SB vs BB ATo, what is BB's push over range? ($16)

[taipan168]

BB's overall HUD stats were 30/23. I hadn't been particularly active but there's a good chance that he thinks I'm stealing from the SB and is defending his blind, but he could also have a hand here.

This is an interesting ICM situation:
- If I fold I have 1120 chips worth 12.55% of the prize pool
- If I call and win I will have 2840 chips worth 25.89% of the prize pool
- If I call and lose I am out.

Therefore I need to be at least 48.5% to win against opp's push over range. Assuming his push over range is 33+, A5+, KJ+ I am 48.7% to win but it's close:

        	equity (%)  	win (%)	tie (%) 
Hand  1:	51.2422 %  	46.56% 	04.68%      { 33+, A5s+, KJs, A5o+, KJo }
Hand  2:	48.7578 %  	44.08% 	04.68%      { ATo }

Two questions:
- What does everybody think opp's push over range is likely to be?
- Given that if my range is right it's only slightly +EV, should I be taking such a narrow chance now or should I fold and hope to outplay these guys elsewhere? (Remembering that ICM assumes that all players are of equal ability).

PokerStars No-Limit Hold'em Tourney, Big Blind is t100 (5 handed) Hand History Converter Tool from FlopTurnRiver.com (Format: FlopTurnRiver)

MP (t1550)
Button (t860)
Hero (t1420)
BB (t1960)
UTG (t7710)

Preflop: Hero is SB with A, T.
3 folds, Hero raises to t300, BB raises to t1960, Hero ????

[Da GOAT]

Its a fold!!!!!

you need opp to be very loose to profit here.

[gingerwizard]

fold.

You've hit the problem with ICM right on the head here. Yes we can do analysis but, being more skillful, you need to put this in to the decision somewhere.

You especially do not need to take coin flips when slight underdog even if it is slightly +ev. Your ability should put an extra weighting to your chips.

I am thinking about amending the ICM to include a chip bias. Basically, based on the level you play at and some assessment of your relative ability it will be quite simple to update the model so that chips you hold are worth more.

I.e. in ICM each chip is like a raffle ticket to the money. So we would put each chip you have as worth x raffle tickets with x>=1 and each chip your opponents have as either 1 raffle ticket or perhaps y<=1 tickets if they are particularly donkish.

It wouldn't take too long to come up with some realistic values of x and y, or to come up with a set of situational x and y. Ie x could be an increasing function of your stack size which means the bigger your stack, the more value your chips have (sensible for good players.)

If anyone is interested in taking this further. Let me know.

[Da GOAT]

fold.

You've hit the problem with ICM right on the head here. Yes we can do analysis but, being more skillful, you need to put this in to the decision somewhere.

You especially do not need to take coin flips when slight underdog even if it is slightly +ev. Your ability should put an extra weighting to your chips.

I am thinking about amending the ICM to include a chip bias. Basically, based on the level you play at and some assessment of your relative ability it will be quite simple to update the model so that chips you hold are worth more.

I.e. in ICM each chip is like a raffle ticket to the money. So we would put each chip you have as worth x raffle tickets with x>=1 and each chip your opponents have as either 1 raffle ticket or perhaps y<=1 tickets if they are particularly donkish.

It wouldn't take too long to come up with some realistic values of x and y, or to come up with a set of situational x and y. Ie x could be an increasing function of your stack size which means the bigger your stack, the more value your chips have (sensible for good players.)

If anyone is interested in taking this further. Let me know.

interesting idea but im not one who knows the inside/out of formulas

[gingerwizard]

I am though!! (Master of mathematics and writing a PHD in statistics.)

This is right up my street and being bored as hell in my office at the mo, it may be the distraction ive been looking for.

With the aid of the insight from a couple of good players i'm sure it wouldn't take me too long to put something together.

The real work will be in tweaking it so that it really represents the skill differences between players in an optimal fashion.

