Election

Do averaging models, markets, and odds produce more accurate predictions?

Elections / / , , , , , ,

In judging state-level forecasts of the 2020 U.S. Presidential Election, we’ve primarily looked at prices from one prediction market (PredictIt), outputs from two forecasting models (FiveThirtyEight and The Economist), and the implied probabilities of gambling odds from three British bookmakers (Sky Bet, BoyleSports, and Betfred.)  Because diversity in predictions can benefit ensembles, I was curious […]

Do averaging models, markets, and odds produce more accurate predictions? Read More »

Do gamblers see NH Senate race moving towards GOP?

Elections / / , , , , ,

When we looked at the betting markets a few weeks ago, gamblers were eyeing Republican wins in 11 of the 12 Senate races where both PredictIt and SBK were taking action.  In the past week, gamblers seem to see the 12th race – New Hampshire – as moving towards the GOP. PredictIt, Smarkets, and SBK

Do gamblers see NH Senate race moving towards GOP? Read More »

How did FiveThirtyEight’s model perform over time?

Elections / / , , ,

We previously looked at Brier scores for The Economist’s 2020 Presidential model predictions over time, and surprisingly found the scores worsened closer to Election Day.  Given the closeness of their predictions, I assumed FiveThirtyEight’s Brier scores would trend the same way but was wrong again. In the chart below you can see Brier scores for

How did FiveThirtyEight’s model perform over time? Read More »

PredictIt and SBK forecasting Republican Senate gains in 2022

Elections / / , , , ,

PredictIt is a U.S.-based prediction market and SBK is a British sports betting app, but both are indicating that political gamblers see Republicans making gains in the U.S. Senate this year (at least as of Friday, February 4, 2022).  The probabilities in the 12 Senate races where both platforms are currently accepting action are extremely

PredictIt and SBK forecasting Republican Senate gains in 2022 Read More »

Did The Economist’s 2020 Presidential forecast model perform worse closer to Election Day?

Elections / / , , ,

As discussed in an earlier post, as of the morning of Election Day 2020, The Economist’s presidential forecasting model had a slightly higher Brier score (lower = better, higher = worse) than FiveThirtyEight’s model, but each model correctly predicted the outcome in the same number of states (49 out of 51).  This made me curious

Did The Economist’s 2020 Presidential forecast model perform worse closer to Election Day? Read More »