Possible issue with BLR Tech casinos craps software

[bb28]

Kudo's to you, jstrike for willing to be so open and transparent!

 

[jstrike]

Kudo's to you, jstrike for willing to be so open and transparent!

Lol. Better hold that thought. Don't think I've been this nervous in years

See, I designed this system from the ground up...I know it's not crooked, so I'm not afraid of some bias showing against the player, but I don't have the Doctor's tools and I really don't know at the end of the day if it's balanced, or if he'll find something predictable about these random numbers that's not visible to the naked eye, something that's less-than-random because of some screwup in my programming. I racked my brains for days when I was building this thing, trying to think of everything to make sure it would be balanced, but full disclosure here, I never took a statistics class, and this is why they say you shouldn't try to design your own RNG or encryption, because someone smarter than you will eventually come along and figure out a way to crack it with something you never thought of.

Which is why I think total openness is the best policy, because if it's gonna be cracked, it's better if it's cracked by someone who'll tell you about it. But it's still scary!

 

[Eliot Jacobson]

The video poker file appears to have instances of duplicate rows of initial cards with the same hand ID, but different resolutions of the hands. Maybe this was just in the SQL. Here are the three examples I found since the casino opened in July.

27811 7/7/2011 0:37 7/7/2011 0:37 4H TC 9S 3S 7C 8H KC AH QS 5H
27811 7/7/2011 0:37 7/7/2011 0:37 4H TC 9S 3S 7C JD QC QD 6H 9C

28414 7/20/2011 0:00 7/20/2011 0:01 8D 7D 4H AH 3S 7S 3D 4C 9D QD
28414 7/20/2011 0:00 7/20/2011 0:01 8D 7D 4H AH 3S 8H 7C TH 5D TC

41430 7/30/2011 21:54 7/30/2011 21:54 7S QS KS 3C JS KS 7S JS QS QC
41430 7/30/2011 21:54 7/30/2011 21:54 7S QS KS 3C JS AC 5H QH 2D 9H

Can you help with this?

Thanks.

 

[bb28]

Lol. Better hold that thought. Don't think I've been this nervous in years

See, I designed this system from the ground up...I know it's not crooked, so I'm not afraid of some bias showing against the player, but I don't have the Doctor's tools and I really don't know at the end of the day if it's balanced, or if he'll find something predictable about these random numbers that's not visible to the naked eye, something that's less-than-random because of some screwup in my programming. I racked my brains for days when I was building this thing, trying to think of everything to make sure it would be balanced, but full disclosure here, I never took a statistics class, and this is why they say you shouldn't try to design your own RNG or encryption, because someone smarter than you will eventually come along and figure out a way to crack it with something you never thought of.

Which is why I think total openness is the best policy, because if it's gonna be cracked, it's better if it's cracked by someone who'll tell you about it. But it's still scary!

My compliment still stands, Obviously you aren't trying to hide anything and that's a great thing. Even if he finds an error, it will be a learning experience for you and hey if he does it won't be the end of the world.

Anyway......I hope he reports back that all is well for you.

 

[jstrike]

The video poker file appears to have instances of duplicate rows of initial cards with the same hand ID, but different resolutions of the hands. Maybe this was just in the SQL. Here are the three examples I found since the casino opened in July.

41430 7/30/2011 21:54 7/30/2011 21:54 7S QS KS 3C JS KS 7S JS QS QC
41430 7/30/2011 21:54 7/30/2011 21:54 7S QS KS 3C JS AC 5H QH 2D 9H

28414 7/20/2011 0:00 7/20/2011 0:01 8D 7D 4H AH 3S 7S 3D 4C 9D QD
28414 7/20/2011 0:00 7/20/2011 0:01 8D 7D 4H AH 3S 8H 7C TH 5D TC

27811 7/7/2011 0:37 7/7/2011 0:37 4H TC 9S 3S 7C 8H KC AH QS 5H
27811 7/7/2011 0:37 7/7/2011 0:37 4H TC 9S 3S 7C JD QC QD 6H 9C


Thanks.

Ooohh. That's weird. That looks like a bug in the game. Are you seeing a lot more of these?

