Application: HoldemLuck/TourneyLuck v0.20 (Updated:2/28/10)

[uconnrounder]

You're right that cEV is more correlated to $ev in STTs than in MTTs, but I think you are overestimating how relevant it is for STTs. If cEV and $EV were at all closely correlated, there would be minimal differences between pot odds and ICM decisions...and as we all know, you're probably going to get crushed if you operate strictly using pot odds decisions in STTs. There's a huge huge difference between losing AK to AQ in the first hand and losing AK to AQ on the bubble as a short stack against another short stack, even if you both have the same # of chips as you started the tournament with. I've played over 30,000 super turbos this year and there have been sooooo many days where I ran above $EV while running below cEV and vice versa. For instance, there was a period of time where I was running 40,000 chips under equity (that's a lot considering you start with only 300 chips in a super turbo) but running 20 buy-ins above $EV over that same sample.

The other really important thing about ICM calculations above chip calculations is that your ICM changes even in hands that you fold. So let's take a bubble situation for example. You're the short stack and you fold while another short stack goes all in with 55 and gets called by the chip leader's JJ, and the 55 sucks out. From a cEV standpoint, nothing happened to you. However, from an ICM standpoint you just got extremely unlucky and your chances of making it ITM have been severely reduced.

I think it would be a huge mistake to put any real stock in cEV as a means of evaluating how good/bad you are running in STTs.

[not_just_ice]

Also, can you explain what hands your program ignores and what hands it calculates EV for?
[list]

It finds all hands where Hero was all in before the river against 1 or more villains.

...

Yeah, like I said, only hands that Hero got all in before the river are calculated so this hand is ignored.

Just stumbled upon this program few days ago and can't help wondering...
So this is probably more of a philosophical question but:

Why wouldn't you consider also cases where all your opponents are all-in before river (even though you have chips left)? [Your "luck" on those hands could be calculated similarly.]
For example if you were sitting on a table with biggest stack then you would have no way to find out how you're doing luck-wise on all-in situations.
Also consider case where you'd only have little bigger stack than your opponent, anyone with a bright mind would take it as a all-in situation and then consider the "luck" on the outcome, except of course someone who'd limit to cases where you actually are all in [like 1bb would be different from 0bb with 200bb stacks].

So could you add this as an option to your program so that one could get luck on those situations if they wanted and still the purists could refuse to use it.

[kraada]

I'm pretty sure that if you are in the pot with only one other person and he is all in and you have him covered, that is still calculated, though I have not done explicit testing to check.

[1p0kerboy]

Wait, so this thing doesn't use the Independent Chip Model (ICM) for tournaments?

Whoever said cEV is closer to $EV in STTs than MTTs couldn't be more wrong. The payout structures in SNGs create a higher bubble factor throughout than in MTTs. For more on this please read Kill Everyone by Lee Nelson.

[kraada]

There was a function added to show EV using ICM, but it wasn't working right and I'm not sure it's even still included to be honest.

[Clockwork Banana]

There was a function added to show EV using ICM, but it wasn't working right and I'm not sure it's even still included to be honest.

Yes some ICM calculations were included (the orange line) but Alon Albert took them out because they weren't working properly.

Also when your opponent is all in and you have him covered, the pot (obviously) counts even if you're not all in. All that is needed is that 1 player is all in and as further players become all in, they are part of a different pot.

Wait, so this thing doesn't use the Independent Chip Model (ICM) for tournaments?

Whoever said cEV is closer to $EV in STTs than MTTs couldn't be more wrong. The payout structures in SNGs create a higher bubble factor throughout than in MTTs. For more on this please read Kill Everyone by Lee Nelson.

That whoever was me and since I recognise your screen name, I'm pretty sure you are more knowledgeable about poker theory than I am since I'm a lazy nitty noob donkey with a chip and a chair (half a brain too but the wrong half!). I will read "Kill Everyone" since you seem to recommend it.

Your arguments sound quite intuitive too.

BUT

1- How off will you readings be?

There may be a higher ITM bubble factor in SNGs than MTTs but there's no denying that there's a higher correlation between the amount of chips won and the amount of money won in an SNG than in an MTT. In an SNG you can triple up and have a third of the chips in play. This would never happens in an MTT. So cEV calculations may be imperfect (it does not accurately show situational circumstances such as bubble factors) but it can act as a useful guide. It simply can't be in MTTs, you don't even get a rough idea.

If you could magically double up in mid SNG, I can guarantee you would see the results show in terms of dollars. Sure you'll get those bubble situations but on the whole you'll do well. This is NEVER the case in MTTs.

