Pollyvote - Forecasting the US Presidential Election

Main Menu

Search

Polly did it again!

Polly's final forecast: Obama 53.0% vs. McCain 47.0%

After her success in predicting the 2004 Election, Polly once again proved the value of combining forecasts.

From the beginning (August 30, 2007), Polly never strayed from the Democratic candidate.

By aggregating forecasts from traditional polls, a panel of American politics experts, a prediction market and currently 16 quantitative models, Polly provided a daily updated forecast of the Republican two-party vote. The table below shows Polly's final forecast as well as the respective forecast of each of her components:

Last updated on	Components	Republican Two-Party Vote
Nov 3	Pollyvote	47.0
Nov 3	Polls (damped)	46.8
Nov 3	Expert Survey	47.5
Nov 3	Iowa Electronic Markets	46.7
Nov 2	Quantitative Models	47.0

The PollyGraph below shows how the Pollyvote forecast has developed over time (click here for more graphs)

Following the principle of combining forecasts Armstrong (2001), Polly provided near-perfect predictions for the 2004 election (Cuzán, Armstrong & Jones 2005).

As Larrick & Soll (2006) demonstrated, most people do not understand that combining leads to substantially more accurate forecasts. As a result, combining is not used nearly as much as it should be for forecasting in business and government.

Pollyvote is part of the Political Forecasting Special Interest Group at forecastingprinciples.com, a resource for scholars and practitioners interested in forecasting elections and other political events. It is supported by the International Institute of Forecasters.

What's New

Ray Fair's post mortem

On his homepage, Ray Fair analyzes the performance of the presidential vote-equation. He also links to a new paper, forthcoming in the American Journal of Political Science.
Tuesday, 18 November 2008