The hand resolution that you see recorded on the later hand, is actually the original deal of the next hand, which I see is missing in all three cases from the CSV. Here's how the next hands were dealt:

41431 AC 5H QH 2D 9H 7S AC AD QH 9C $0.05 $0.05
28415 8H 7C 10H 5D 10C 5S 10C 7D 9D 10H $0.25 $0.00
27812 JD QC QD 6H 9C 3H QC 7S QD JD $1.25 $1.25

But there's something screwier happening. What I'm seeing in these cases is that the shuffle triggered after the next deal, but before the re-deal. So the first cards dealt in 41431 were dealt from what was in the deck after 41430 was done, but the resolution came from a new deck. This could have led to players getting two of the same card. I'm trying to figure out how that could have happened; basically it looks like the game gave the OK to run the next hand asynchronously, but the shuffle got delayed somehow (must have been delayed for a few seconds...maybe a freeze on the server). Maybe -- maybe that could have made a user's button get lit a little early and they clicked at just the right moment before the new deck was finished being prepared.

Time to start hacking I guess...it's gonna be a long night...

[EDIT]
I'm showing four times it happened since opening. Hands # 27811/2, 28414/5, 41430/1 and 75420/1. That last one, weirdly enough, happened today, to me, since I posted these results when I was messing around on the site. And this hasn't happened since late July. (Any other duplicate hand numbers you see in the 20000's range are from a brief switch-over to a backup server, with a different numbering system, and they're completely separate hands).

[RE-EDIT]
I've confirmed the problem was that in a rare case, the game dispatched a fresh "deal" button to the player at a moment when the database froze up and the newly shuffled deck hadn't been recorded yet. This only affected video poker jacks & 5 hand video poker. As it was, there are no recorded cases in the 5-hand version, probably because it offers bonus rounds and so it takes a second longer to reset the "deal" button for the player.

Thank you for catching it.

 

[Eliot Jacobson]

Early on there were triplet examples that spanned multiple days, like this:

692 3/8/2009 10:55 3/8/2009 10:56 2C 4S 4H 6C 8C 4S 6S 4H TC 8S
692 3/8/2009 10:55 3/21/2009 0:26 2C 4S 4H 6C 8C 4H TC 7H TD 5D
692 3/8/2009 10:55 3/21/2009 0:26 2C 4S 4H 6C 8C TD 9S TC 2D 5H

Around the time the casino came on line, there were additional triplet examples:

28014 6/3/2011 14:35 6/3/2011 14:35 7D AH 5H 9C 9H 9C 2H 9H 2C 8C
28014 6/3/2011 14:35 7/16/2011 2:28 7D AH 5H 9C 9H JD TC 9C 8S AH
28014 6/3/2011 14:35 7/16/2011 2:28 7D AH 5H 9C 9H TC TS 9C JD 8S

Then there were the three examples that I presented above, with data exclusively after 07/01/2011.

I can work with the log files and just delete those data lines, they have very little impact on the overall statistical analysis. But, if you want to fix the log file and create a fresh version without duplicates, containing data just since the site opened, I would appreciate that. Either way, I'm done for the day.

Best regards.

 

[jstrike]

No problem -- here's the list since opening, with the four duplicate hands removed.
Link Removed ( Old/Invalid)
Not a huge sample source I guess...

 

[Eliot Jacobson]

I've done about as much auditing as I feel like doing on JOB video poker and the 1M roulette spins. I conducted quite a few statistical tests on each. A couple of things to note:

On JOB, I assume that most players play a game that is reasonably close to optimal, though very few will play optimally. Based on that, I looked at the distribution of final hands for the players and compared those to expectation. With two exceptions, these results were statistically normal. The number of expected straights was 59.94, the actual number was 85. This is 3.26 standard deviations above normal, making it about a 1-in-2000 shot. The other exception is in the RTP of the data. At RTP = 88%, it is well below the expected 96% to 97%. If I were to conduct ongoing audits, I would pay attention to these things, but for this audit, it is not a big deal. My conclusion is that JOB is operating normally. Finally, I mentioned above that there was a bug in the log files with repeated lines. I believe you figured it out and it is now fixed.

