A team from the University of Oxford’s physics department has claimed the first prize of £10,000 in the Climate Prediction Market competition, which saw academic teams attempt to predict monthly UK temperature and rainfall over six months from April to September 2018.
Other Oxford teams also took second and fourth place, with a team from the London School of Economics in third. The top 10 performing teams shared a combined £55,000 of cash prizes that went towards their institutions, and also included entries from Durham, Edinburgh, Leeds and Bath universities. In total, 24 academic teams from 10 UK universities participated.
The project’s lead scientist, Mark Roulston, said, “The competition has been a great success. It shows that with a well designed market and the participation of engaged experts you can generate quantitative consensus forecasts, including information about uncertainty. It will be interesting to see in what other areas this method for organising expertise can be applied.”
The Climate Prediction Market competition saw 20,000 trades completed, and coincided with the UK’s hottest May on record, and second hottest June and July.
Prediction markets are markets that are explicitly designed to reveal and aggregate information. They can be used to produce consensus forecasts based on the judgments of multiple experts. The market generated joint probability distributions for rainfall and temperature, incorporating information about the relationship between them, rather than simply producing distinct probability forecasts for each variable.
The Climate Prediction Market’s design could itself be applied to longer-range climate prediction, as well as non-climate phenomenon. In June, Winton spun out the company Hivemind. One of Hivemind’s products is a platform enabling organisations to run their own prediction markets.