The opening poster – going by the handle ‘katie91’ – reported receiving abnormally bad results when playing the ‘Reel Deal’ and ‘Hi Lo Gambler’ games. It has to be said right at the start that all credit should be given to katie91 for the initial investigation as it was very well laid out, consisted of both video logs and game history logs for all playing sessions and a full mathematical analysis of the results that they received.
To summarise, katie91 firstly looked at Reel Deal and placed a total of 1936 wagers mostly on the ‘Red’ bet with results of Red occurring 9282 and Black occurring 10074. As the game is based on a deck of cards with the 10 value cards removed (10s, 10h, 10c, 10d) the distribution of colours in this game should be 50/50. The odds of losing a 50/50 bet 10074 times out of 19356 trials is 1 in 154028680. A similar but smaller sample was taken from the Hi Lo Gambler game which once again show a very unlikely loss (1 in 41987).
At this stage renowned gaming expert and gambling author Eliot Jacobson PhD agreed to verify these results. Below is the results of his analysis;
My understanding of “Reel Deal” is that it is advertised a fair game, with RTP = 100%. In this case, Red (R) and Black (B) should be equally likely, as should any sequence of R & B.
I conducted a “correlation” test on the data provided by Katie91. First, I took the raw log file and sorted it chronologically. Then I considered all sequences of five consecutive games. In each game, either a R or B card was drawn. A fair game would give a roughly equal distribution for each of the 32 sequences BBBBB, BBBBR, BBBRB, … RRRRR. Altogether, there were 19352 such sequences. It is worth noting, in looking at this data, that Katie91 wagered on Red over 19197 times.
The following table tabulates the data for chronological sequences of five consecutive games:
Jacobson then gave a statistical analysis of these results, demonstrating how extraordinarily unlikely this bias was to have occurred by chance. His work verified and extended the conclusions of Katie91’s original post. Jacobson continued:
Michael Shackleford also analysed these result and stated on his site that “The probability of luck this bad or worse is less than 1 in 200 million, or five standard deviations south of expectations.” (see http://wizardofodds.com/online-gambling/blacklist/).
After this damning report, Dr Jacobson followed up his analysis with some further research;
I wrote some primitive C code to explore the outcomes produced by the game Reel Deal as played by the OP. In this code, I assume that a player always wagers on Red and simulated 19197 rounds, the same number of rounds the OP wagered on Red. The intent of the program was to produce an RTP of 96% (the approximate RTP of the OP) and to get a distribution of outcomes that I could run a chi-squared test against the distribution of the OP.
This code has two steps. First it chooses a color, Red or Black. It is biased towards Black, so that 48% of the time the program chooses Red and 52% of the time it chooses Black. Once a color is chosen, then a number is chosen at random for the card. Thus, if the color is Red, then the program chooses a number at random from 1, 3, 5, 7, 9, 11. If the color is Black, then the program chooses a number at random from 2, 4, 6, 8, 10, 12.
Every time I ran this code, the even numbers (Black) were the top 6 numbers by frequency and the odd numbers (Red) were the bottom 6 numbers by frequency. The results of the OP are in the yellow column. The results of three sample runs are in the next three columns. Along the bottom are the p-values obtained by comparing the OP’s results with the sample runs of the biased program.
Note, this analysis does not prove that the code for Reel Deal was written in the manner given in the code below. This analysis shows only that the method used in the code below cannot be excluded.
Here are the results:
At this juncture, the fact that the Reel Deal game was not function within expected parameters has been completely established.
As the real investigative work had been completed by katie91, Eliot Jacobson and poster ,strong>thelawnet (we’ll discuss their input shortly) the issue had now been clearly established this was the point where the mop up work like contacting the relevant authorities needed to be addressed.
The rest of this article is specifically going to look at the response of the affected casino. For the discussion of the reaction of the various regulators please see our ‘What does a Regulator do?’ article.
During the course of the thread Betfred made two statement regarding these games and the action they would take as a consequence.
The investigation into the complaint raised here in regard to the Reel Deal game has started yielding results, and as promised before they will be disclosed in this thread. The below is confirmed by Finsoft – the supplier through which Betfred receive Realistic Games products.
Analysis has revealed that Reel Deal was indeed returning at 96%, despite being advertised at 100%. Finsoft’s review revealed that this was the result of an administrative error on the game’s deployment to Betfred, where the wrong help file was attached to the game.
While mistakes do happen, Betfred realise this is not acceptable. Neither Finsoft or Betfred would purposely mislead players, and will therefore actively compensate players on losses derived from the game over the last 6 months. The amounts should be in accounts by next Tuesday. Claims beyond 6 months will be accepted and honoured, too, but must be submitted individually.
