the best market creators look more like editors than traders. a new study analysed 802,863 resolved prediction markets to understand what determines whether a market attracts volume. the strongest signal was not sophisticated wording or complex market mechanics. it was the topic. using only information available when a market launched, researchers could explain 74.7% of the variation in its eventual log trading volume. what the market was about contributed far more predictive value than its wording and resolution rules. but there was an important catch. when the model was trained on older markets and tested on later ones, its ability to identify markets that would cross $1,000 in volume fell sharply. 76.9% of the later markets used question formats the model had never seen. demand moves faster than historical templates. good market creation is therefore partly editorial judgment: understanding what a community cares about now, which events are creating genuine disagreement and whether the moment will still matter by the time the market exists. no central listing team can possess that context for every football club, creator, game, token or niche community. the people already inside those conversations can. that may be the strongest argument for permissionless market creation: not simply that anyone can create a market, but that the people closest to emerging demand can create it while the demand still exists.