In this day and age of AI, I think it's worth trying to build a better economy simulator without having any existing economic models in mind. Addressing this from a programming point of view could lead to interesting results.
There’s no consistent way to make money. You can exploit small details about how each market works, but in a short time everyone will notice and copy you. It’s like a market without patents. People talk, the trade ideas spread. Like wildfire. Too fast even for a quant to make sense - good managers can tell if a piece of info is hot. No manager will ever be able to tell whether a stock will outperform.
The objective is to maximize profit per unit of risk taken. If hedge fund made 20% less than market but took 50% less risk, that's a MASSIVE value proposition.
(now, one could argue they fail at that too, but that's more difficult to prove)
It’s just really hard to get precise predictions. Really hard. Not like settling P=NO hard, but like moto perpetuo hard.