I volunteered to write some code to do a test for one poster.

Although this is probably a futile exercise since true rigtards will find a million reasons to discount the test it may be of some interest to those of a less paranoid persuasion.

In the first instance the following data will be produced for a 40k HH sample:

The number of the following pocket cards dealt to hero:

AA
KK
QQ
JJ
TT
AKs
AK
AQs
AQ
KQs
KQ


Percentage of times A, K, Q, J appear on FTR Overall

Percentage of times A, K, Q, J appear to improve hero's hand creating a pair, three or four of a kind.

Unfortunately there is not so much that can be done with villian's hand since we only know his cards if he goes to showdown. However, provided the second two sets of data match and are according to expectations it's a reasonable indication that all's well.

I'm just waiting for the full dataset from jagnje then i'll post some results as soon as possible. Any sensible comments will be welcomed.

I made an excel spreadsheet to calculate these numbers, they are based on the first graph posted by the chap above.

Equity s.d per 2612 hands
0.51 626003.146
0.52 312813.6405
0.53 208333.4456
0.54 156030.3893
0.55 124597.9753
0.56 103600.6521
0.57 88565.99443
0.58 77257.74765
0.59 68433.50947
0.6 61347.80219
0.61 55526.19623
0.62 50652.37164
0.63 46507.30482
0.64 42934.51777
0.65 39819.22738
0.66 37075.31344
0.67 34636.88579
0.68 32452.66212
0.69 30482.12074
0.7 28692.80715
0.71 27058.40957
0.72 25557.35884
0.73 24171.79285
0.74 22886.7792

The poster observed a result 80k away from the mean if his equity was less than 0.57 on average this result is less than 1 s.d. from the mean and thoroughly uninteresting. Even if he was running at 63% which seems incredibly unlikely to me it is still only 2 s.d.s from the mean.

I doubt anyone runs at 0.57 equity for a sustained period. If we assume 0.55 is somewhere around the max sustainable in SNG, then the observed result is around two thirds of a s.d. away from the mean (i.e. about 25% of all players will run worse).

Sorry to break it to you but your data isn't even *surprising* let alone proof that you are doomswitched.

Why would you expect these two numbers to be similar? When hero holds A-A there are going to be significantly less {AKQJ} that can be dealt, as there are only 14 of them lying unknown somewhere, rather than 16 normally. Same thing for all of the hero's hands except TT (but that is not an interesting case since an {AKQJ} never improves him.

Actually I'm a bit confused what you are trying to do.

You are assuming a far too simplistic approach.

Obviously these factors will be taken into consideration.

Evidently.

Well, we can't give any constructive comments unless we know what the methodology you are using is, beyond saying 'this could work in theory'

edit - I did think about this a reasonable amount but I can't see any good way to make it work. Initially I thought we could just calculate the probability of KQ (say) improving to a pair of queens or better by showdown then comparing it to the poster's data for his hands where he has KQ. Sadly this doesn't work either - if the poster flops a pair of Ks or Qs he will tend to showdown, if he flops 8-5-2 he may fold on the flop, even if we would have made a pair by the river sometimes. Thus the set of hands reaching the river will be heavily weighted toward him having some piece, but how much depends on how lightly he peels.

You could treat each stage of the action entirely separately. i.e. Holding KQ, how often does the poster flop a pair or better compared to the expected value. Holding KQ without a pair on the flop, how often does he improve to a pair on the turn (ignore straights/flushes here they make the math beyond annoying).

Well, I've told you the methodology.

What you are questioning is the interpretation.

i.e. You assumed that once the counts had been made (e.g.) the number of times one or more aces appeared on the flop would be directly compared to the number of times that the same thing happened on the flop when the hero held AA.

This seems an odd assumption. Why assume that a stupid comparison will be made rather than one that makes sense?

That's it!

Obviously there are a lot of things that it would be nice to do but are impossible without knowing other hole cards.

You'll notice that I ommited straights and flushes from the improvement test list.

Ahaymadoots

So let me see if I have understood this now:

We look at the set of all flops and see how often an Ace flopped.
We look at the set of all flops hero saw where he was dealt A-A, and see how often an Ace flopped.

We compare the ratio of the two and see if it is close to the expected value. We can calculate the ratio explicitly with elementary probability.

Is this better than just calculating the probability hero flops an Ace when he holds A-A explicitly, and comparing it to the observed data?

Well this should work fine... looking forward to seeing the data.

When you create a public poll how do you check what people have voted?

The theoretical calcuation will obviously be sitting there with the observation but I think it better to observe the totals for the flop as well because it would show us if there is anything else untoward going on.

Actually I worded the original post about this rather badly because really the comparisons would all be to the theoretical expectations. The fact that two observations should match (after appropriate adjustment) is derivative and thus not of any great interest.

