Mail   Mail   Mail  fortank_facebook  fortank_twitter 
Research Seminars in USA Oct 2018: DePaul Chicago, Univ Illinois Urbanah-Champaign, Uni Wisconsin - Madison: Looking for the needle for the needle in the haystack: Evidence of the superforecasting hypothesis when time and samples are limited. PDF Print E-mail
Announcements - forLAB Announcements

Prof Kostas Nikolopoulos

Research Seminars in USA Oct 2018: DePaul Chicago, Univ Illinois Urbanah-Champaign, Uni Wisconsin - Madison: 

Looking for the needle for the needle in the haystack: Evidence of the superforecasting hypothesis when time and samples are limited

Abstract

The success of the Good Judgmental Project in harnessing the power of superforecasting naturally leads to the question as to how one can implement that approach on a smaller scale with more limited resources as in less time and fewer participants. Small(er) corporate environments and SME-type decision structures are prime examples where the modified superforecasting approach can be used. In this research we focus on a hybrid approach of judgmental forecasting on special events where we combine training of superforecasters-to-be via the concept of a modified version of structured analogies (s-SA), a staple of judgmental forecasting in the literature. We call the resulting approach structured superforecasting and illustrate its efficacy over samples of participants from the wider public sector and the academic community. In particular, with a proper experimental design that includes a training and a control group, we apply the above methodology and compare performances. More importantly we do find evidence of the superforecasting hypothesis even when we are working with smaller samples – a few hundred experts - and when the selection of super forecasters needs to be done much faster – in less than a year.