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
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
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.
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.
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.
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.
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.
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.
7 comments
[ 3.8 ms ] story [ 37.2 ms ] threadyou 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
http://www.techcrunch.com/2007/01/02/use-weatherbill-to-bet-...
Not sure how it's going for them, though they haven't been deadpooled.
Local. I don't care about the temperatures all across the Midwest. I care about the weather where I am right now (Boston).
> short-term or long-term?
short-term, but longer than a couple of hours -- upcoming weekend would be nice.
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.
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.
If you don't have that, why bother?