The remaining items brought up in this thread are still under investigation by our suppliers; they are:
The suggestion that the game in question’s RTP is adaptive
The suggestion that the game performs differently in fun and real modes
It’s been suggested here that the parties involved have blocked Eliot’s investigation. This is simply not true. However, it is possible that some people contacted directly were away over the Holiday period or were not in a position to comment on confidential information to none permitted individuals. Let me assure you that all concerned take this matter very seriously, and I hope this was conveyed in the game being removed and in the compensation now offered.
If you have a claim for losses on Reel Deal then please email [email protected] and we’ll process the amounts in cash and in full to your Betfred accounts ASAP.
Analysis of Statement 1
While Betfred acknowledge in the above statement that the game was in fact not performing at the listed RTP, the RTP that they claim the game should have had would be impossible to achieve using a fair deck of cards.
To clarify, when electronic gambling game depicts a real device – like cards, dice, coins or a roulette wheel – the most fundamental rules of gambling and the contract of trust between players and operators is that the device that is depicted should behave in the same fashion as the real device. This is actually one of terms of Betfred license.
If the cards depicted in the Reel Deal game behaved in the manner that real cards would the RTP of this game would be 100%. The fact that the wrong help file may have been attached to this game only implies that the game was programmed to produce non-random. Rather than resolving the problem this statement raise the possibility that this game was intentionally designed in a manner that is seriously ethically compromised.
Apologies for the delay, but the analysis of significant amounts of data and liaising with multiple suppliers takes time. We have moved as quick as possible without jeopardising the accuracy of the results, which you will find below.
Realistic Games provided the assets and rights to the Reel Deal game but SPIELO G2 developed the game for their operators and in doing so changed a number of core features. As such, it is not right to identify Realistic Games as responsible for how the game performs.
On developing the game, SPEILO G2 developed two version: fixed odds and fixed price. The latter was in operation at Betfred. Fixed price meant that randomness could be introduced via a certified (GLI and TST approved) RNG and an RTP was introduced. In this case, at 96% RTP.
The development of the game in this way resulted in SPELO G2 inadvertently running the fun version of the game on a fixed odds model and not a fixed price, and therefore it ran at a different RTP.
Finally, during the deployment of the game to Betfred the wrong help file was associated with the game and reported the wrong RTP.
Our initial offer of compensation in regard to the help file was made. However, having reviewed the analysis from SPEILO G2 and our own, we accept that Betfred Games has been running two versions of the same game for free and money play respectively and that is simply not acceptable. Based on that we will be refunding all losses on the game from when the game was introduced to Betfred, and will be removing other Realistic Games provided by SPIELO G2 to complete a review of their configuration, help files and RTP and until we’re confident in their accuracy. Compensation payments will be issued within 7 working days.
We would like to take this opportunity to apologise to our players and to thank the OP and the Casinomeister forum as a whole for bringing this to our attention. The integrity of our games and operation is of paramount importance to Betfred and value any feedback that strengthens or corrects our operation.
Finally, our logs and cooperation will remain open to authorised parties to further any part of this investigation.
Analysis of Statement 2
First off, discussion of the integrity of Betfred’s RNG is absolutely irrelevant in this specific instance. The RNG could be functioning perfectly but the game could still be programmed to produce bias result.
If I have a RNG designed to select a number at random between 1-26. It functions perfectly, but I assign the numbers to a suit of cards as follow;
# RNG Selects
Now even through the RNG functions perfectly, the set up of the game uses the RNG results in such a way as to ensure 2-6 turns up 3x as often as T-A when every card should turn up an equal number of times, which is a clearly misleading practice. Moreover, even stating the correct RTP of this game wouldn’t make it fair. The use of a deck of cards intrinsically implies a probability to the customer that this game has distorted.
Alongside this Betfred’s statement once again ignores the fact that to produce a 96% RTP the cards in the game would have to behave in an unnatural fashion, something that is both unethical and in specific violation of Betfred’s GRA license. What this statement did achieve was to release Realistic Games from any responsibility for the way these games functioned.
Instead we turn our attention to FinSoft – another UKGC regulated company. Aaron’s statement above explicitly states that FinSoft “changed a number of core features” and that they “two version: fixed odds and fixed price”. Without changing the paytable it would be impossible to create two different fair versions of this game. As the paytable was never changed it’s now fair to conclude that FinSoft – an arm of the Spielo G2 company – intentionally programmed and distributed a game that while using cards as its gaming device, failed to comply with the natural odds of the device, a practice which is completely unethical and unacceptable.
Free Games – Another Huge Problem
While all of these issue with the Reel Deal and Hi Lo Gambler games were being investigated another issue surfaced with games being offered by FinSoft/Spielo G2.
