just presenting it as an idea, and hopefully someone smart can come along and make sense of it.

[curtains]

[ QUOTE ]
Seriously, it is not worth the time to prove that this whole idea is completely off.

But, Curtains is correct when he says that it is not sufficient to rebuff somebody's mathematical argument by just saying "this is wrong and I don't care what anybody says."

So, in an effort to put this nonsense to bed while still presenting a mathematical reason for why this is nonsense... I will suggest the following demonstration:

-Use Nicholas' model to calculate how much your ROI should increase if you start with 7991 chips and the other 9 players start with one chip.

Irieguy

[/ QUOTE ]

One note, I think that if you double up immediately, that theoretically the chips you gain are worth about 85% of their value. This discounts any advantageous situations you may find due to your large stack, but is from a straight ICM viewpoint. This percentage should be higher when the chips you gain are smaller, thus when you gain 50-100 chips, the chips should be worth very close to their actual value. This means that if you double up right away every tournament, and you are playing in a tournament with all equally skilled opponents, your ROI should be about 85%.

I know there are reasons why it might be higher or lower, than this above figure but the correct percentage should be at least in this ballpark.

The example of having all of the chips but 9 and giving one to each opponent doesn't really prove anything other than the fact that the more chips you gain the less they are worth in theory, which I believe that everyone involved in this debate already understands.

[curtains]

[ QUOTE ]
Seriously, it is not worth the time to prove that this whole idea is completely off.

But, Curtains is correct when he says that it is not sufficient to rebuff somebody's mathematical argument by just saying "this is wrong and I don't care what anybody says."

So, in an effort to put this nonsense to bed while still presenting a mathematical reason for why this is nonsense... I will suggest the following demonstration:

-Use Nicholas' model to calculate how much your ROI should increase if you start with 7991 chips and the other 9 players start with one chip.

Irieguy

[/ QUOTE ]

Also when you say that the idea is "completely off", do you mean that the chips you gain are worth almost nothing? Are they worth 2% to one's ROI? 3%?

In each of these cases you are assuming the chips are worth very little compared to their actual value, yet the amount of ROI percentage points that you gain is quite signifigant.

If you think that these chips will increase your ROI by less than 2% I think that this is clearly incorrect. There is no way that your chips should be worth that much less than their actual value and I see no logical reason to believe this should be the case.

If you think the 45 chips are worth just 3% then fine, 3% is not that much different than 4% in any case.

COuldn't bother to read all responses. Two things:

1) ROI is not 0% if you bust, it's -100%.
2) your odds of winnning a push on your first hand are much less than 50%, as your opponents will choose good hands to call you with.

[Irieguy]

[ QUOTE ]
Nic is not the only one to propose any math in here.

[/ QUOTE ]

I'm not sure, because I didn't really put a lot of thought into it, but I think the problem with your model is that it doesn't account for the fact that a SNG is a percentage-payout tournament and that chip value decreases as the number of chips that you have increases.

In other words, if you gave me 45 extra chips everytime I went broke, my ROI would skyrocket. But if you add 45 chips to my 1000 chip stack, it wouldn't matter.

Maybe everybody already understands this relationship, as Curtains says, but it is blatantly disregarded in these ICM-based estimates of how much these extra chips would be worth.

Irieguy

[Irieguy]

[ QUOTE ]


The example of having all of the chips but 9 and giving one to each opponent doesn't really prove anything other than the fact that the more chips you gain the less they are worth in theory, which I believe that everyone involved in this debate already understands.

[/ QUOTE ]

Really? I thought the fact that the result is an impossible ROI figure shows that the formula is flawed.

If you say "use this formula to show how 45 extra chips increases your ROI by 5.86%," but the same formula will yield nonsensical results when you plug in other values that are within the real limits of your model... then the formula is bogus.

Do you really think Nic's calculations are correct?

