5 ms·

Ask YC: How self-made weather forecast can be monetized?

3 points by nelud 18y ago

assuming it's good enough. Which forecast is better - local or global, short-term (couple of hours) or long-term (hundreds of years)?

utnick 18y ago
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
- JacobAldridge 18y ago
  I read about a company doing this on TechCrunch last year http://www.techcrunch.com/2007/01/02/use-weatherbill-to-bet-on-the-weather/ 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.
rms 18y ago
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.
aneesh 18y ago
> local or global? 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.
dkokelley 18y ago
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.
michael_dorfman 18y ago
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 don't have that, why bother?
jonp 18y ago
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.

3 points by nelud 18y ago

assuming it's good enough. Which forecast is better - local or global, short-term (couple of hours) or long-term (hundreds of years)?

utnick 18y ago

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

JacobAldridge 18y ago
I read about a company doing this on TechCrunch last year http://www.techcrunch.com/2007/01/02/use-weatherbill-to-bet-on-the-weather/ 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.

rms 18y ago

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.

aneesh 18y ago

> local or global? 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.

dkokelley 18y ago

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.

michael_dorfman 18y ago

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 don't have that, why bother?

jonp 18y ago

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.