Bubble situations in SNGs may be important but how much divergence will you find between ICM and cEV given the fact that you are likely to experience as many positive (lucky) situations as negative (unlucky) situations. This occurs because of the number of bubbles you are likely to see as a regular SNG player. This just can't be said of MTTs. You can get screwed over all your life in crucial MTT situations but I've never seen a player who's bubbled all their SNGs because of bad bubble luck.


2- About the importance of an ICM type of reading in MTTs compared to SNGs:

Like I said above bubble factors are likely to go in your favour and against you so in the long run they are likely to equal themselves out. I do understand that the point of having statistical tools is that they can tell you what happens with a higher degree of precision than that and can even be used to see shorter term trends too but cEV is likely to get things somewhere close because 1- it doesn't take long for things to equal out over time (the number of bubbles experienced in higher) and 2- the payout in SNGs is never a huge multiple of your buy-in whereas in MTTs they are (there are 2 bubbles in MTTs), so any divergence is not likely to be a massive one.

I have to tell you some bad beat stories to make my point so bear with me a second. How do you think an elaborate ICM reading would have behaved when I lost KK against TT in the biggest final table I ever made in a pot that would have made me the tournament chip leader? How about that bogey $11 tournament (I will wear black for the rest of my life on the 8th of June) where I lost 3 "all in" hands dominating my opponent at the final 2 tables and where in each instance I would have been (or comforted my place as) the tournament chipleader? I stood to make 100, 200 or maybe 400 times my buy-in even, depending on where I had finished should I have won (In the end I finished 10th). Since I refuse to think I'm the unluckiest player on Earth (maybe just to reassure myself and put the gun down), this must happen to quite a few people. I sometimes watch replays of the Sunday Million and I see some very very very expensive beats!! Much more expensive than the biggest string of bubble beats you'll ever see in SNGs.

Are you to tell me that because of higher bubble factors in SNGs (as I understand, this is the case because you are always closer to the bubble in an SNG that you are in an MTT at any given point), those would be more significant than those crucial beats at the final table or final 2 tables of an MTT? Those payouts are MASSIVE relative to the buy-in whereas in SNGs, you're really trying to double or triple your money.

I don't know I've got to think about it further but it seems to me that in MTTs there are some key moments that change everything. In SNGs those key moments are much less significant and get lost in the multitude of tournaments a regular SNG player engages in.


3- How useful is your luck?

This isn't specifically related to this discussion but related to luck in tournaments. It occurs to me that if you have a big stack in an SNG, it's much easier to win than if you have a big stack in an MTT, perhaps partly because the payout structure is much simpler.

If you are the chipleader at your SNG table you will remain so until you lose chips or someone wins some. In MTTs you can lose your chipleader status when you are moved to another table. This can affect your ability to boss the table. Also you can instantly get into a situation where people can have you covered without any chips changing hands (hence putting your tournament life at risk). This wouldn't happen in an SNG.

Also closing stages of an MTT seem a lot trickier than closing stages of an SNG where moves rather appear obvious when you have a big stack. That is to say, give a donkey a commanding chiplead in an SNG and he probably won't fail. Give a big stack to the same donkey in an MTT and he's no closer to the win as he will have to find creative plays, not simply ICM dictated ones.

[Tract0rDan]

Apologies if this has already been mentionned, but this thread is huge and unsearchable!
The other day I had a guy covered and put him all-in, but without pressing 'all-in' myself. For example, I have $25 left, he has $10. I use the scroller and end up betting $13 and putting him all in. He calls and is all in, but this doesn't appear in my equity graph - is it because bet an amount rather than pressed 'all-in'?

[kraada]

I thought those spots were counted - alon?

[clarkrussell]

I can't v. 15 to work. It says "application failed to initialize properly". Any ideas?

[PJ of TheGame]

The taller the payout structure, the closer cEV is to $EV. In a STT you can get half as much as first by cruising into the money, while cruising into the money in a big MTT will get you like 1/100th of what first gets. This would make it much more correct to take risks to gain chips (closer to cash game strat) in a big MTT then in a STT.

However, it doesn't really matter for a simple "am I running lucky" graph. You had a certain amount of equity before the confrontation, and a certain amount afterwards- and it would be pretty simple to do the math on it. The exact correlation matters for making decisions while playing, but won't really affect a graph. This will NOT tell you anything useful though, since everyone will run extremely unlucky on that graph... since the chips you win when you get lucky aren't as valuable as the chips you lose when you get unlucky (basic ICM stuff).