When you play a Hi/Lo game, the number that is currently display (the last result) appearing again – i.e. the same number twice in a row resulting in neither a higher or lower result – would be considered an automatic loss for the player. In the free play mode the various Hi/Lo games provided by FinSoft/Spielo G2 have been excluding the last result from the range of possible results on the next wager. This would dramatically increase the amount of funds that is returned to the player. This only happens in the free play versions of the game.
To confirm these issues, we did some of our own testing at both Betfred and Bet365 on the Aladdin’s Treasure Hi/Lo game.
In this first video we looked at the Real Play version of Aladdin’s Treasure at Betfred Casino. We played 100 spins and during that 100 spins there were 4 occurrences of the same number appearing on two concurrent spins. As this game uses 20 numbers, if all of the numbers are weighted evenly the chance of a number landing a second time after the first occurrence is 1 in 20. As such we would expect to see 5 occurrences of the same number appearing on two concurrent spins – 4 is well within the expected range.
In the second video we repeated this same test on the Free Play version of Aladdin’s Treasure. This time round we played 500 rounds without a single occurrence of the same number occurring twice in a row. If the numbers are weighted evenly the odds of this happening are (19/20)^500 = 7.27×10^-12 or 1 in 137,466,652,000.
There is no chance that this game is behaving in the same way when playing for free as it does when playing for real.
Now Bet 365
The same test is implemented on Bet 365’s Real Play Aladdin’s Treasure with exactly the same 4 doubles in 100 trials.
And again we repeat the test on the Free Play version, this time only playing 350 spins, again with no double numbers. The odds of this happening are (19/20)^350 = 0.000000015 or 1 in 62,623,617.
There is no chance that this game is behaving in the same way when playing for free as it does when playing for real.
The practice of setting free play games to ensure that players win more than they would if they were playing for real is an obvious attempt to mislead the player and encourage them under false pretences to deposit and risk funds, it’s a behaviour that is considered by this site to be rogue and it once again breaches the GRA license.
Unquestionably FinSoft/Spielo G2 have crossed both ethical rules and the licensing agreements of many of the casinos they’ve provided games to. If this could be explained away as a bug, removing the games and compensating affected players would most likely be a reasonable response. This can’t be explained away as a bug. Between Betfred’s statement clearly outlining that Spielo G2 have altered games and programmed them to produce non-random distributions of cards and the work done by thelawnet showing that the FinSoft platform was altering the way the various free play games respond, these games were without doubt programmed with the intention of misleading players about their likelihood of success. As such, trusting any game that comes from FinSoft/Spielo G2 is no longer an option. Nothing short of complete removal of ALL games provided and reimbursement of funds to all players who played on any game that has been proven to have been affected is a suitable response for any reputable and responsible casino.
Response of the Casinos
The first venue to take any action beyond suspending the real play games, Nordic Bet casino went the full distance to protect their reputation, dropping FinSoft/Spielo G2 software completely as a provider and refunding customer losses on any affected games. This is the sort of responsible behaviour that the very best casinos in the industry should have and we could not commend Nordic Bet more highly.
BetVictor’s name was raised on a couple of occasions during this investigation, firstly as they offer some Realistic game (the games are provided directly via Realistic and don’t involve FinSoft/Spielo G2 at all) and secondly as they offer games from Dynamite Idea. Dynamite Idea are a casino game development company that were bought over by Spielo G2 several years ago and are now to all intents and purposes Spielo G2 by a different name. We discussed this at length with BetVictor casino, who explained that the games were acquired prior to the takeover of Dynamite Idea by Spielo G2, are hosted on BetVictor’s own servers and run the original code having never been updated. Therefore Spielo G2 has never had any influence these games and BetVictor should not in any way be connected to this issue. I’d personally also like to stress how forthcoming and communicative the BetVictor team were in dealing with some very serious issues. Again another top venue showing what it means to be responsible.
As with BetVictor, Grosvenor and the other Rank Interactive properties (Blue Sq and Mecca) do offer a small selection of Dynamite Idea games however these games – again as with BetVictor – were purchased prior to Spielo G2 buying Dynamite Idea and the code has not been changed since. In dealing with this issue Grosvenor were easily the most proactive of the groups involved, getting on the phone as soon as they were alerted and following up with further discussion as soon as further information was available.
Virgin casino did offer a selection of Dynamite Idea games, but during their recent takeover by GameSys have got rid of them. It should be noted that this decision likely had nothing to do with this issue as Virgin got rid of the vast majority of their game catalogue at this time.
Having looked at the 888 casino game catalogue they do not appear to have any games provided by any of the affected providers and their rep claims no knowledge of any relationship between then and Spielo G2. As such their inclusion on the Casino City list is a bit of a mystery.
