RE: sweetspot responsiveness - one issue is that market capitalization is cumulative to the number of bets, and the number of bets rises very quickly as the time horizon on the prediction approaches. A $1million bet (I wish) 24 hours before prediction close will barely budge it, but find a new prediction six weeks out on which you feel reasonably informed and that same bet pushes it 30%. (Which has two immediate effects - people immediately bet against you and the market recognizes your bet at 80/20%, limiting your payout if correct.)

Lately, the rising number of overall predictions has made it easier (for me) to keep money in the market spread into smaller bets, but the lag between stated circumstances/deadline and judging (which I've noticed is greatly improving) often has me cashing out ahead of time.

One possible solution is an escrow system or other limits on money movement over time, or maybe a maximum bet on any one prediction equal to average net worth (which would also limit the geometric growth of the networth leaders). The option for incremental cash out would help the situation.

ETA: The pool of money absolutely grows through time, both because of new users and because even the most obvious bet (recent Google stock predictions, for example) at the last minute will have a 2-5% yield. The new user at 100k is now about 10% below average?