The result will be analyses that mean we are no longer making slightly +EV calls against bad players, we would be making vastly +EV calls. This will certainly reduce variance and ensure that you, as better players, are making the most of your skill advantage.

Things to capture in the function:

Higher Buyin = better opps = lesser advantage. If Goat was to play $5.50 his big stack is worth more than if he is playing $55+5.

Laggy player = lower chip values. As he leaks much more of them away playing OOP and calling with less/nothing.

Taggy player = higher chip values. Once he has them he keeps them.

Winning player = higher chip values. Well you can't assume its all luck.

tight/passive = lower values. He isn't making chips, the blinds are swallowing his stack. When he does catch a hand, most people will avoid him so he has less opportunity to double up than everyone else.

Strong heads up = higher values. If heads up you are more likely to take down first so we must account for this.


Im sure there are other things, and I would like to know what the good players think are the important features for them. Where are your biggest edges?

[Da GOAT]

With the aid of the insight from a couple of good players i'm sure it wouldn't take me too long to put something together.

just tell me what you think you need. my play is more instinct than maths based tho. i dont know exact odds or use M.

The result will be analyses that mean we are no longer making slightly +EV calls against bad players, we would be making vastly +EV calls. This will certainly reduce variance and ensure that you, as better players, are making the most of your skill advantage.

But also slightly +EV becomes -EV. wonder how much variance you think you can kurb at turbo's. not much since i shove so much crap anyway coz i need chips. most good player advantage is spotting push spots which fish dont so i dont know how my stack of 8xBB is worth more,especially if i know opp range since +EV is +EV and vice versa.

Things to capture in the function:

Higher Buyin = better opps = lesser advantage. If Goat was to play $5.50 his big stack is worth more than if he is playing $55+5.

Sweet

Laggy player = lower chip values. As he leaks much more of them away playing OOP and calling with less/nothing.

Taggy player = higher chip values. Once he has them he keeps them.

Winning player = higher chip values. Well you can't assume its all luck.

tight/passive = lower values. He isn't making chips, the blinds are swallowing his stack. When he does catch a hand, most people will avoid him so he has less opportunity to double up than everyone else.

Strong heads up = higher values. If heads up you are more likely to take down first so we must account for this.


Major blind dependant tho ive adjusted my own game recently at high blind levels to gain a bigger edge (i feel) than taking +EV pushes. +EV pushes arent nessarily the MOST EV you can get.

Im sure there are other things, and I would like to know what the good players think are the important features for them. Where are your biggest edges?

If your asking here, its a long list. and hard to quantify. any specific?

[gingerwizard]

With the aid of the insight from a couple of good players i'm sure it wouldn't take me too long to put something together.

just tell me what you think you need. my play is more instinct than maths based tho. i dont know exact odds or use M.

The result will be analyses that mean we are no longer making slightly +EV calls against bad players, we would be making vastly +EV calls. This will certainly reduce variance and ensure that you, as better players, are making the most of your skill advantage.

But also slightly +EV becomes -EV. wonder how much variance you think you can kurb at turbo's. not much since i shove so much crap anyway coz i need chips. most good player advantage is spotting push spots which fish dont so i dont know how my stack of 8xBB is worth more,especially if i know opp range since +EV is +EV and vice versa.
Yes but EV is expected value in a FAIR game when everyone has equal chance of winning. You have more chance so you need to adjust EV. Therefore slightly +EV becomeing slightly -EV is exactly what we want. Thus pointing to the fact that making borderline calls negates our skill advantge

In turbos most of your edge is located in finding good push spots, therefore Turbo/standard should have an effect on our function. Further finding good push spots is not your ONLY advantage. You are not a donk. You don't minbet. You are not likely to push or call with total crap early. You don't think QTo is a good hand. You understand c-betting, c-raising. You know when to lay a hand down. etc etc etc. If you know opps range, why should slightly +EV be a reason to push and potentially bubble out?

