The problem with forecasting is how its done. Much of forecasting has less to do with the actual material explanations of events and more with observed historical conditions and probability.<p>A 50% chance of rain means "in all of recorded history, when conditions were similar, it rained 50% of the time," which leaves a lot to be interpreted by meteorology advancements. It does not mean that the "rain trigger" has a 50% chance of being pulled, since as we know, past results do not affect future outcomes in a pure probability, and the past conditions can only be correlated to rain, not directly tied to causation.<p>Also, I think there's a sweet spot somewhere in between a couple of hours and hundreds of years. Maybe around a month would be good to start.
Don't worry about the monetization here, worry about the forecasting! It's still an unsolved problem. 3 day forecasts on local/national news aren't that accurate. Do better and your technology is worth a lot of money.
um assuming its good enough... if you could predict 1 yr in advance and sell the knowledge to brides to be for a couple hundred a pop... you could make quite a killing<p>you could even offer a money back guarantee... and if the probabilities and prices line up you dont even have to be that good at forecasting
> local or global?<p>Local. I don't care about the temperatures all across the Midwest. I care about the weather where I am right now (Boston).<p>> short-term or long-term?<p>short-term, but longer than a couple of hours -- upcoming weekend would be nice.
If you've got a good algorithm that can predict local weather 3-5 days out with a significantly better accuracy than the other players in the industry, you won't have any problem monetizing it.<p>If you don't have that, why bother?
If you can do medium-term (over the next year) then consider trading weather derivatives. Such contracts pay out on eg the number of warm days being higher/lower than consensus.