PDA

View Full Version : SNG Luck Factor question



Rvg72
02-13-2009, 03:35 AM
As some of you know we are about to release this functionality and the last couple of betas are calculating $ expectation for all SNG all-in situations (although currently not saving it do DB, just to a log file for testing) but I do have some questions on how you think the OVERALL tourney luck should be calculated

Scenario 1: Player plays a few all-in situations and overall is a little bit lucky but then busts out in 4th place so $0 winnings. Should his luck adjusted winnings be set at $0 to avoid a situation where it looks like he lost more than the buyin

Scenario 2: Player is very unlucky in all-ins but still wins the tourney. Say first place is $100 and if you total up all the ICM differences in all-in hands he was unlucky by $12. Do you really want the adjusted winnings to be $112 or should this be capped to the win amount?

The issue with adjusting the values is that the $EV adjusted winnings for all players won't add up properly but I'm thinking it needs to be done this way.

Normally the values will fall in normal ranges although I'm sure this will come up pretty frequently. Any feedback on how this should be handled? I'm specifically talking about $EV adjusted winnings for entire tourneys, so in the winnings graph for example.

Roy

vlsup
02-13-2009, 03:43 AM
I think the turney Luck Factor need to appear in two reincarnations:
- in chips (i.e. "pure luck" in allins)
- and in win/lost money (i.e. "final effect" of allins)

Rvg72
02-13-2009, 04:14 AM
I think the turney Luck Factor need to appear in two reincarnations:
- in chips (i.e. "pure luck" in allins)
- and in win/lost money (i.e. "final effect" of allins)

Yes, both are there but the question is what if the final effect of allins gives you an adjusted $ amount that is not possible?

vlsup
02-13-2009, 04:42 AM
make a build-in template of letter to poker-room support with request of more money due to their bad random generator :)

GrumpyDadPoker
02-13-2009, 06:52 AM
I suggest to cap it in both ends, I won't be able to understand a SNG or a MTT where I lost more than the buy-in or won more than the 1st place payout.

my 2 cents.

dave_w11
02-13-2009, 11:24 AM
I think that it should not be capped as this will cause long term skewing of results for 'luck' one way or another depending on which causes most capping. If you do this please at least add an option for it to be done without capping.

A long term tendency to report players more 'unlucky' than they are will be hard to detect without a lot of data so I think it is important not to bring about potential for such bias.

Rvg72
02-13-2009, 01:23 PM
Yeah, I think I'll need to do uncapped by default and provide an option to use capped otherwise like you said it could create longer term issues.

Iggy23
02-13-2009, 03:34 PM
2 questions...

1) Does the luck factor include situations where you are against another player that is all in or just situations where you are the one who is all in?

2) What about a situation where you are down to 4 players, you fold but another player goes all in? Wouldn't the luck factor of that hand have an effect on your winnings? Will that be included in your winnings graph?

chilin_dude
02-13-2009, 04:11 PM
Uncapped definitely - That is how jukbaks works and is the best way imo

Rvg72
02-13-2009, 06:39 PM
2 questions...

1) Does the luck factor include situations where you are against another player that is all in or just situations where you are the one who is all in?

2) What about a situation where you are down to 4 players, you fold but another player goes all in? Wouldn't the luck factor of that hand have an effect on your winnings? Will that be included in your winnings graph?

1) It includes any situation where there is no bias and you have no actions left but there are cards to come

2) Yes - generally when you are not involved in an all-in your equity change is very close to $0 but in spots like this it can be very large despite you not being in the hand. These will be included although this situation might not be in the beta 8

Roy

LirvA
02-14-2009, 02:35 AM
You might want to start a thread about this at 2+2 or something, lots of people over there know a ton about this sort of stuff. David Sklansky might even be able to offer some good input, he's a hell of a numbers guy.

You're talking about bubble play right? If for instance, you're not abusing the bubble, .. when you fold and let other people play against each other, your EV increases by just folding and letting them play pots against each other. The whole thing like ... If everyone folds to the blinds who get all in against each other with coin flipping hands, their EV is essentially +/- 0, but everyone elses EV increases because one of them has a chance to bust out.

.... of course I'm not the most knowledgable person in this area, ... I mean I understand the differences between chip EV and tournament EV, and all that, but I'm not an expert who's studied this shit for a long time and all that. I've never screwed around with sng wizard or sng powertools or whatever .... and I tend to ramble and follow ,s with ...s, so you might want to start a thread on 2+2 about this.

guitarizt
02-14-2009, 06:21 AM
Yeah I have no idea ask on 2+2 math forum or something imo. The second example seems ok to me but I don't really know anything. It'd prob be time consuming to try it both ways on a sample test of hhs right?

dave_w11
02-14-2009, 01:32 PM
I did some messing about with excel using the results from juks all in testing program. The graph shows about 11k $16s using capped and uncapped luck.

I think it would be better not to even ever draw a graph using capped results as it would likely result in a lot of confusion.

Could you have both capped an uncapped calculated and stored for viewing as individual results but only have an option to plot the uncapped results?


http://img23.imageshack.us/img23/5701/cappedluckvr1.jpg (http://imageshack.us)

Rvg72
02-16-2009, 07:11 PM
We're going to go uncapped initially since that is most accurate

Beta 8 with this functionality should be out later tonight

Roy

l1ttledb
05-20-2009, 07:19 PM
So can anybody point me in the direction of an explanation of this luck adjusted winnings?