Irieguy

[gumpzilla]

[ QUOTE ]

I'm not sure, because I didn't really put a lot of thought into it, but I think the problem with your model is that it doesn't account for the fact that a SNG is a percentage-payout tournament and that chip value decreases as the number of chips that you have increases.

[/ QUOTE ]

I haven't looked at it much either, but if we're talking about the original post I think it's pretty clear that it's the ad hoc way a skill advantage is handled that is the problem. The ICM alone is good enough to handle the percentage payout issues, the problem is that no matter how skilled you are your ROI is going to be pretty much the same once you've somehow assembled 99% of the chips. You can't just say that a 15% ROI player is always going to have a 24% or whatever better edge than a theoretically average player dropped into his exact same position.

[Irieguy]

[ QUOTE ]


If you think that these chips will increase your ROI by less than 2% I think that this is clearly incorrect. There is no way that your chips should be worth that much less than their actual value and I see no logical reason to believe this should be the case.

If you think the 45 chips are worth just 3% then fine, 3% is not that much different than 4% in any case.

[/ QUOTE ]

I think that increasing your true expected ROI by an additional 3-5 points demonstrates a pretty significant improvement in performance.

45 chips added to 1000 is not enough to accomplish this. 45 chips added to 1 chip would be more than enough to accomplish this.

This is not just incredulousness on my part. It is based on a thorough understanding of other value and equity relationships in percentage payout structure tournaments. I'm relatively sure I could explain it, but it would take a moderate amount of consideration and math... more than I feel this project is worth.

If my unwillingness to prove my point in detail disqualifies me from further comment, I can accept that and I will withdraw from the conversation.

But I will suggest that any mathematical "proof" demonstrating that a 45 chip addition to a 1000 chip stack will increase your ROI by >3% can be easily shown to be flawed... and that type of debunkery takes little effort and is something I may be willing to do for at least a short while longer.

Irieguy

[microbet]

ICM does discount the additional chips. The 45th chip added is worth less than the 44th chip added to your 1000 stack. (or all chips are devalued, same diff)

I also don't think this is worth the trouble, IF we are talking about 3% vs. 4% or even 5%. So, if you think around 3% is fine, then ok, whatever.

I'm just wondering if anyone is thinking it is more like .3% as I think that big a difference would have clear implications on playing style.

Also, I just want to make sure everyone knows, as far as I'm concerned it's all a good natured discussion. I, and I'm sure many others, are very interested in issues where two of the best posters (in my opinion, if not yours) disagree.

[curtains]

[ QUOTE ]
[ QUOTE ]


If you think that these chips will increase your ROI by less than 2% I think that this is clearly incorrect. There is no way that your chips should be worth that much less than their actual value and I see no logical reason to believe this should be the case.

If you think the 45 chips are worth just 3% then fine, 3% is not that much different than 4% in any case.

[/ QUOTE ]

I think that increasing your true expected ROI by an additional 3-5 points demonstrates a pretty significant improvement in performance.

45 chips added to 1000 is not enough to accomplish this. 45 chips added to 1 chip would be more than enough to accomplish this.

This is not just incredulousness on my part. It is based on a thorough understanding of other value and equity relationships in percentage payout structure tournaments. I'm relatively sure I could explain it, but it would take a moderate amount of consideration and math... more than I feel this project is worth.

If my unwillingness to prove my point in detail disqualifies me from further comment, I can accept that and I will withdraw from the conversation.

But I will suggest that any mathematical "proof" demonstrating that a 45 chip addition to a 1000 chip stack will increase your ROI by >3% can be easily shown to be flawed... and that type of debunkery takes little effort and is something I may be willing to do for at least a short while longer.

Irieguy

[/ QUOTE ]


Okay please commence with your debunkery if you have the time.

[FlyWf]

Well, I'll try a couple non-mathematical answers for why the 4% estimate is wrong:

45 chips is less than half a BB at level 4. Typical 2+2 play is to essentially play extremely tight early. Thanks to the extra 45 chips, instead of a 6 or 7 BB stack when you are carddead you get a 6.5 or 7.5 BB stack. That is not a significant source of folding equity, and FE is the weapon by which good players gain an edge.

