Average bet size doesn't really say much because one highroller betting 20 coins amounts to 20,000,000 minimum size bets,
ah, yes. I see that now.
Profit has actually been higher than expected value and the bigger the biggest bets are, the bigger likelihood of the actual profit differing from the EV. For expected profit just put total bet amount x 0.019... Total wagered is available in the global stats API call.
the total wagered from that dicesites.com is clearly wrong given it is little more then the amount won (they apparently show total won + profit; it ignores the total amount that was bet and lost (except for the profit)) (plus it shows the data available on your site that starts in 5/11/14, yet gives the start date of 18/04/12; which from an investor POV is deceivingly very unflattering; it would make one think that the wagered and profit of the last year and 3 months was actually spread out over the life time of the site; and thus RoI per year would be much less (roughly 1/3)).
the total wagered can easily be calculated from 2 stats on your site, total won and profit: as I explained before, "total wagered=total won+total loss=total won + (total won + profit)= ~202342" (currently is about 203518); this is because obviously total wagered is total won + total lost (there are no ties correct?) and total lost is equal to the amount that was won plus the profit cause that's the only way the profit could be what it is. you understand this, yes?
thus: total bet amount x 0.019 is 3866.842. 1586 less then the actual profit of 2280, like I said actual profit is about 60%(59%) of expected.
anyway, maybe this is perfectly reasonable variance IDK; although if I were you as a casino I would like to know what the variance and standard deviation is for the games on my casino.
Where do you get the total loss from? Of the total wagered, some rolls win and some lose. From what has been wagered, 0.019 is expected to be taken off from bet payouts as profit. We provide the total wagered amount and house edge is accomplished by taking a sum off winning rolls' payout.
Payout: ( betAmount * (65535+1) / numberToRollUnder ) * ( 100 - 1.9 )
Total bet amount as seen in the API: 104469.02610966
104469.02610966 * 0.019 = Expected house profit = 1984.911
Actual profit: bets 104469.02610966 - wins 102119.66460453 = 2349.36150513
From the house's perspective, the EV is the same no matter the bet size, but players can place more bets with low probability before running out because profit is taken from payouts and if payouts occur less often, it takes longer for the player to run out of funds (on average).
I also see that the current stats include (or ignore) slot game rounds as session game rounds. EV on those is slightly higher for the house, and they are credited to investors. I will make the global stats API call give more data about this.