Simple example:
Country X decides to do something dastardly against the interests of Futarchy Y. Confidence in the economic outlook wavers, as happens in any country where there's an anticipated reduction in trade and maybe even war. Country X, could have made a lot of money shorting "future GDP per capita" on the Futarchy prediction exchange prior to their dastardly act though.
It gets worse because not only is Country X able to directly profit from making the lives of Futarchy Y worse, but the futarchic system actually makes retaliation impossible. Much as the government of Futarchy Y talks tough, and would like to take action, like sanctions, but when they actually put their sanction policy to the test, these proposed sanctions probably have a negative impact on "future GDP per capita", which is one of a perfectly reasonable set of metrics Futarchy Y has chosen. Despite the man on the street tending to agree that sanctions would be a just and moral cause of action Futarchy Y is forced into retreat because a priori metrics matter more than what the government and the public actually want to happen, and there's no way the market agrees that sanctions are going to help them achieve the given goal of growing GDP.
Smart people outwit crude algorithms in simple games like Go if they have an infinitude of possible outcomes. Now imagine that crude algorithm is calibrated by simpletons voting on parameters they don't really understand