Consider the following imaginary lottery (motivated by Massachusetts' Cash WinFall):
Ticket money obtained in the previous draw are given to the participants of the next draw proportionally to the ticket price they pay.
For example, suppose that in the first draw people bought tickets costing 5 million dollars, and in the second draw - costing 1 millions; if I bought a 1 dollar ticket in the second draw, then I win 5 dollars; and that's for sure, no minuscule probabilities, cool! People will like it and buy more tickets in the third draw, say 10 millions, and win 10 cents for a dollar spent, that is lose 90 cents overall; in the fourth draw almost nobody will like to take part, people buy tickets costing only 0.5 million, and win 20 dollars for one dollar spent, cool again! And so on...
Of course, the organizer will have to take a share, say 10% of ticket price, but it seems that it does not really change the mechanism, that is ticket money fluctuating up and down.
Such a strange lottery looks like something from a stock market - so can one really model it there?
P. S. English is not native to me, so please feel free to correct the question.