One of the big benefits SharkScope brings to online poker is ability to check for player collusion. The head to head statistics feature allows you to see how often 2 players play together and how their ROIs compare when playing with and without each other.
This feature has been used by many users over the years, who have seen a suspicious play and want some further evidence to prove their case and alert the poker sites to the problem.
The most interesting example of these came fairly recently. A user of ours who happened to be professional statistician spotted some strange statistics of a player who had made it onto one of our leaderboards ahead of him. The player's name was T049078 on the Cereus network and he had achieved an ROI of over 100% in 327 $20 ultra turbo 6 handed games. Our user then posted his suspicions on a forum (under the name Sharkscoper, which I admit is a big coincidence given the name of this site, but I assure you he is not connected to us in anyway other than being an unknown user of the site). His post was then promptly laughed out the forum which is sad indictment of how much the average poker player understands statistics - though a great sign for us trying to make a profit from our poker.
Although an average ROI of over 100% is possible for a good player in the larger tournaments, it is completely impossible for a 6 handed game as possible ROI goes down with the size of the game. In addition the faster the game, the lower the possible ROI as a good player has less time to make his poker advantage count. So for an ultra turbo game this was more than 10 times the expected ROI of around 10%. And yes, 327 is enough games to draw conclusions.
So how was this ROI being achieved? Clearly there was some form of cheating going on. Sharkscoper posted on a couple of the more intelligent forums, TwoPlusTwo and PocketFives, and started getting a few more helpful comments. Initially of course, the shout was for it being a super user (given the history of Absolute Poker and Ultimate Bet this is now the default cry for pretty much anything on Cereus), but there is simply no way someone with a super user account would waste their time playing $20 6 handed games to earn a measly $7500.
It was then spotted using the SharkScope head to head statistics feature that T049078 had played a disproportionate amount of games with another user DRAGONEN8.
Clearly there was some form of collusion going on. But things still didn't add up as T049078 also had a huge ROI in games not played with DRAGONEN8. Then the plot thickened as forum posters started checking their own hand histories for games played with T049078 and discovered highly suspicious plays between other users and T049078. Again the use of the SharkScope's head to head statistics feature confirmed these users were part of a collusion ring with T049078. As time went on the number of confirmed colluders kept rising. By the end a staggering 17 colluders were found and you have to assume there were plenty more out there. In virtually all games containing T049079 had at least 2 other colluders present.
This is by far the biggest collusion ring I have heard of being uncovered. Normally in collusion cases it will simply be at most 3 players, colluding at the same table, but this was something on a bigger scale and clearly quite special.
Cereus naturally launched an investigation and promptly confirmed that these users were in fact colluding and to their credit began refunding the players who had lost out due to this collusion.
So is that the end of it? Do we fully understand what went on here? The answer is no, there is still some investigating to do to see what was going on here. For starters, T049078's ROI was only 100%. So if he was also providing the stake for two other players in the tournament who were shown on inspection to have an average -50% ROI in those games, he was roughly breaking even every game. Some forum posters suggested that they could have had rakeback, but even with this and other benefits this would not have been particularly profitable.
Furthermore if you are colluding with another player surely you'd expect this advantage would mean that you would more than break even on each game. The fact is they were having a net neutral effect on the game, and therefore getting no benefit at all from the collusion. What kind of collusion is that? Are they just very bad at poker?
So at the end of day this wasn't a case of collusion against players and therefore had to be sophisticated money laundering scheme. One of the biggest problems online poker sites face is credit card fraud. A person will create a new account and deposit money with a stolen credit card. They then have to figure out how to get the money out of the account.
Most sites force you to withdraw to the same location as you deposited from, so this adds a little protection. The best alternate route you'd think is to lose that money to an account in good standing and then withdraw the money from that account. The difficulty of course comes when the Poker site learns that the credit card was stolen and promptly closes down all accounts related to the fraudulent one and that have played suspicious heads up games.
Repeatedly creating accounts in good standing to get the money out must be pretty difficult and time consuming. So this scheme was clearly was an effort to get round this. Presumably they thought, how could a poker site detect you losing to specific players in multi player games? And for a time they were correct.
The most interesting part of the whole story is that users spotted the problem, complained to Cereus and got compensated by Cereus, but in reality, based on the statistics the only real losers here were Cereus themselves. The colluders had a net neutral effect in the games they played in and so took no money from other players and Cereus would have lost all the money that was no longer in the T049078 account when they closed it and the refunds they paid out. So it was really quite generous of Cereus to provide these refunds (although politically they have no choice here as they are trying to restore their reputation and no one would have accepted that the colluders didn't benefit at their expense from what they were doing).
It must be very tough as a business to run a poker site, particularly the smaller networks, when you have to suck up these kinds of costs, and maintain constant vigilance looking for new ways that people are trying to steal from you. Presumably all the poker sites are rushing to add automatic detection (if they haven't got it already) for this new kind of laundering, so it too will be wiped out soon. People tend to always assume that collusion is the poker sites fault for not detecting it but as we've seen here it can be extremely tough and they certainly have the most to gain from eliminating it.
Makes you think that a lot of the delays we experience in receiving withdrawals from sites must be partially down to the extensive fraud checking they have to do. The longer they have your money, the more chance they have time for something fraudulent you did to come to light.
The moral of the story is that not everything is quite as it seems.