ICM ignores blinds. If blinds were 10/15 the entire tournament the 45 chips would produce a much larger edge, especially if the tourney didn't pay 2nd and 3rd place. That leads into another reason:

There are successful endstates for a SNG that result in you having 0 chips.

Not to sound too much like that rambling and incoherent Gigabet post about lines and chunks, you don't have a stack of chips. You have a stack of BBs. So you aren't gaining 45 chips, you're getting 3 BB when you already had a 53 BB stack. But there's no strategic edge to that. You can still stack off a player. You're still absolutely [censored] if you lose an allin.

As the tourney progresses your edge in BB decreases and decreases and by around level 5 it no longer matters at all.

[Apathy]

Another assumption that is incorrect here that I don't think anyone has pointed out yet is if you are assuming someone is a 10% roi player to start with and then does this push first hand thing THEY WILL NO LONGER BE A 10% ROI PLAYER BECAUSE THAT IS A TERRIBLE LOSING PLAY.



Also why do they get 45 chips I thought the blinds started at 10/15 (Not that that matters a ton to the discussion, just saying.)

[curtains]

Apathy this all threads from another thread where I suggest to raise on the button with 88 and 15-30 blinds. A few counterarguments were that the 45 chips in blinds don't actually mean anything, and so I said that I disagree because of this.

[Apathy]

[ QUOTE ]

Apathy this all threads from another thread where I suggest to raise on the button with 88 and 15-30 blinds. A few counterarguments were that the 45 chips in blinds don't actually mean anything, and so I said that I disagree because of this.

[/ QUOTE ]

Gotcha I'll check that one out.

My point still stands that it is faulty logic to justify a play that you wouldnt usually do by saying what it will add to your CURRENT roi because your CURRENT roi is not the same since you are changing the way you would usually play in certain situations, so you can't just add roi points for the balance of the increase in chips from your new play.

[curtains]

Dude I dunno what you're talking about, at least that's nothing I've ever argued!

[microbet]

My part in the discussion was purely based on just starting with different chip stacks. Academic. I'm not talking about crazy allins on the first hand.

[Apathy]

[ QUOTE ]

Dude I dunno what you're talking about, at least that's nothing I've ever argued!

[/ QUOTE ]

Meh never mind is was in the original post and the replies on the first few pages, you never said it.

[Apathy]

[ QUOTE ]
My part in the discussion was purely based on just starting with different chip stacks. Academic. I'm not talking about crazy allins on the first hand.

[/ QUOTE ]

It was pretty much the NicolasP posts at the start of this thread discussing some sort of all in early to double up strategy. I just came into this discussion at a weird time so nevermind.


Basically someone makes that post about taking flips early once every few weeks and I have some strong views about it and I thought that was what this discussion was based around after skimming through it.

[Insty]

Here is some mathematical proof that the original post was erroneous:


[ QUOTE ]
take a 15% ROI player in 10/1 tournaments
1.15*11=12.65/tournament (-11 buyin = 1.65 profit)


[/ QUOTE ]
Ok.

Some defintitions:
ROI(1) = 15%
$ENTRY(1) = 11
$WON(1) = ROI(1) * $ENTRY(1)
$WON(1) = 1.15 * 11
$WON(1) = 12.65

[ QUOTE ]

ICM says they should only bring in 10/tournament if even skilled
12.65/10=26.5% advantage due to player skill


[/ QUOTE ]

Ok, you have made up a new measurement called skill advantage.
Which you have defined as $WON/$EV

SA(1) = $WON(1) / $EV(1)
SA(1) = 12.65 / 10
SA(1) = 1.265

[ QUOTE ]

If you start with 845 chips and everyone else has 795 chips, then ICM says you should bring in 10.51/tournament


[/ QUOTE ]

ok.. It does. (This is also approximately -4.4% ROI)
$EV(2) = 10.51

[ QUOTE ]

10.51*1.265=13.29515 = what you expect to bring in due to skill advantage


[/ QUOTE ]

This is where things go wrong.
I dont think you can use skill advantage like this. But the multistep way makes it hard to see why.
This is what you have done so far.
$WON(2) = $EV(2) * SA(1)
$WON(2) = 10.51 * 1.265
$WON(2) = 13.295


[ QUOTE ]

13.29515/11=1.20865=20.865% ROI
that's 5.865 points higher than his usual 15%!!!