For the roulette spins, I conducted statistical tests until I couldn't stand in any more. Everything came back normal.

A couple of comments about game returns. For JOB, you have the 8/5 pay table, which returns 97.3% with optimal play. You offer double-0 roulette, with it's 5.26% house edge. These are on the low end for competitive online casinos. Many casinos offer 9/6 on JOB, and single-0 roulette. You may want to consider up-leveling your returns on these games to be competitive as a newcomer.

By comparison, you offer better than the best standard returns on your Three Card Poker. This surprised me, in light of JOB and RO. The Ante Bonus is 1/4/6, and the Pair Plus is 1/4/6/30/40. I wonder if there is a typo on the return for the Ante Bonus, it is usually 1/4/5. I ask this because 1/4/6 is a non-standard pay table, it isn't even listed on wizardofodds.com. I'd like to hear more about how you decided on the 1/4/6 pay table for the Ante Bonus on 3CP.

Please let me know if you have any questions.

 

[jstrike]

Wow. You're good. I mean man you're good. I can see why you get paid the big bucks. I can't believe you spotted that in three card poker

It's not a typo, what happened was when I started writing the code I cribbed my pay table from

You do not have permission to view link Log in or register now.

. Later on I realized it was better than average, but I'd already almost finished, and I decided to go with it. I know the payouts are relatively low on most non-skill games for the site. I'm trying to draw a market of skill players in games like ScubaCube and SapphireRummy (no luck yet, btw). But the goal in the end is to raise the video poker tables up to 8-6 and then to 9-6 when we've got a wider margin to play with and are comfortable raising the game stakes at the same time. One thing about the Bitcoin market is, there's not much competition. Not much play either, for that matter. As far as Roulette, the weeks it would take to rebuild it as a European style table might not be worth it while I'm still working on new games; but I did develop a 36:1 / 17.49:1 / 8.24:1 etc. (inside) American roulette wheel that can be played for a 1¢ minimum, (2¢ if you want to take better odds on edges, 4¢ to make better odds on corners -- that's why the odds are rounded down) that I'm thinking about deploying when the time is right. The idea was suggested by someone on this board, but some Roulette purists I've talked to have been bitterly opposed to it, I guess on the grounds of tradition.

As far as why there are 58% more straights while actual RTP is lower than expected, I can only guess... I get to watch live hands, and I've seen a fair number of VP players who don't play perfect strategy. Most of them that I've witnessed have some weaknesses. I've spotted a certain kind of player who'll almost always hold any three cards to a straight or a flush, especially late at night, when they're drunk I guess. Players chasing straights might explain those results. I'm not sure. I even watched one guy throw out a pat flush with three picture cards (it was something like JQA43) to chase a royal... that was on the 5-hand table, so it's not the worst bet I've ever seen (he had something like a 10% chance of hitting one for a pretty good payout), but still a little crazy.

Anyway, thank you for the time and energy you put into this... I really wish I could make it up to you. I'm sure this was a pain in the ass and not what you like to do on a Sunday (unless you're like me, lol). But the hell with iTech -- I want the Jacobson seal on my site as soon as I can afford it... and in the meanwhile, I might go for some consultation on my newer games (scubacube, mayan gold) where I've only been able to simulate the odds but still don't really know if I might've screwed up the calculations. You have a great eye for detail. And, you're a mensch. Thank you.

 

[Eliot Jacobson]

It's not a typo, what happened was when I started writing the code I cribbed my pay table from

You do not have permission to view link Log in or register now.

. Later on I realized it was better than average

From what I can tell, the pay table referenced through the link you gave is 1/4/5 on the Ante Bonus, not the 1/4/6 you have.

I'm sure this was ... not what you like to do on a Sunday

I enjoy this type of work. I like to invent new tests for data. I like trying to find something wrong with data -- to try and break it somehow. So this was no problem.

 

[binary128]

Wow. You're good. I mean man you're good.

I agree. Galewind has relied on Eliot for almost 3 years. IMO, there's no one better.

