How do the best forecasters update their beliefs? Do markets or prediction polls elicit the best forecasts? In this week’s episode of GGWP, Pavel Atanasov answers these questions and discusses his postdoc experience at Good Judgement under Phillip Tetlock and Barb Mellers. We also talked to Pavel on the strengths and weakness of machine forecasting models and different ways to handle median-spamming behavior on prediction platforms such as Metaculus during aggregation.
Pavel Atanasov is the co-founder of Pytho.io–a boutique R&D shop which uses decision science to improve predictions and decision making. Pytho’s current focus is Human Forest, a double-NSF award (patent-pending) project which combines data-driven base rate automation and collective human insight to deliver on key objectives on which machine algorithms and human forecasters can fall short. Pavel has a Ph.D in Psychology and Decision Science from The University of Pennsylvania and is the author on numerous forecasting, judgement, and behavioral data science papers.
If you want to learn about the intersection of research, entrepreneurship, and forecasting, this is the episode for you.