[/ QUOTE ]

This is just getting wronger.

What you have done.
ROI(2) = $WON(2) / $ENTRY(2)

Lets expand $WON(2):
$WON(2) = $EV(2) * SA(1)
ROI(2) = ($EV(2) * SA(1)) / $ENTRY(2)

Now lests expand SA(1):
SA(1) = $WON(1) / $EV(1)
ROI(2) = ($EV(2) * ($WON(1) / $EV(1)) ) / $ENTRY(2)

Lets expand WON(1):
$WON(1) = ROI(1) * $ENTRY(1)
ROI(2) = ($EV(2) * ((ROI(1) * $ENTRY(1)) / $EV(1)) ) / $ENTRY(2)


Since
ENTRY(1) = ENTRY(2) = 11

We can cancel them.
ROI(2) = ($EV(2) * ((ROI(1) / $EV(1)) )


Which gives us ROI(2) in terms of EV's and ROI(1) that is bad.
Which of these are you claiming is your real ROI?

You can transpose the above to:

ROI(2) / ROI(1) = EV(2) / EV(1)

Which just gives you some unsurprising ratios.



What you CAN do is recalculate skill advantage SA(2):

If you start with 845 chips:
SA(2) = $WON(1) / $EV(2)
SA(2) = 12.65 / 10.51
SA(2) = 20.4%

So if you always start with a 45 chip advantage you have a smaller skill advantage than before.

[ QUOTE ]

45 chips early on DOES make a difference


[/ QUOTE ]

Yes, you have a smaller skill advantage.

[Nicholasp27]

i'm not a proponent of doing the double-up first hand thing...i was simply doing an exercise to see what this argument would say about doubling-up first hand

if u got to my post with results, it showed you can push 99+,ato,ajs,kqs only from the sb if bb calls with top 15% of hands and everybody else folded first

it doesn't say to push any2 from any position with any number of limpers...it is a very small % of the time that you can do it...but it's based on the theory that 45 chips give you 4 roi points

[Nicholasp27]

yeah, that's why we later abandoned the whole skill advantage issue and just looked at icm only, so an average player...icm DOES discount the value of added chips as your stack gets bigger...


it says you go from 10 to 10.51 if you get 45 chips to start tourney (5 from each player)...that's 4.4 roi points

now, someone tell us why that is wrong for an average skilled player...

[Nicholasp27]

yeah, that's why we later abandoned the whole skill advantage issue and just looked at icm only, so an average player...icm DOES discount the value of added chips as your stack gets bigger...


it says you go from 10 to 10.51 if you get 45 chips to start tourney (5 from each player)...that's 4.4 roi points

now, someone tell us why that is wrong for an average skilled player...

[Nicholasp27]

yeah, that's why we later abandoned the whole skill advantage issue and just looked at icm only, so an average player...icm DOES discount the value of added chips as your stack gets bigger...


it says you go from 10 to 10.51 if you get 45 chips to start tourney (5 from each player)...that's 4.4 roi points

now, someone tell us why that is wrong for an average skilled player...