Creating an adjustment to ICM will mean that slightly +EV moves mean you are taking your skill into account and will reduce variance since in a skill-less game your pushes would be +++EV. Meaning you will be called and beaten less often.

Things to capture in the function:

Higher Buyin = better opps = lesser advantage. If Goat was to play $5.50 his big stack is worth more than if he is playing $55+5.

Sweet

Laggy player = lower chip values. As he leaks much more of them away playing OOP and calling with less/nothing.

Taggy player = higher chip values. Once he has them he keeps them.

Winning player = higher chip values. Well you can't assume its all luck.

tight/passive = lower values. He isn't making chips, the blinds are swallowing his stack. When he does catch a hand, most people will avoid him so he has less opportunity to double up than everyone else.

Strong heads up = higher values. If heads up you are more likely to take down first so we must account for this.


Major blind dependant tho ive adjusted my own game recently at high blind levels to gain a bigger edge (i feel) than taking +EV pushes. +EV pushes arent nessarily the MOST EV you can get.
exactly how to think of all of this. +EV isn't necessairily the most you can get. Your experience and skill counts for something.
Im sure there are other things, and I would like to know what the good players think are the important features for them. Where are your biggest edges?

If your asking here, its a long list. and hard to quantify. any specific?

[Da GOAT]

What do you reckon of SNGPTs minimium edge factor????

also +EV push spot on bubble is just that , i know opp range. i fail to understand exactly how you can give me an added adv. in these exact situations where i know its +EV/-EV anyway.

[gingerwizard]

Don't have SNGPT. When my BR supports it im there though.

Ok so you work out that a push is +$Ev like taipan has at the start of his post.

However it is only JUST +EV. so barely better than a break even bet. The correct play is to push because its +EV. But still you crash out more than 50% of the time, and over time you will make money this way but you are not taking into account everthing else.

Taipan is a better player than this guy yes? If we took that into account in someway, this push would quite rightly be seen as -EV. I.e. you can find a better spot than this.

A good way to think of it is this: Suppose you are a winning player. Your skill and the time you've spent acquiring it mean that you expect to make $x per week. So if you play 20 SNGs per week thats $x/20 per sng. So a move that has EV<$x/20 is negative FOR YOU. It may be great for a losing player or a break even player but YOU have just LOST money. You EXPECTED to make more.

See?

[Da GOAT]

Don't have SNGPT. When my BR supports it im there though.

Ok so you work out that a push is +$Ev like taipan has at the start of his post.

However it is only JUST +EV. so barely better than a break even bet. The correct play is to push because its +EV. But still you crash out more than 50% of the time, and over time you will make money this way but you are not taking into account everthing else.

Taipan is a better player than this guy yes? If we took that into account in someway, this push would quite rightly be seen as -EV. I.e. you can find a better spot than this.

A good way to think of it is this: Suppose you are a winning player. Your skill and the time you've spent acquiring it mean that you expect to make $x per week. So if you play 20 SNGs per week thats $x/20 per sng. So a move that has EV<$x/20 is negative FOR YOU. It may be great for a losing player or a break even player but YOU have just LOST money. You EXPECTED to make more.

See?

Oh i get you, but i know its an easy fold here

[gingerwizard]

Ah but why???

Surely its +EV? By your argument +EV is +EV right?

You know its a fold but for an intangible reason. I think it may be possible to quantify it so that (by some revised version of ICM), we know +EV means push or call!!

[taipan168]

I think there may be a relatively easy proxy for the donkishness of a player by measuring the degree (number of standard deviations perhaps?) by which their VP$IP diverges from, an optimal VP$IP of, say, 20%. A player whose VP$IP is 0% is likely too tight, one with 40% is likely too loose (and one with 80% is a complete and utter donk).

I know that VP$IP changes as the blinds increase but we are looking for a quick and easy measure of likely skill here as a rough adjustment to ICM as a lot of the factors that we should take into account (skill HU, not minbetting, not missing easy pushes) aren't really knowable but VP$IP is the one stat that is very easily tracked.