Chris

 

[jetset]

Back to BLR, guys.

I found this comment from Gambling911 interesting in the BLR context.

"Rachel Miller, General Manager of Legends, said less than 10 percent of Legends' 7,500 active gamblers use the casino.

"She said Legends continues to use BLR, in part, because the software works so well with the site's main function of handling sports bets. Miller insisted that from an operations standpoint, none of Legends players would have been adversely affected by the BLR setup. She said the company has invested thousands of dollars in new random number generators to work with the BLR software."

I guess you better not be one of the 10 percent with that sort of approach....

 

[binary128]

Back to BLR, guys.

I confess that I was confused, and remain confused, by this "new random number generators" statement from Legends. We are talking about an online Sportsbook that has achieved a

You do not have permission to view link Log in or register now.

, which is NOT an easy thing to do. I think that the starting position for an investigation into such a highly rated company should be one of respect. However ...

1.) The problem report started in

You do not have permission to view link Log in or register now.

on Oct. 28.

2.) Mike published his

You do not have permission to view link Log in or register now.

test results on Oct. 31.

3.) Eliot published his

You do not have permission to view link Log in or register now.

"too many 7s" test results on Oct. 31.

4.) Eliot started this Casinomeister thread on Oct. 31.

5.) boymimbo started

You do not have permission to view link Log in or register now.

, and then a

You do not have permission to view link Log in or register now.

, on Nov. 1, and wound up feeding extensively into a

You do not have permission to view link Log in or register now.

started Nov. 2. (My hat is off to boymimbo for "Service Above and Beyond".)

6.) And finally, The Wizard of Odds site published the

You do not have permission to view link Log in or register now.

on Nov. 2.


(Yes, there are a lot of links here, but that's because the documentation and data trail is huge.)


It has been repeatedly stated, on ALL of these forums, that the problem has nothing to do with the RNG.

It has been repeatedly demonstrated that the problem is the software "working around" the RNG, perhaps by simply calling for numbers until it gets one that it "wants", rather than working with the first one that it gets. (IMO, Eliot's "too many 7s" test is definitive here.)

As best I can determine, Legends first made their "new random number generators" reference on Nov. 2, several days after the RNG had been removed as a variable from the equation. However, I think it is fair to say that they made this statement before they were fully up to speed on the details of the issue.

But to then repeat that statement within the article in the Pittsburgh Tribune-Review column of Nov. 18, a full 2 weeks later ... confusion runs rampant.

 

[vinylweatherman]

They would rather carry on cheating the 10% of players that use the casino, rather than throw away all the money and effort they have spent on integrating BLR software into their platform, as well as having to spend even more money finding and integrating a new casino software.

They must know by now that the new RNG makes no difference, but are trying to "whitewash" the issue by pretending that they have dealt with the problem, and the casino software is now working fairly.

The one thing that could bring them to justice is to repeat the tests on the new RNG implementation, and prove that the cheating still takes place as before.

Any operator that takes the attitude that running gaffed software is OK because it "affects only 10% of our players" does not deserve a premium A+ rating, even though the cheating has nothing to do with the sportsbook product. The problem being that they may take the same attitude if another problem emerges that only cheats a small minority of players, yet is disproportionately expensive to fix or to pull the product altogether.

If it is only 10% of players, why not pull BLR altogether, and forget about spending money just for those 10% of players, who can just as easily sign up to a casino. It can only mean that this 10% of players are generating a significant amount of profit through their casino play, more so than normal, because they are being cheated at the craps table.

 

[Westland Bowl]

It's ironic I discover this post with many things I've been saying for the past two years until after I gave an ultimatum post six weeks ago. I'm bored of my government auditing job. I now want to audit online casinos!!! I know where to look for these hidden manipulations of game results.

 

[Casinomeister]

Back to BLR, guys.

I found this comment from Gambling911 interesting in the BLR context.

"Rachel Miller, General Manager of Legends, said less than 10 percent of Legends' 7,500 active gamblers use the casino.

