The chart looks really bad. Though i dont understand why the chart at bitcoinproject is working so fine.
bitcoinproject started collecting data after nakowa's lucky streak. His 'expected profit' is calculated from the day he started collecting data.
at least there should be a pressure to the top... but that the bets from nakowa are "forgotten" by luck... i dont know what to think.
If you mean the 37k 'gap' should tend to close to zero, then you're mistaken. The law of large numbers says that the profit percentage will get arbitrarily close to 1% not because the gap gets smaller, but because the profit and expected profit get larger. The 37k gap stays at 37k, but 37k becomes less and less significant. When the profit is 100 million BTC, a 37k gap won't be noticed, and the profit will be 1% of amount wagered. Turns out "the long run" is longer than we thought...
Isnt the site profit calculated solely from profit divided through house?
It is simply the profit divided by the amount wagered.
Currently we have a profit of 12,941 and a wagered of 4942k. 12941 * 100 / 4942e3 = 0.26185%
But for example when JD runs with 1000 house and makes 10 profit it would mean 1% profit. But if you add 30000 to the house
The size of the bankroll isn't relevant. A bigger bankroll allows us to take bigger bets; that's all. The profit is calculated as how much we kept from how much was wagered.
That would mean that your 3000 more fee would have been earned and taken from the investors that were invested while nakowa played. The fee would have hit the investors very hard additionally i guess. 70% house crash? Wouldnt the theoretical profit that high that any investment would be wiped out with the fee then? Probably even going into negative then...
If the fee was taken per-bet, the bankroll would have reduced quicker than it did, and so would the max profit. Investors wouldn't have been wiped out, or had their investment go negative, but it would have cost them more. The new scheme of charging investors 0.1% per bet instead of 10% on profit costs them more whenever the house does worse than expected, and costs them less whenever it does better than expected. When the house does much worse than expected, like at the end of September on nakowa's biggest day, it costs them a lot more than the current system (which costs them nothing if they stay invested, since they only pay on profit; interesting aside: that was my most profitable day by far, ever. So many investors were 'day trading' the massive swings that I took huge commissions when they managed to divest at the top)
The house was crashed 70%? And the graph Peter R posted supports that it was quite normal? At 1% kelly, which means nakowa only could win max 1% of the house at one bet means that he had to win more than 70 times at full profit. And that should be a quite normal happening? Of course im not a genius in math so i might think wrong here somehow...
I'm not sure where you're getting that 70% number from. At no point was the bankroll depleted by 70%, even including people divesting. Nakowa bounced the house profit between +7k and -7k (roughly) on his biggest day. A swing of 14k. The bankroll was around 50k or 60k at the time. 14k is 28% of 50k. So at most he took 28% of the bankroll. Bear in mind the profit didn't start at +7k and end at -7k. Those are just the extremes of the profit through the day. So he took less than 28%. And that doesn't mean he won 28 times in a row. He played hundreds of bets over a very long session. Also, I don't think Peter's graph suggests his result was "normal", just that it wasn't exceptionally unlikely.
Really? I thought implementing risk settings would be impossible because the server would have to calculate too much. I thought till now you calculate the profits only when divesting or weekly and that the profit shown at the website is some javascript running on the users computer that is guessing the profit from the bets.
Then the server load wouldnt be an argument against risk settings anymore too...
I don't do per-investor calculations after each bet. I just recalculate the size of the bankroll after each bet. I know each investor's share of the bankroll, so can calculate their investment size when I need to. Charging a fee per bet is simply a case of adjusting the bankroll calculation I already do. It doesn't introduce a per-investor calculation per bet. Variable rate investment DOES.
I hope i can use JD with HMA again and its only temporary. But the nakowa thing, the luck he had to have and that the profit isnt moving back makes me think. I dont want to bring back a "maybe" old discussion but i dont see how thats possible. Practically winning 70 times in a row?
* What's HMA?
* We don't expect the gap to close; we just expect it to become a smaller percentage of profit as profit increases.
* Why 70?