Thoughts Ginger?

[Staresy]

Whilst I generally agree with what you're saying here Taipan, you would have to factor in how many hands that VP$IP is based upon.

It is entirely feasible that someone picks up a surge of good hands, but has to let some of them go post-flop because of the donktastic nature of micro-stakes SnGs.

[taipan168]

Whilst I generally agree with what you're saying here Taipan, you would have to factor in how many hands that VP$IP is based upon.

It is entirely feasible that someone picks up a surge of good hands, but has to let some of them go post-flop because of the donktastic nature of micro-stakes SnGs.

Agreed, but this will only ever be a rough measure. It's just that I can't think of any other easy and relatively objective way to adjust for the skill differential between players. Plus by the time that it gets to push/fold you should have a big enough sample size to give you a reasonable VP$IP estimate.

[gingerwizard]

Well.

I don't know how VP$IP works and i don't have a HUD thingy (is it a tracker or a display??). I understand its some measure of aggression based on number of hands played and from what position.

TBH I think we can do better than that by looking at cards they've shown down and remembering their actions pre and post flop. (Pre especially.)

First you would rate yourself.
Then you ask a few questions of your opponents: e.g. does he minbet? what have i seen him raise with pre flop and in what position? does he steal? has he pushed and sucked out? Has he limped a lot. What are is HUD stats? and (say) use that info to rate him from 1 to 10.

(Not that 1:10 is the best way, im simply suggesting using as much info as you have as possible and a little judgement.)

There will be a default rating for no-reads and the average player.

As soon as we figure a way to quantify a rating system, it will be pretty simple to adjust the ICM calculations for skill.

I suggest we start by getting something down and improving it as we go. Perhaps A four tier system for Now:
1. FTR poster
2. Solid steady, no real visable mistakes
3. Loose/ aggressive
4. Donk.

(Not that LAGGY play cannot be optimal but i think its bad in your standard SNG.)

See how this works and add to it from there.

Ill come up with version 1.0 of the DEPENDENT chip model at some point during the weekend.

GW

[taipan168]

I don't know how VP$IP works and i don't have a HUD thingy (is it a tracker or a display??). I understand its some measure of aggression based on number of hands played and from what position.

VP$IP = % of times a player voluntarily puts chips in the pot preflop. It's a PT stat, and probably the most important one. Measures the % of total times you call or raise preflop excluding checking the BB (whether or not you see the flop).

As I said, this obviously needs to change as the blinds increase (my VP$IP for the first 2 levels is an average of like 11-13% whilst at 100/200 and above it is more like 35-40%) but you can be pretty much certain that if a player has a VP$IP of 80% they are a total donk. Plus the further they are away from the "optimal" number the more of a donk they are - eg. a player with 80% VP$IP is much more of a donk than one with 40%.

I think this is a good, objective, quick overall measure as to the donkishness or otherwise of a player. I agree it doesn't take into account factors like "the guy minbet 30 into a 600 pot" and "the guy minraised with QJo after 5 limpers" but generally I think there will be a reasonably strong correlation between VP$IP and these other factors.

IMO the rating system is a bit too subjective for my taste and will be difficult to use for multitablers.

[gingerwizard]

Statistics and indeed poker is ALWAYS subjective. I think using VP$IP could certainly be one way very useful for multitablers. Single tablers should be able to get better reads and therefore do better.

So for now I could set up a default rating system based on VP$IP, but build in the ability to use reads.

ATM my idea is to start from some discrete function using (say) just the four "ratings" i gave above. (How you get to those ratings will depend on you but could use VP$IP). See how this works then gradually build to a continuous function, perhaps even of VP$IP itself!

[Da GOAT]

First you would rate yourself.

Can the formula account for EGO?

[gingerwizard]

Well your beliefs will account for ego. The formula will be an updated version of ICM taking into account your beliefs.