If I compare the screenshots then there was roughly (very) a 350 BTC profit for the site with 13,000 BTC wagered. That is a house edge of 2.7% in about 150.000 bets. What I would like to know is how many bets would be enough to be statiscally significant? What I mean is, do these kind of swings in the house edge happen over those number of bets or do we need millions to converge to 1% or even more? Dooglus, any idea?
Here are the last 2 data points I have:
Wed Sep 17 07:42:31 PDT 2014 profit: -297.32139723 invested: 1,531.01564543 wagered: 92,186.6599 bets: 51,133,608 BANK: 944.47462087
Wed Sep 17 16:43:29 PDT 2014 profit: 326.75800395 invested: 1,658.95634765 wagered: 121,997.1963 bets: 51,810,085 BANK: 1,696.31436053I don't know how to answer your questions, but I can tell you that Just-Dice had over a billion bets and ended up with a profit of less than 0.4% of the total amount wagered even though the house edge was 1% throughout.
So even a billion bets isn't enough to get the actual profit close to the theoretical profit.
Part of that is because a huge percentage of that billion bets are tiny in comparison to the big bets. Another part seems to be that the "large" in the "law of large numbers" is bigger than you might naively expect it to be.