"She said Legends continues to use BLR, in part, because the software works so well with the site's main function of handling sports bets. Miller insisted that from an operations standpoint, none of Legends players would have been adversely affected by the BLR setup. She said the company has invested thousands of dollars in new random number generators to work with the BLR software."

I guess you better not be one of the 10 percent with that sort of approach....

Pure roguish in my opinion. If merely one player is allowed to play on cheating software, they should be drummed out of the business.

 

[Mousey]

Pure roguish in my opinion. If merely one player is allowed to play on cheating software, they should be drummed out of the business.

Where the heck is the 'regulation' and player protection? Looks like it would be illegal for regulated casinos and bookies to knowingly and willingly use gaffed software??

 

[Eliot Jacobson]

Pure roguish in my opinion. If merely one player is allowed to play on cheating software, they should be drummed out of the business.

I made this post at SBR in an attempt to have them downgrade their rating of Legends -- copy here:

It is outrageous that Legends defends its use of the corrupt BLR casino software.

First, the RNG has nothing to do with rigging the software. It may be that they opted out of using the feature that allows the results to be rigged, if such a feature exists. However, the way the software cheats is that the software weights the outcomes of craps in such a way that it guarantees a huge winning edge for the casino. If the RNG chooses a number according to that weighting, it makes no difference that a new RNG is used. It makes no difference how the random number is chosen, if it is already determined that a "7" is rolled over 50% of the time.

Second, Legends continues to use software that they know has the potential to be rigged, and has been used to offer rigged games at other casinos. They continue to use a rogue product offered and supported by a rogue company. They continue to accept real money wagers on a product that is known to have cheated people at other casinos and may be cheating people at their own casino.

Third, giving Rachel Miller the benefit of the doubt is no better. If she is not technically savvy and simply does not understand the issue, but insists that changing the RNG fixed the problem, then the implication is that Legends knew the software was rogue. It means that Legends was aware that the software is able to cheat players, and that Legends made an attempt to disable that feature. They may call that fixing the RNG or whatever, but it makes no difference. They went through the trouble to fix the RNG because they knew that the off-the-shelf product could be used to cheat players.

These three points make the continued use of BLR software indefensible for Legends.

 

[jetset]

Pure roguish in my opinion. If merely one player is allowed to play on cheating software, they should be drummed out of the business.

I couldn't agree more - knowingly allowing one player to run the risk of being cheated is completely unacceptable....standing by and being prepared to watch 750 casino players use suspect software even more so, if that was possible.

To me, this speaks to a lack of care, professionalism and integrity.

 

[boymimbo]

I couldn't agree more - knowingly allowing one player to run the risk of being cheated is completely unacceptable....standing by and being prepared to watch 750 casino players use suspect software even more so, if that was possible.

To me, this speaks to a lack of care, professionalism and integrity.

the way I see it, World Wide and 5 Dimes was using the rogue software with the same results, meaning that the software provider (BLR) was at fault, not the sportsbook. Legends makes a claim that they have a different RNG. The only way to prove the issue is to do a test at Legends and see what the results are.

 

[binary128]

I made this post at SBR in an attempt to have them downgrade their rating of Legends

For those that are interested, the post to which Eliot refers

You do not have permission to view link Log in or register now.

.

Sadly, that forum's Administrator lumped Eliot (and apparently the article in the Pittsburgh Tribune, as well as the Wizard of Odds) into the "all online Casinos are rigged" crowd.

This "all online Casinos are rigged" kind of post is all too common on the SBR forum, which I think goes to explain this administrator's rather aggressive response.

Also true is that this speaks to the "wall of awareness" between Sportsbooks and Casinos (or vice versa). SBR has a quality reputation for dealing with Sportsbook-related issues. Casinomeister has a quality reputation for dealing with Casino-related issues. Casinos that are part of a Sportsbook fall into a kind of "nether region".

Again, kudos are due to boymimbo for once again stepping into the fray and trying to bring the SBR discussion around to the available hard data that is available.

Personally, I hope that his efforts are fruitful, and that the data, which IMO is conclusive, causes this forum to address the facts, rather than merely abusing the messengers.

Any Casinomeister members who are also on SBR would, I think, do well to assist boymimbo in his efforts.

