That's where the analogy starts to fall apart then. Because the variance in those decisions is not very similar, since you're sampling very different underlying distributions. And estimating the priors for a problem like "what is the optimal arrangement of tables to maximize throughput in a cafe" is very different from a problem like "what is the current untapped/potential demand for a boardgaming cafe in this city, and how profitable would that business be".

The main reason why professional poker players are playing the long-game, is because they're consistently playing the same game. Over and over.