It should be noted that it’s difficult to determine who is using FinSoft/Spielo G2/Dynamite Idea from game selection alone. We’ve been using this list – http://online.casinocity.com/software/spielo-g2/ as a guide, but the list has proved to be rather inaccurate. We have been informed that Ladbrokes, Coral/Gala, Grosvenor and BetVictor have no relationship with Spielo G2. If you know of any group we had not addressed that is using this provider please let us know at [email protected].
Not only do Betfred continue to work with a group who have been intentionally cheating, their statements to this date on this matter have shown a woeful lack of understanding of the main issues here – that being them offering games that are in clear breach of their license and are without question cheating customers by misleading them to believe they had a higher chance than they actually did, but their initial offer to refund losses only up to six months unless directly contact by the customer was entirely unacceptable. They have bungled their way through this issue and rather than do the responsible thing and admit their mistakes and renew their commitment to only work with reputable and honest providers, have instead decided to pretend there is no further issue to address. Of the bad, Betfred are the worst.
While Bet 365 casino never offered any of the cheating versions of FinSoft games themselves, they’ve made it very clear that they see no ethical issues with continuing to work with a group that have been intentionally cheating player. This is hardly the behaviour that players should expect from a market leading organization.
After extended conversations with Boyle casino, it was confirmed that they do work with FinSoft, but are unwilling to terminate that agreement. They have stated that they’ll be keeping a very close watch on their FinSoft games from this point on, but that still leaves them in the position of choosing to work with a company they know has intentionally cheated players. Far from an industry leading example.
Paddy Power appear to have some level of relationship with the Spielo G2 group but were unresponsive to multiple attempts to communicate with their affiliate team.
Another group that despite repeated attempts to discuss this issue with them, failed to get back in contact with us. Very disappointing.
Unfortunately Casino.net never really stood a chance in this situation. Their entire casino platform is Boss Media software, the former name for Spielo G2.
Gold Club Casino
Not to be confused with Club Gold Casino – who are a Playtech brand – Gold Club Casino, like Casino.net, use Boss Media for their entire casino platform.
Sky Vegas Casino
Sky Vegas offer a range of Dynamite Idea games and unlike BetVictor or Grosvenor make no claim to have attained them prior to their take over of Spielo G2 and show no willingness to remove them. In a case like this the rules are very simple – if you continue to work with a company that’s been shown to be dishonest, then you are no longer trustworthy yourself.
Stan James Casino
Stan James were directly affected by the FinSoft game issue and while they appear to have pulled them they continue to work with Dynamite Idea. Simply put you can’t have it both ways – you’re either only working with companies the player can trust or you’re not trustworthy.
This investigation is still ongoing as many more properties that we don’t work with directly are likely still to be working with FinSoft/Spielo G2/Dynamite Idea. This case is serious enough that we no longer feel that it is a good idea for any casino property that cares at all about their reputation to be involved with this group and strongly encourage all players – for their own safety – to avoid any casino that is prepared to stock their games. As such we will be Blacklisting all casinos in ‘The Bad’ section and any other property that continues to offer games from this supplier.
Time Line for important CasinoMeister posts
Update – June 2015
We have recently been challenged by a GTECH associated venue, with a weak understanding of the course of events that lead to our Blacklisting of GTECH, to justify the opinions and evidence that our position was based on. As such I spent some time going through the 2 significant CasinoMeister threads and extracting the important posts. The list of threads below is in rough chronological order and provides notes on the specific information that can be drawn from each post.
It should be noted that there the 2 main threads represent around 90 pages with approximately 1800 posts. There are a massive number of high quality posts and a large number of credible posters making valuable contributions to the discussion. Anyone really looking to challenge the conclusions drawn from this issue should ensure that they are familiar with the entire conversation. While this list focuses on high profile groups/people participating in the conversation, there are a large number of other posters expressing dissatisfaction with the way this issue was managed that are significant in their own right.
The United Kingdom of Great Britain and Northern Ireland sits to the north-west of mainland Europe. The United Kingdom shares a border with The Republic of Ireland and has coasts on the Atlantic Ocean, Celtic Sea, North Sea, Irish Sea and English Channel. The population of the UK is approaching the 67.6 million mark leading to a fairly densely populated land mass. The gambling sector in the United Kingdom is entirely regulated and licensed by the UKGC – the United Kingdom Gambling Commission. Should players resident In the UK wish to gamble with foreign based operators there is no history of this being treated as a criminal offence, but high levels of protection exist for UK residents playing with UK licensed operators.
The above information is what we believe to be the the legal status of online gambling, however information on this topic is limited and hard to find. We accept no liability for any errors or ommissions. It is the reader’s responsibility to ensure that they know the legality of online gambling in their country before engaging with any online gambling service.