Either vote yourself or click 'view poll results'.

lol good 1!

LOL begging for your doomswitch to turn off. Just get good enough to beat the players and the doomswitch when its turned on, that way you don't lose long term and have no excuses.

Melina

Where do I click for View poll results?

before the streak my average equity was more than 51%.
during the streak it was 49% something.

here's excel file with all the all-ins.
http://ifolder.ru/13070752

it's 1.49 s.d. for last 2600.
and 2 s.d. for last 560. it's been getting worse.

of course, it doesn't proof anything.
but i lost enough money. i don't want to lose more waiting until it becomes 3 s.d. i'm out of PS.

my feeling is PS takes some luck (0.5-1%) from winning players in favor of losing ones. implemented by a system of coefficients. the idea is to keep the fish playing (i.e. paying rake) as long as possible.

it's too difficult to trace. must analyze correlation between winnings and all-in luck.

There were some #VALUE in the O column where you were certain to win or lose the hand - not sure if you see the same thing. I manually changed these to 0 and the corresponding N column to 1 or 0. Not sure why the formula didn't work on my version of Excel.

After I changed these I got 1.46 s.d. from the mean, not sure why that's different than your 1.49 s.d.

Is counting all hands as 0 for a loss and 1 for a win the same as calculating the variance per hand explicitly by using the actual pot-size and the possible gains or losses? It seems to me that if the pot-size is variable rather than fixed this ought to increase the variance markedly. edit. on second thoughts it appears to be definitely different. Is it possible for you to calculate these variances using the actual pot-size?

Certainly if you don't feel comfortable on a site then you should leave, but I'm surprised that a 1.49 s.d. from average result would lead you to believe something is amiss. After all, well over 10% of all players would run worse than this.

The 500 hand data doesn't seem relevant as it is a cherry-picked sample where you ran particularly badly.
Since you mention the last 2600, it would be interesting to analyse the complete sample assuming there are more than 2600 ?

Anyway it makes a nice change to converse with a riggedologist who actually understands statistics

Dear Full Tilt player,

Please don't waste out time with this nonsense or we will have to increase our rake to cover the cost of replying.

The statistics that you have fabricated from your patchy memory regarding your all ins is almost certainly wrong and if you could use one of the many tools available to prove otherwise I will gladly investigate the hands in question.

Please stop implying that there are doomswitches or that we favour "bad" players in order to make more money. If you are not happy with our software or the assurances that we give you of the security and integrity of our games you are welcome to play elsewhere.


You almost always get your money in as 70% favourite or better? Doesn't this mean you are either lying or you play bad?

WTF are you talking about? You continue to reference the UB scandal in relation to rigging the game in favour of bad players when they are completely unrelated. How is a software engineer going to benefit financially from rigging the game in favour of bad players? Stop clouding the issue with your rambling conspiracies.

Is it that difficult? If you can't work this out how can you expect to be taken seriously?

I'm also pretty sure there are occasionally players from those countries sitting at the same tables as those who believe the game is rigged. Does the riggedness engine work by country of residence now? And what about players living in countries that do regulate online gambling (quite a few EU countries)? This U.S.-centric worldview is pretty narrowminded.

It's kind of funny that the same **** happens live. Lots of people complain about how "unlucky" they are in poker and blame the dealer, the gods, their life, etc.

I excluded 1 and 0 allins as they don't have anything to do with luck.
Of course if you count them, you get less than 1.49.

I don't think pot size should be in the formula. For example, being a short stack on bubble, 400 chip pot is as important for you as would be 6000 pot (especially then you play DoN).

Last 560 allins it was even worse - 2 s.d. They took almost everything I won for 2 months.

Overall result: 6000+ allins, av. equity 49.34%, av. win 48.91%, 0.75 s.d.
To compensate this I have to win 52 flips in a row.

It's not the first streak. Every downstreak was steep, every upstreak flat. I gather upstreaks were not actually upstreak, rather normal play.
Anyway I don't trust the site so I've left.

The overall result is all that matters, anything else is just cherry-picking.

Of course if you don't trust the site as a result of running 0.75 s.d below equity for 6000 hands then you should leave, but I respectfully submit that gambling may not be for you as 30% of the time or so you are going to encounter an even unluckier result (assuming all sites are perfectly fair)

He's saying there are rigtards and then there are sensible people who just want an unbiased 3rd party to audit the sites so the rigtards can STFU and the sensible people have their reasonable concerns assuaged.

Petty insults aside, what's your point? Are you saying that they rig it so everyone breaks even? How do they decide who wins and who loses? How do they know if you're going to cash out if you win? How do they know you're not going to deposit again after going broke?

I wish someone would just log 1 million live hands, put it into PT or something and compare it to 1 million online hands from each site.