You are more or less correct.
The math model is different though. The key here is the increase of the deposits by 50% each cycle. In this case no payments are missed. if it does not go like that - system slows down waiting for new funds.
Thanks dooglus. Kind of a silly miscalculation there. It also shows that the system will bust really fast after a few initial returns(unless profits are reduced).
I have updated the Excel Sheet to also reflect(scroll down) how the payouts will stop happening or happen at less intervals even if people continue investing after that.
It will stop/ paid out in less intervvals after a few more rounds of returns(unless the subsequent investments are much bigger than 100).
Also, looks like the current cycle has come to an end with only 0.3BTC left in the address, and has paid out around only a 50%-60% back who deposited on 9th July. Payouts will now happen only if the balance goes above 0.3 BTC, and will probably happen after bigger intervals(in the case more deposits occur).
The 50% increase is correct, in my earlier simulation it looked like the scheme needed two new Users to pay to an older User, but the two new Users, after deducing what needed to be paid, were left with more or less half of their initial deposit. In the real scheme they are not left with anything, as all the funds go to payments, but the math adds up. The number of the new participants needed every cycle may be impossible to reach, unless there is an ever-increasing interest in this operation.
http://s12.postimg.org/ymagrwogt/10_Level_Pyramid_v22.jpg