[Nicholasp27]

yes, but don't build skill advantage into the equation anymore...

just go off icm, which is an average player...icm DOES discount additional chips more and more the larger your stack grows...

icm says the average player gets 4.4 roi points from those 45 chips...do u agree or disagree with that?

if everybody agrees with that, then we all agree icm is ok and that it's just my attempt at putting skill advantage into the equation that is off (which, btw, i realize that the skill advantage is NOT linear...but i was assuming that it was pretty close to normal when you add only 45 chips...i realize that if icm says your equity is .49, then you can't then multiply that by 1.295, as that'd give you over .5, which is impossible...but 45 chips seemed low enough to keep them close)



so, is ICM right/wrong that the average player would gain 4.4 roi points if they were given 5 chips from each player to start the game? (assuming everyone at table is average)...that is the actual question that sparked this thread and that should be discussed skill advantage and playing styles is further touched

[Insty]

[ QUOTE ]

so, is ICM right/wrong that the average player would gain 4.4 roi points if they were given 5 chips from each player to start the game? (assuming everyone at table is average)...that is the actual question that sparked this thread and that should be discussed skill advantage and playing styles is further touched


[/ QUOTE ]


Where are you getting this gain of 4.4 roi points from? What do you mean?

I believe that starting with 800 chips your $EV via ICM is $10.00
starting with 845 chips your $EV via ICM is $10.51

I'm not sure I could find 9 opponents to play who would agree to start with 795 chips though.

Inst.

[Nicholasp27]

[ QUOTE ]
[ QUOTE ]

so, is ICM right/wrong that the average player would gain 4.4 roi points if they were given 5 chips from each player to start the game? (assuming everyone at table is average)...that is the actual question that sparked this thread and that should be discussed skill advantage and playing styles is further touched


[/ QUOTE ]


Where are you getting this gain of 4.4 roi points from? What do you mean?

I believe that starting with 800 chips your $EV via ICM is $10.00
starting with 845 chips your $EV via ICM is $10.51

I'm not sure I could find 9 opponents to play who would agree to start with 795 chips though.

Inst.

[/ QUOTE ]

ev of 10 in an $11 buyin, so -1/11=-9% roi
ev of 10.51 in an $11 buyin, so -.49/11=-4% roi

[JJKillian]

[ QUOTE ]


Where are you getting this gain of 4.4 roi points from? What do you mean?

I believe that starting with 800 chips your $EV via ICM is $10.00
starting with 845 chips your $EV via ICM is $10.51

I'm not sure I could find 9 opponents to play who would agree to start with 795 chips though.

Inst.

[/ QUOTE ]

ROFL, I keep thinking the same thing as I keep reading this post. I just figured I missed something along the way and there was a reason everyone had 795.

JJ

[Nicholasp27]

it's just an exercise to examine the effects of 45 more chips...if we can conclude the actual effect of those 45 chips, we can then use it in more practical applications

[Insty]

[ QUOTE ]
ev of 10 in an $11 buyin, so -1/11=-9% roi
ev of 10.51 in an $11 buyin, so -.49/11=-4% roi

[/ QUOTE ]

This is correct.
I believe in ICM.

[JJKillian]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]

so, is ICM right/wrong that the average player would gain 4.4 roi points if they were given 5 chips from each player to start the game? (assuming everyone at table is average)...that is the actual question that sparked this thread and that should be discussed skill advantage and playing styles is further touched


[/ QUOTE ]


Where are you getting this gain of 4.4 roi points from? What do you mean?

I believe that starting with 800 chips your $EV via ICM is $10.00
starting with 845 chips your $EV via ICM is $10.51

I'm not sure I could find 9 opponents to play who would agree to start with 795 chips though.

Inst.

[/ QUOTE ]

ev of 10 in an $11 buyin, so -1/11=-9% roi
ev of 10.51 in an $11 buyin, so -.49/11=-4% roi

[/ QUOTE ]

I am not the math guru as many are on here, but I am wondering something here. Lets say that 4.4% actually does exist (not arguing that it does not). But it seems to me the flaw in this is it is figured with 10 people left. When the next person busts out you gain nothing except more math. Shouldn't this be figured from the bubble. So if by some weird force of the unknown we have 4 people left at this lvl (I guess 15/30 since the 45 chips), then it would be correct right? I am sure I am flawed in my thinking here some how just not sure how.

