I need help with the Math, PLEASE!

[LinusE]

Hi all

I have been playing poker fulltime for 8 months and have had some problems with one of the sites I play at. I have almost won nothing at this particular site so I started to check stats and decided to look into 5 keyareas and guess if I was surprised to see this (below). It is actually amazing that I've won anything at all.
I have played NL with 6 players all hands and mostly NL400 (2/4).
Is this normal fluctuation? I think NOT and need someone with very good knowledge in the Matharea to convince me that this is normal.

I think this particular site steal from winning tight players and give to loosing players to keep them playing.
Here goes:

After 55088 hands at battlefield these are the stats

Saw flop with pocketpair:
2787 times hit set 273 (should be 11,86%, 330,5 sets) thats 9,79%,
missing 57,5 sets ( -17,5%)

Saw turn with open ended straightdraw :
486 times hit straight 41 times ( should be 17,0%, 82,6 times) thats 8,4%
missing 41,6 straights (-50,4%)

Saw river with open ended straightdraw:
531 times hit straight 86 times ( should be 17,4%, 92,4 times) thats 16,2
% missing 6,4 straights (-6,9%)

Saw turn with flushdraw:
654 times hit flush 84 times ( should be 19,1%, 125,2 times) thats 12,8%
missing 41,2 flushes (-33,0%)

Saw river with flushdraw:
716 times hit flush 102 times ( should be 19,6%, 140,3 times) thats 14,2%
missing 38,3 flushes (-27,3%)

These are just the stats I have looked into so far but since they were
pretty remarkable I stopped and sent you this.
I will also look into pair on flop becoming twopair or trips should be 10,6%
and gutshutdraws, hitrate for AK, AQ on flop and so on....

Please Help me

LinusE

[GeeseHoward]

I will let someone else do the the maths but what I can tell you you something from working in the poker industry for the last 3 years, in two well known poker rooms.

Poker rooms make thousands of dollars a day, they don't need to jeopardize their entire business to make a few $$.

"steal from winning tight players and give to loosing players to keep them playing".

[MasterLJ]

55k may seem like a lot of hands, but you are talking about events that happen 300ish times out of 55k times (0.545%) you need a HUGE sample size to see your numbers converge to the "real" percentages.

This is why the consensus for figuring out your "real" win rate is after 200k+ hands.

I know from creating several hold'em sims that you don't start to see strong statistically correlation until each individual situation has occurred 1k+ times. Flopping a set is certainly part of the individual events that make up holdem, so to see convergence you'd need 1000/0.00545 or 183,486.29 trials if my model is even remotely correct (I've never done a mathematical proof to figure out how many trials lead to convergence, so I'm just going from experience).

At any rate, keep it up. If you are anywhere in the green with your win-rate given the stats you posted, you are doing very well. And I hate to use the term, but you are "due" for some better luck.

[MasterLJ]

Btw, where are you getting these statistics?

To my knowledge, PT does not have a feature that breaks it down like you describe (I could be wrong here). It's very possible that whatever your using is not identifying the scenarios correctly, thus counting incorrectly.

[Richter]

standard deviation, google it

[LinusE]

I am getting theses statistics from a program that is called pokeroffice and the statistics are correct.

I am not just a badbeat-tired dude with a paranoid personality. I have always believed that any pokersite messing with the RNG would be insane, but theses figures are pretty extreme. I went through 6 years of college in 2 years and got straight A:s so I am not that stupid either.
I am pretty lazy though soo I stopped the studies becasuse I got a very wellpaid job and therefore I do not have good enough knowledge in combinatorics.

Please refrain from the standard answer and look at it from a mathperspective. This scenario might be possible but to get a picture, compare it with rolling a dice and hitting 6 that is 16,7% and if u roll a dice 486 times you would expect to hit six 81 times but what are the probabilities that u only hit it 40 times. Here you have 5 indepentent scenarious within the hands 1 to 55088. Note that this is NOT a worst scenario it is all hands played at this site.
And these 5 scenarious are all pointing the same way (4 of them pretty extreme and only one that appears to be in the normal range)

Could on add these after each other since they are independent of each other? then u would get a bigger samplesize and part of why theses figures are remarkable is that they all point in the same direction

[LinusE]

[ QUOTE ]
55k may seem like a lot of hands, but you are talking about events that happen 300ish times out of 55k times (0.545%) you need a HUGE sample size to see your numbers converge to the "real" percentages.

This is why the consensus for figuring out your "real" win rate is after 200k+ hands.

I know from creating several hold'em sims that you don't start to see strong statistically correlation until each individual situation has occurred 1k+ times. Flopping a set is certainly part of the individual events that make up holdem, so to see convergence you'd need 1000/0.00545 or 183,486.29 trials if my model is even remotely correct (I've never done a mathematical proof to figure out how many trials lead to convergence, so I'm just going from experience).

At any rate, keep it up. If you are anywhere in the green with your win-rate given the stats you posted, you are doing very well. And I hate to use the term, but you are "due" for some better luck.

[/ QUOTE ] tyvm for your opinion

[Louie Landale]

Yes, US and EU folks interchange commas and periods...

Those do seem unrealistic. Is it possible to parcel your stats up into bundles; perhaps stats for your first 5,000 hands, your second 5,000 hands etc? Then you'd have 12 sets of stats.

If these 12 sets of stats are reasonably close to your total, then you greatly increase the chances that the stats accurately reflect your true chances, and therefor support your claims of cheating. If those 12 sets of stats vary widely, then you greatly increase the chances of being a 'victim' of unlikely random variations.

- Louie

[LinusE]

I have checked the pair becoming set stats in 5 equal periods and they are between 8,4-10,4 % instead of 11,56%

If one look at the straigtdraw from flop to turn it is very close to hit a 6 rolling a dice 17% vs 16,7% and I would like to get some help with figuring out whay the chances are that if one roll a dice 486 times ( avg. should get 81 sixes ) one only get 40 sixes? How often or how many attempts rolling the dice( in series of 486 ) will one need on average?

[LinusE]

I have had the scenario 16,7% in series of 486 run through mathematica (math SW) and the lowest number of sixes was 46 in 300 000 series which indicate that I would have to run almost astronomical amount of series to reach as low as 40.