An election prognosticator who previously predicted a landslide victory for Vice President Kamala Harris has shifted her prediction to former President Donald Trump.
Thomas Miller, a data scientist who accurately predicted the outcome of the 2020 presidential election, previously backed Harris to win in 2024, stating that Newsweek In September, Democrats were on track to win more than 400 votes in the Electoral College, according to their model.
The model, which relies on betting odds as opposed to aggregate polling data, is now saying Trump is the favorite, with its most recent prediction that the Republican could win the Electoral College by 345 votes.
Newsweek has mapped out what a possible 345 Electoral College win might look like for the former president. In that scenario, he sweeps the South and Midwest, winning all the states where Harris had only a single-digit lead, according to FiveThirtyEight.
The model first began predicting a Trump victory on Oct. 7, when betting odds began to shift toward the former president.
Miller’s model prioritizes betting odds over polling data and also uses historical trends from previous elections to balance the data with what he calls “fundamentals,” which other models do.
For the 2024 election, Miller is using betting site PredictIt to fuel his model, which was the same data he used to make his successful 2020 prediction.
However, the data and underlying principles are predicting different results. “Technical and fundamental analysis are not congruent,” Miller wrote on his website Oct. 20, as historical trends point to a Democratic win, contradicting betting odds that Trump will win.
This matches what fellow pollster Alan Lichtman has also said about the election. Unlike other forecasters, Lichtman’s “Keys to the White House” model relies almost exclusively on long-term historical trends, and has consistently predicted a Harris victory.
Although polling still indicates the race is tight, the betting odds are in Trump’s favor, with poll market odds giving Trump a 60.3 percent chance of winning nationally, compared to Harris’ 39.7 percent. Harris is shown leading the way.
Although betting odds have historically consistently predicted the correct winner of a presidential election, they are not based on representative statistics like polling and are therefore more prone to bias.
Miller said Newsweek That he corrects for potential biases in his model, which currently favor Trump.
“There is a Republican bias in the forecast markets,” Miller said. “We estimated the degree of Republican bias in 2020, and we currently correct for that bias. What is uncertain, however, is the level of Republican bias in forecast markets in 2024. Highlight the degree of bias in 2024.”
Miller’s recent forecast actually indicates a slight improvement for Harris. On October 26, the model indicated that the vice president would receive only 171 Electoral College votes, the lowest result for a candidate since Republican nominee Sen. Bob Dole in 1996.
“The 2024 race for president has gone from tossup to Republican landslide, tossup, possible Democratic landslide, tossup, and now possible Republican landslide,” Miller said. “With less than two weeks to go until the end of voting on November 5, could the prediction markets and related election forecasts change again? Yes. We expect to see increased trading and higher volatility in the final week of the race. expect.”
Of the 11 presidential elections since 1980, the only race where the winner had worse odds than the loser was in 2016, where both the betting market and traditional polling failed to predict a Trump victory.
Update, 10/30/24, 11:29 am ET: This article was updated with additional information.
Do you have a story that Newsweek should cover? Do you have any questions about this story? Contact LiveNews@newsweek.com.