Also, once again using the 15/30 lvl wouldn't it also matter when the blind is raising next? For example if this happened to be the last hand of the blind lvl, wouldn't you have to figure these numbers at 25/50 since the amt you just gained actually went to the next lvl once you got the chips back in your stack? Or are blind lvls not considered in this model?

More asking if anything else. Once again most of this math is initially behind me until I read it a few times.

JJ

and now since it has been brought up I won't feel like a clown for asking. How did this whole scenerio come up where we are at 15/30 blinds, all players are left, and every one has 5 less chips?

[Nicholasp27]

in another thread, curtains said that he'd seen math showing that 45 extra chips to start tourney would add 4+ roi points to your ev...i didn't believe him, so i did the math

this isn't a practical application, but if we can conclude from this discussion the effect that the 45 chips has on your roi, then we can use that info in more practical ways


icm is strictly stack sizes and payout structure...it doesn't use blinds,position or anything else...sngpt uses blinds, position, cards, calling ranges to come up with inputs for icm and then compares icm's results to say that pushing/folding is +/- ev

[Irieguy]

[ QUOTE ]
yeah, that's why we later abandoned the whole skill advantage issue and just looked at icm only, so an average player...icm DOES discount the value of added chips as your stack gets bigger...


it says you go from 10 to 10.51 if you get 45 chips to start tourney (5 from each player)...that's 4.4 roi points

now, someone tell us why that is wrong for an average skilled player...

[/ QUOTE ]

ICM calculations would be correct if everybody paid the exact same amount in blinds, and if all players saw every hand to the river (or played in exactly the same way... whatever way you want that to be.)

ICM is a mathematical model based on equivalent play and conditions. That model has utility for solving certain problems. But people commonly confuse "equivalent play" with "equal skill."

What ICM doesn't account for is the fact that individual play varies widely from one player to the next, and depends greatly on things like stack size and blind position. So, as the blinds get high and stack sizes become disparate some players will make enormous mistakes and some players will make consistently winning plays. This creates a sizeable rift in tournament EV between good players and bad players. This is where "skill" comes into play and it is precisely what determines one's expectation... or expected ROI. The effect of 45 starting chips on the play of meaningful hands in a SNG is negligible.

This is the flaw in this argument. People are saying that if everybody played the same way, 45 chips would add 4% ROI... so even if everybody is playing differently those 45 chips should still add "about" 4%. The reason why this is incorrect is because a player's ROI is determined almost exclusively from his ability to make important decisions with enormous mathematical repercussions. Some players fold aces preflop, and others will min-raise and then fold to a push when heads-up. These types of errors, and conversely the ability to exploit these errors, constitute a player's expected $ outcome to such a large extent that it dwarfs any theoretical "equal play" based model of expectation.

I think that once you understand that ICM assumes identical play, you can understand how erroneous expected value estimates will be when you apply them to a game where play is never, ever identical from player to player and hand to hand.

So, does an ICM-based calculation show that 45 extra starting chips will add 4%+ to a player's ROI? Yes. Is it therefore accurate to assume that such an addition would add about 4% to a winning player's ROI in the real world? No, of course not. If you understand poker, this will be clear.

Irieguy

[Nicholasp27]

ok, so it sounds like you see a fundamental flaw in icm...it not only assumes equal skill level (ie all 10 players are -9% players), but equivalent play

so if
3 of the donks are lags that have a -9% roi
3 of the donks are weak-tight that have a -9% roi
4 of the donks are loose passive that have a -9% roi

and everyone starts with 800 chips

you think icm is wrong in stating that everyone's ev is .10?

or only that as chip stacks change, icm is incorrect in predicting everyone's ev because different people play different ways, even if they are all -9% in the long-run?

[murfnyc]

[ QUOTE ]
Now, someone tell me where the flaw is in the following double-up first hand strategy....


1600 stack vs 8 800 stacks...

