Announcing the Sept 2020 Covid-19 prediction market settlement

Mark Roulston, Senior Data Scientist •

Blog 15

We have been working in conjunction with Lloyd’s Lab to investigate the use of prediction markets—using AGORA, our prediction market tool—as a method for combining expert judgements into consensus forecasts relevant to the insurance industry.
Some of the experimental prediction markets we have been running aim to predict the number of newly confirmed cases of COVID-19 in the US for each of the remaining months of 2020.
Our call for experts to participate in this prediction market drew a variety of expertise from areas such as statistics, mathematics, epidemiology, infectious disease modelling, other areas of medicine and general data science.

On Thursday 1st October, Johns Hopkins Coronavirus Resource Center published the number of new cases in September which meant that market participant Matthew Chantry, Postdoctoral Research Assistant in Physics at Oxford, finished the month having made the largest single contribution to the forecast in terms of accuracy and timeliness.

We asked Matthew what had interested him in taking part in the market, if he had any tips to share and if he thought his particular area of expertise was useful when predicting particular outcomes.

What was it that drew you to participate in the market?
I had participated in similar markets modelling other phenomena in the past and found them to be a fun modelling challenge for my spare time.

What is your background, and which of your skills and experience did you find most helpful when participating?
My background is in mathematics and computational fluid mechanics. My skills in understanding the mathematics of the market and how to build probability density functions were particularly helpful.

What was your approach when participating in the market?
The market rewards participants trading towards their belief distribution. Buy when the price is too low and sell if the price is too high (even if the event is likely).

What external sources did you utilise to form your judgements?
Just the freely available forecasts that can be found online.

Do you have any tips or insights for other participants in the remaining markets?
Yes, trust your probability distribution, and don't follow the herd!

Matthew and the two runners up will receive cash rewards for their participation, courtesy of Lloyd’s Lab. If you’d like to know more about how these types of markets might be used for your own prediction problems, or are interested in taking part in any of the upcoming markets, you can find more information and a participant application form here.