It also continues to astound me - NO ONE is complaining about the RNG. The RNG has NO PART in this problem. Why does anyone questioning this issue's applicability to Legends KEEP TALKING about the RNG? I'm stunned. I'm just stunned. I'm usually pretty good at coming up with analogies, but for this one, an analogy simply escapes me.

 

[jstrike]

It also continues to astound me - NO ONE is complaining about the RNG. The RNG has NO PART in this problem. Why does anyone questioning this issue's applicability to Legends KEEP TALKING about the RNG? I'm stunned. I'm just stunned. I'm usually pretty good at coming up with analogies, but for this one, an analogy simply escapes me.

I think the difference between a corrupt RNG and corrupt middleware, or a corrupt game client for that matter, is basically meaningless to anyone who isn't a programmer. We understand that if any of those things were crooked, the game would be just-as-crooked as if all of them were corrupt. I don't think most players realize how RNG numbers get processed into the spins or cards they're actually seeing, any more than most McRib lovers think about what succulent part of a pig they're biting. What basically matters to the consumer is: (A) is it FDA approved? (B) Do I personally know anyone who got ill from eating this, and (C) how many calories. Everything else you can or can't say about it is bullshit, but just like fast food, the casino industry has chosen a few little tag-words that everyone can latch onto, so as long as those things "sound okay", the marketing goes, the product must be good for you.

Trans Fats, low cholesterol, "organic", "natural ingredients" -- whatever the hell that means =
Expected RTP, an expensive RNG, "Flash", "secure", etc.

But the point is, I'll eat a sausage McMuffin for breakfast and not trouble myself about how it was made, basically 'cause I know no one else is dying from it. I don't care about how they got that into a bag for me this morning. I just want to know everyone's getting the same shit and it's not poison.

This is why I advocate publishing hand results. The industry gets mileage out of this RNG ridiculousness, which doesn't mean anything; and at this point it's a placebo that rogue casinos can use to calm players down, so let's do away with it. All RNG's out there now are random enough that unless you're a 3rd degree ninja hacker, you're probably not going to win or lose because the RNG's "off". We should be talking about whether the sausage is edible, not whether the ingredients are "natural".

 

[Nifty29]

It's ironic I discover this post with many things I've been saying for the past two years until after I gave an ultimatum post six weeks ago. I'm bored of my government auditing job. I now want to audit online casinos!!! I know where to look for these hidden manipulations of game results.

Quick aside here....

Given your return after the passion and drama of your "close my account" post....are you now saying that you have changed your mind, or do you still believe that "all casinos cheat"?

 

[Westland Bowl]

Quick aside here....

Given your return after the passion and drama of your "close my account" post....are you now saying that you have changed your mind, or do you still believe that "all casinos cheat"?

How about "all US-FACING online casinos cheat"?? Yes, I still post here as my Casinomeister account wasn't closed but I still do not bet online anymore. Well, I take that back, Sports Book Review forum gives out free points each day to use in their DGS casino so I once in a while see if I can out-smart that rigged casino with a new strategy. As I use various different strategies with brick-and-mortar casinos as I have with online casinos, I do not wish to reveal how I play. However, I can tell you that if you knew my strategy, you will readily see how these online casinos are soooo adaptive to it and it was very noticeable to me since November 2009. I wouldn't say that these online casinos go as far as BLR has but they do so in varying degrees but it is still all the same b.s.

 

[Jasminebed]

No disrespect Westland, but Heroes Casino that you "cheated" by employing a system was BS, and Galewind (i.e. Binary) stepped up to the plate and said their software didn't cheat, only non-random software could be beat LONG TERM (emphasis mine), and truly stepped up to the plate by yanking their licence and paying you out of their own pockets.

For those members not familiar, you can find that thread here: https://www.casinomeister.com/forums/threads/heroes-casino-blackjack-issues-slow-pay.28007/

IMO systems can help you lose slower, cashout more often OR larger, or keep you more entertained by employing them. Almost anything that slows your play will limit your loses including calculating systems, or kissing your lucky troll doll, or getting out of your chair and circling three times counterclockwise.