To decide who wins the bet, the scientists have agreed to compare the average global surface temperature recorded by a US climate centre between 1998 and 2003, with temperatures they will record between 2012 and 2017.

If the temperature drops Dr Annan will stump up the $10,000 (now equivalent to about £5,800) in 2018. If the Earth continues to warm, the money will go the other way.

Maybe there are details missing in the article but it seems like there should be a significance threshold. How likely is it that a rise (or fall) of x over a period of time is due to chance alone?

I still like the idea of a small series of contracts. Aside from “rolling” transaction fees (which might be counteracted by tighter spreads) and volatility, betting that the average temperature will rise is basically the same as always being long the front contract.