Kristian Fagerlie reviews his developing forecasting system’s estimates for the US midterm elections, comparing its projections for Senate control and individual races with prediction marketsA prediction market lets participants trade contracts whose value depends on the outcome of a future event.. He highlights where its estimates differ and plans to assess them against eventual resultsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria..
He asks the model to estimate the next-day S&P 500 response under alternative Senate-control outcomes. The demonstration produces probability distributionsProbabilistic forecasting describes possible future outcomes using probabilities or a probability distribution. It expresses uncertainty rather than giving only one predicted outcome. rather than certainties, and Fagerlie describes the reasoning presented by the system.
The demonstration is an early test of his own forecasting startup. The election and market outcomes were still unknown when recorded, and the examples do not establish predictive accuracy or a repeatable trading edge.
Watch on YouTube
