New prediction website
In our last newsletter, we pointed to a recent post by Tim describing Seven Guidelines for Better Forecasting. One of the main point in that post was the critical role of accurate tracking of predictions using a mathematical approach known as proper scoring. In current work we are exploring how these concepts can be applied to important problems such as the evaluation of fund managers. The key idea is that by using proper scoring, we can more effectively separate genuine skill from luck, with implications for hiring and firing, bonuses, and the overall performance of a fund.
As part of this, in recent weeks we have created a website, RBA.Tips. RBA.Tips is like a footy tipping competition, but instead of tipping the outcome of football games, you tip what the Reserve Bank of Australia will decide about interest rates at their next monthly meeting. Once the decision is known, tips are scored using proper scoring, and a tipper's scores over time combine to form a skill rating. Tippers are then ranked by skill rating (though they can choose to remain invisible). We invite you to visit the site and participate even if you're not particularly interested in RBA interest rate decisions, because the main function of the site is to help people learn about skill in forecasting, probabilistic prediction, proper scoring, and potential applications of these concepts.
Melbourne Meetup, Tues 19th May, 5.30 On Tuesday the Insights Guild of Asia-Pacific is hosting a meetup in Melbourne CBD, with the title Crowdsourcing Insight in Forecasting. Tim will be leading the discussion, and will be using RBA.Tips to take a close look at the interplay between analytics and expert human judgement in a new platform for forecasting key economic events. It would be great to see any of our Melbourne-based network there.