Predicting the outcome of the US presidential election is hard. There are dozens of polls, none of which seem to agree with the others. Even betting markets, which are at least as predictive as scientific polls, list different prices.
To get around this problem, election forecasters combine poll results, betting market prices, as well as economic and historical factors to produce predictions about who will win the presidency. But even the forecasters don’t always agree. Often, they assign different odds to the candidates; sometimes, they even predict different winners. So how do you know which forecast to trust?
One approach is to trust none of them individually, but to trust all of them collectively. I’ve selected some of the leading election forecasters — the Upshot, FiveThirtyEight, Daily Kos, and PredictWise — and put their predictions all on one chart that shows you where the race stands and where it is heading.
Which forecasts are we tracking?
Daily Kos — The Daily Kos presidential forecast combines public opinion polls and three other variables which appear to have had an affect on previous presidential election outcomes. Those are (1) the incumbent president’s net approval or disapproval rating in June of the election year, (2) the percent change in America’s gross domestic product from the first to the second quarter of the election year, and (3) whether the incumbent party has held the presidency for two or more terms. Read more about the Daily Kos’s methodology.
FiveThiryEight — FiveThirtyEight (the name comes from the number of electors in the Electoral College) produces three general election forecasts: one based only on the polls, another based on the polls and other factors, and a “now-cast” showing what would happen if the election were held today. I have shown only the polls-based forecast here. FiveThirtyEight rounds its predictions to the nearest one-tenth of a one percent. For the sake of simplicity, I have furthered rounded it to the nearest percent. FiveThirtyEight also includes a prediction for Libertarian candidate Gary Johnson, whose chances are negligible and who has been removed from this page. Read more about FiveThirtyEight’s methodology.
PredictWise — PredictWise’s general election forecast is based not only on public opinion polls, but also on betting markets, where “investors” can buy and sell “futures” in candidates, the prices of which are determined by market activity. Such betting markets have been shown to be at least as predictive as polls. Read more about PredictWise’s methodology.
The Upshot — The Upshot is the New York Times’s data blog. Its general election forecast is mostly based on hundreds of state and national polls. It also incorporates, to a lesser extent, the voting history of every state. Read more about the Upshot’s methodology.