Haven't a clue what it is

morny
05-20-2009, 07:54 PM
ICM stands for individual chip model. It is a way of estimating your $ expectation in a tournament, given the chip counts and the prize structure. The button calculates what your expectation was before the hand and what it was after.

If somebody was all-in before the river, it will calculate the probability of each hand result (e.g. player 1 beats player 2, player beats player 1, player 1ties with player 2). Then it calculates how many chips each player has after each result and from that the ICM-based expectation for the players for the result. It then weights the resulting expectations by the probability of each result to come up with an overall $ expectation.

The difference between your expected $ value over all the hand results and the actual $ value of the chips is luck...

l1ttledb
05-20-2009, 08:58 PM
..

so if the red line is below the green your unlucky.. and if its above your lucky ( over that sample of course )? or is it the other wat round?

morny
05-21-2009, 11:36 AM
Red line is your winnings adjusted to account for luck, so if for example you won $2k and with average luck you should have won $2.5k then the red line will be up at 2.5k and the green your actual winnings at 2k so when the red is above the green line it shows your running bad and if the red line is below then it shows your running good

l1ttledb
09-25-2009, 11:43 PM
hi again, i was just wonderin if this makes sense. Over the last 100 or so games i would say im running terrible. For example i have had at least 12 in the last 40 all in with overpair shoves where they have hit sets, i have also had alot of miraculous beats all in eg KK v Kjo came T63-A-Q over 90 percent on flop.. The thing is i would expect my red line to go up but instead is steadily falling.. Does this sound right?

fozzy71
09-26-2009, 10:51 AM
Can you please post the graph or email it if you prefer it not be public?

Please email it to me at fozzy@holdemmanager.net, with a link to this thread./

l1ttledb
09-26-2009, 06:56 PM
emailed it to ya

fozzy71
09-26-2009, 07:18 PM
What version of HM are you using currently? What site is this for? What type of tourneys?


*Try creating a new DB - http://www.holdemmanager.net/faq/afmviewfaq.aspx?faqid=146
*Now import a small portion of your \HMArchive so you can see if the problem exists in the new DB - http://www.holdemmanager.net/faq/afmviewfaq.aspx?faqid=173
*If the new DB seems to work properly, you will want to import the rest of your archives to the new DB, and export/import the hands from the old DB, and then delete the old DB. - http://www.holdemmanager.net/faq/afmviewfaq.aspx?faqid=21

l1ttledb
09-26-2009, 11:36 PM
all fulltilt, 1.09 beta 34, all the games in the swing im talkin about are 9man $10+1 NLH.. I even think the red line might be off versus the upswing as although i ran better than normal my game review + wizzard told me a different story ( ie only ran marginally better so the red line should be upward also)

Thanks again

3steptrap
11-07-2009, 04:29 AM
OK so I have a question about luck adjusted winnings.

Below is a graph of my tourneys since August 2009 in Holdem Manager:
http://s3.postimage.org/onYnS-bd24ba29a4d44837f233f736b2eee044.jpg

This is a graph of the same games using "Tourney Luck" from the Poker Tracker 3 site that plots changes in chip EV:
http://s3.postimage.org/oo2n9-bd24ba29a4d44837f233f736b2eee044.jpg

According to the first graph I have run bad and played bad. According to the second graph I ran bad but played good. I understand that chip EV is different than $EV but how can my "Expected" line be going up in one and down in the other?

Anyone who would like to comment or discuss (not just mods) please feel free to do so.

fozzy71
11-07-2009, 10:37 AM
Before I ask the developer to reply can you tell me what sites you are playing on, what types of tourneys primarily, and what version of HM you are using? If it is an older version we may want to update and test in a new DB to see if it changes the EV at all.

Rvg72
11-07-2009, 11:17 AM
OK so I have a question about luck adjusted winnings.

Below is a graph of my tourneys since August 2009 in Holdem Manager:
http://s3.postimage.org/onYnS-bd24ba29a4d44837f233f736b2eee044.jpg

This is a graph of the same games using "Tourney Luck" from the Poker Tracker 3 site that plots changes in chip EV:
http://s3.postimage.org/oo2n9-bd24ba29a4d44837f233f736b2eee044.jpg

According to the first graph I have run bad and played bad. According to the second graph I ran bad but played good. I understand that chip EV is different than $EV but how can my "Expected" line be going up in one and down in the other?

Anyone who would like to comment or discuss (not just mods) please feel free to do so.

Hi, Poker Tracker EV badly miscalculates pots where there are more than 2 players (at least they did recently which is the last time I checked). Because of this their EV calculations are all completely wrong especially with tourneys since there are much more 3+ way all-ins.

Roy

3steptrap
11-07-2009, 01:59 PM
Before I ask the developer to reply can you tell me what sites you are playing on, what types of tourneys primarily, and what version of HM you are using? If it is an older version we may want to update and test in a new DB to see if it changes the EV at all.


Sites: Full Tilt
Types of tourneys: 6 max Speed/Turbos
Version: Trial 1.09 Beta41