2020

Comparing the different 538 models in U.S. Senate races

FiveThirtyEight’s model is actually three different official model versions built on top of each other with polling, fundamentals, and expert forecasts: Version Component(s) Lite Polling Classic Polling, Fundamentals Deluxe Polling, Fundamentals, Expert forecasts Polling: Polls are weighted based on their sample size, recency, and pollster rating.  (Pollsters are rated by methodology and past accuracy.)  In […]

Comparing the different 538 models in U.S. Senate races Read More »

PredictIt prices: Warnock 2020 v. Warnock 2022

Senator Raphael Warnock’s two U.S. Senate races in Georgia were viewed very differently by PredictIt traders. In the 2020 U.S. Senate special election, all candidates ran in a “jungle primary” on the day of the General Election, with a runoff vote scheduled for January 5, 2021.  The price for Democratic “Yes” contracts in the Georgia

PredictIt prices: Warnock 2020 v. Warnock 2022 Read More »

Did Yahoo Fantasy Football projections outperform the naïve forecast in 2021?

Anyone who plays in a Yahoo Fantasy Football league understands the frustration of relying on Yahoo’s fantasy forecasts.  The projected stats always seem to be wrong.  To quantify exactly how bad Yahoo is at projecting future performance, we’re going to compare Yahoo’s projected 2021 quarterback statistics to the naïve forecast of 2020 statistics.  Our hypothesis

Did Yahoo Fantasy Football projections outperform the naïve forecast in 2021? Read More »

Comparing Economist & 538 Models over time

We previously looked separately at both The Economist’s and FiveThirtyEight’s respective 2020 presidential models in the months leading up to the 2020 Presidential Election, and are now directly comparing their state-level predictions.  This is challenging because the models made different predictions.  The Economist’s model made forecasts in all 50 states plus Washington, D.C., while FiveThirtyEight’s

Comparing Economist & 538 Models over time Read More »

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

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 »