awesome program kutos ^^

I hope you made a mistake and over-adjusted it, because otherwise im running uber-lucky, lol.

Same here, running +3639% Luck for +1837.69 bucks over 900+ stars tournies. I thought I was a winning player, maybe not. Anyone else getting some unexpected results?

Nice program. I think there's some problem in how you treat your ICM calculations since the results still only seem correct for HU hands.

This sounds like you are just distributing the chips in the pot by hand equity then plugging that into an ICM calculator, that wont work you need to treat the outcomes separately. Here's a hand I went through to compare with the output:

Poker Stars $15 $1 No Limit Hold'em Tournament - t200/t400 Blinds - t25 Antes - 4 players
The Official 2+2 Hand Converter Powered By DeucesCracked.com

CO: t3190
BTN: t3025
SB: t3715
Hero (BB): t3570

Pre Flop: Hero is BB with A A
1 fold, BTN raises to t1600, 1 fold, Hero raises to t3545 all in, BTN calls t1400 all in

Flop: (t6300) 7 5 9 (2 players - 2 are all in)

Turn: (t6300) 9 (2 players - 2 are all in)

River: (t6300) 4 (2 players - 2 are all in)

Final Pot: t6300
BTN shows Js Jc (two pair, Jacks and Nines)
Hero shows Ad Ah (a flush, Ace high)
Hero wins t6300

Plug stacks for the two outcomes (win/lose) into ICM calc to get Heros equity:

Stacks - Win
CO: t3165
BTN: t0
SB: t3490
Hero (BB): t6845 EV_win=38.53%

Stacks - Lose
CO: t3165
BTN: t6300
SB: t3490
Hero (BB): t545 EV_lose=6.21%

For AA vs JJ, win% = 81.15%, lose% = 18.85% (ignoring ties)

Expected Equity
= win%*EV_win + lose%*EV_lose
= 81.15%*38.53% + 18.85%*6.21%
= 32.44%

Outcome Equity = EV_win (since I won)
= 38.53%

Luck = Outcome Equity - Expected Equity
= 38.53% - 32.44%
= 6.09%

Compare this to the output which gives Luck = 7.3%:

Ah, there might be hope for me.

You're absolutely right in what you are saying and your math is correct.

First I thought SnGLA interprets this SnG with a 6-max payout-structure (60% |40% | 0%). This would give you 7.31% Luck. On the other hand in this case your equity gain should be 17.5% instead of 12.5%

So, I have no idea what's going wrong there.

Do you think you could post or pm me the original HH of this hand? Then I could debug it step by step.

That would be great, thanks!

Here's the original hh.

PokerStars Game #14682390222: Tournament #74438820, $15+$1 Hold'em No Limit - Level VIII (200/400) - 2008/01/19 - 21:52:45 (ET)
Table '74438820 1' 9-max Seat #6 is the button
Seat 1: amg55cls (3190 in chips)
Seat 6: papa lloyde (3025 in chips)
Seat 8: scottwire (3715 in chips)
Seat 9: IFoldPktOnes (3570 in chips)
amg55cls: posts the ante 25
papa lloyde: posts the ante 25
scottwire: posts the ante 25
IFoldPktOnes: posts the ante 25
scottwire: posts small blind 200
IFoldPktOnes: posts big blind 400
*** HOLE CARDS ***
Dealt to IFoldPktOnes [Ad Ah]
amg55cls: folds
papa lloyde: raises 1200 to 1600
scottwire: folds
IFoldPktOnes: raises 1945 to 3545 and is all-in
papa lloyde: calls 1400 and is all-in
*** FLOP *** [7d 5d 9d]
*** TURN *** [7d 5d 9d] [9s]
*** RIVER *** [7d 5d 9d 9s] [4d]
*** SHOW DOWN ***
IFoldPktOnes: shows [Ad Ah] (a flush, Ace high)
papa lloyde: shows [Js Jc] (two pair, Jacks and Nines)
IFoldPktOnes collected 6300 from pot
*** SUMMARY ***
Total pot 6300 | Rake 0
Board [7d 5d 9d 9s 4d]
Seat 1: amg55cls folded before Flop (didn't bet)
Seat 6: papa lloyde (button) showed [Js Jc] and lost with two pair, Jacks and Nines
Seat 8: scottwire (small blind) folded before Flop
Seat 9: IFoldPktOnes (big blind) showed [Ad Ah] and won (6300) with a flush, Ace high

Okay, I see the problem. It's a Pokerstars issue.

You bet 545 chips more than papa lloyde can call.

