How do you know if something is a good bet?

In the short run it's always unclear. Probabilities suggest that evidence or proof is pretty thin in the beginning.  Good ideas can struggle.  Bad ideas can succeed.  If we focus too much on the short run performance we run the very real risk of losing the signal of truth.

The only real way to tell if the idea, product, team, or business is a good bet is to observe performance over a period of time.  This requires patience.  Patience when building anything, especially in the pressure-cooker of a startup is hard to come by.

Above average performance over a long period of time, is always the result of structural reasons.  And structural reasons make it possible to predict future performance over the long run.  We are an impatient bunch.