.1844 is your new ev


18.44*1.265=23.3266
23.3266/11=2.1206 or 212% roi!!



[/ QUOTE ]

Maybe I'm missing something, but if you win every tournament so only have a 354.5% ROI (39/11), so it does not make sense to me that you would expect an ROI of 212% if you have only 20% of the chips. I think it is due to the fact the each additional chip gained is worth less than the previous one.

[Irieguy]

[ QUOTE ]
ok, so it sounds like you see a fundamental flaw in icm

[/ QUOTE ]

It's not the ICM that is flawed. It's the use of the ICM to draw certain conclusions that is fundamentally flawed.
[ QUOTE ]

so if
3 of the donks are lags that have a -9% roi
3 of the donks are weak-tight that have a -9% roi
4 of the donks are loose passive that have a -9% roi

and everyone starts with 800 chips

you think icm is wrong in stating that everyone's ev is .10?

[/ QUOTE ]

The ICM is not wrong. It is just evidence.

If a winning player sat down at the above game, he would be able to identify the leaks in his opponents and exploit them. He would beat that game for a significant expectation and giving him 45 extra chips at the start wouldn't mean jack squat.


[ QUOTE ]

or only that as chip stacks change, icm is incorrect in predicting everyone's ev because different people play different ways, even if they are all -9% in the long-run?

[/ QUOTE ]

Yeah, that's a component of the problems that you will encounter when you try to apply ICM-derived expectations to real-world conditions.

Irieguy

[Insty]

[ QUOTE ]
ok, so it sounds like you see a fundamental flaw in icm...it not only assumes equal skill level (ie all 10 players are -9% players), but equivalent play


[/ QUOTE ]

This is not a flaw of ICM - it is a feature.

[ QUOTE ]

so if
3 of the donks are lags that have a -9% roi
3 of the donks are weak-tight that have a -9% roi
4 of the donks are loose passive that have a -9% roi

and everyone starts with 800 chips

you think icm is wrong in stating that everyone's ev is .10?

or only that as chip stacks change, icm is incorrect in predicting everyone's ev because different people play different ways, even if they are all -9% in the long-run?

[/ QUOTE ]

I think it is wrong to use ICM in this manner.

Perhaps you should do some (more) reading of previous ICM threads.

[lorinda]

Deleted.

Plenty of good posts on the subject.

Lori

[Nicholasp27]

we aren't talking about winning players here, just average, -9% roi players


do you think is it accurate to say that everybody has .1 equity to start the game? or is this inaccurate because certain styles do better against some styles so some people at the table should have more than .1 equity even if everyone is -9% roi?

[Degen]

Wow, great discussion.

I see merit on both sides of this and i'm not really mathematically qualified enough to add anything new, but I would like to know how some people think this (http://forumserver.twoplustwo.com/showthreaded.php?Cat=&Number=2610396&page=&view=&s b=5&o=) changes the debate.


Gigabet says there (sort of) that any amount of chips you have over the average (he calls it a line on top of blocks or something) are essentially useless at that point in the tourney yada yada yada

Again I'm not saying this changes anything, just wanted to toss that amazing post in here and see if anybody thinks it changes or adds to this discussion at all.

[ QUOTE ]
I think the extra chips are worth more to the average player than to the +ROI player. (that's based on ICM, player's styles could affect whether or not this is true in a specific case) I'm not sure what happens to the -ROI players.

[/ QUOTE ]

It has to be symmetric, no? Zero(ish) sum and all that, if the +ROI player is getting less from the chips, the -ROI players at that same table are losing less from the lost chips.

[microbet]

I don't see it having to add up. The +ROI players are getting less because there is less to get. They are closer to perfection and you can't get better than that. The -ROI players might get more. On the other hand, if someone could be bad enough to lose all the time, they might still lose all the time and the chips would be worthless to them as well.

It doesn't have to add up because we are talking about different situations, like one game with 1 good player and 9 average players and a separate game with 1 bad player and 9 average players.