Usually, at least at FullTilt, there should be an extra line saying:

Uncalled bet of 545 returned to IFoldPktOnes

However, even Pokertracker has now a pre-flop pot of 6845 instead of 6300 in its database and, more importantly, has IFoldPktOnes all-in pre-flop, which reduces your EV_loose to 0%. Pokertracker corrects this in a later step, kind of a summary-line which I didn't consider yet, since this wasnt an issue for FullTilt games.



I fixed it and will release an update soon.

New version / patch available:

SnG Luck Analyzer v0.1.5

Users of v0.1.4 can use the automatic online update feature.

When I connect, I have an error "you don't have sng tournaments".
It's maybe due to the fact that I have no tournament summaries ? How can I get them ?

Yes, the problem is, that Pokertracker doesn't recognize your SnG's as SnG's if there is no tournament summary. They are safed as MTT's in your database, without any buy-in. This is different if you are playing at FullTilt or Party though. Don't ask me why...

You can either import tournament summary emails from your poker-site into pokertracker or update them manually. See Pokertracker help for further info.

The outcome was now very close to what I would expect = running just a little hot.

So I think Pokerstars is now working.

Awesome-o.

Pokerstars will only send you 60 summaries in 1 hour, so its a bit of a hassle...

Thanks for the update. I really hope we are getting closer to a fully working program!

I have two new features in mind. Which one do you guys think should have higher priority?

1. Calculation of ROI and luck-compensated ROI
2. PostgreSQL support

[*]PostgreSQL support for sure.

If you have info what your current ROI is, you only need to add or subtract the Bucks amount to get your 'real' ROI, right?

Would it be a lot of work to insert the file menu opening hand history files (Full tilt) again besides connecting to an acces database, like the earlier versions? I don't use pokertracker and therefore don't have such database available.

example:
hero is firstin @sb and pushes allin with KK
bb calls with AA
was that a -ev move? or does the program calculate that the bb got lucky to catch aces?

ok i imported 1 sng
11$ i finished 3rd for 20$
Luck: 50.2%
Bucks: 55.27$ <- can anyone explain this to me?

SnG LA doesn't have it's own hand parser at the moment. Are you using any other kind of poker tracking software?

Was it a 10-Max SnG? Than it should actually be 22$ for 3rd place. A screenshot would help.

Your figures tell me, that you should have busted out of this SnG more than once

nope u win 20$ @ partypoker


and answer to this please =)

So it was a 10$ SnG with 1$ rake? Could you post a screeny?

Neither. SnGLA doesn't calculate whether it was a +EV push or not. This would require the user to set calling ranges.

The program expects a certain outcome, in this case that the aces win. If you suck out with your kings it will calculate a positive luck-factor and the bucks which you gained against statistical expectation.

screeny of what?
yep the PP sngs are: 10+1, 20+2, 30+3 etc

so if the opponent catches a good hand everytime i try to steal iam a fish according to the prog? then its no wonder thats my expected graph is so low =)

Could you mke it compatible with Tourneymanager, bodypull?

So I dont have to ask for tournament summaries from Pokerstars,

because Tourneymanager makes the summary when I import the HH of the Tournament.

You already added the screeny I was asking for, thanks.

There is an issue with Party. Pokertracker doesn't assign returned pre-flop bets at all for Party games. I will have to build in a routine which filters them out.

Until then the results for Party are wrong.

It doesn't do it for any street - overbets are stored as part of the bet fields. The PokerTracker 2 database is full of issues like this - it's majorly screwed up. Fields like flop_bet, turn_bet etc contain all kinds of inaccuracies, and the bugs are never fixed because the data isn't displayed in PT. In addition, site hand history changes will import corrupted data for a period until the user downloads the patch, and often reimporting isn't done, or the fix by the patch doesn't fix the other fields. I've spent many weeks mapping out and working around these issues for ring games, and having to do it all again for tournaments to get an accurate program was the main reason I didn't add tournament support a long time ago. I know it's nice to have something which appears to work, but if it's not fairly accurate then it's worthless.

To BodyPull: I'd strongly suggest going with your own hand parsing for the main sites, or at least limiting imports to sites that you know work (Stars, FT maybe). EV analysis is really sensitive to accurate data, and a few games played at another site (or Party, given your current algorithm) makes the data totally unreliable.

Also, having your software connect to the internet is a bad idea, and I think your program will do a lot better if people know they can trust it. You could be anyone, and even if the program is clean now, any one of your updates could contains a trojan dropper or send private information once the program is given free internet access. There's no reason to connect to the internet to get the date/time - it's available on the local machine and in the hand histories for real time scanning. If you want it for a beta timeout, you're far better off getting that from the hand histories or database- the current system is so easily hacked by sandboxing or using a local proxy server. Give users the option of running it without internet access and you'll get more customers

Nice program BTW and good luck with it.