...
By Marni Rose McFall
Reporter
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
A data scientist from Northwestern University has gained attention for his unconventional election predictions by using betting markets rather than traditional polls.
Thomas Miller accurately forecast the 2020 presidential election. Now, he's back with a new forecast: that Vice President Kamala Harris will win against Donald Trump with a landslide. Newsweek spoke to Miller to find out more.
...
Read More on Politics
"I don't rely on polls. I rely on prediction markets," Miller said. "A political prediction market is a leading indicator of what will happen in an election. People are putting their money down, which means they believe something's going to happen in the future."
Miller believes that because polls typically survey small samples of the population and take several days to publish results, they don't capture shifting voter sentiments accurately.
"The poll is asking, 'Who do you want to win?' The prediction market says, 'Who do you expect is going to win?' There's a lot of research to justify that the prediction markets are more effective as forecasters than polls," he said.
"You can wait for the polls, or you can just look at the prediction markets... In fact, we have visibility of the market prices all the time. We could have watched that debate and seen, with every point in the debate, these numbers increasing."
"The prediction market is forward-looking. Just as the stock market is a leading indicator of what will happen in the economy, a political prediction market is a leading indicator of what will happen in an election."
...
What does the model say about the election?
Miller's analysis indicates that this election has gone from being a landslide for Trump—to a landslide for Harris.
"It's a landslide," he said. "It's going to be a landslide."
Miller's model began by predicting a victory for Trump. He said, "You can see early on, from June through July, certainly through the Trump-Biden debate, we saw increasing probability, higher votes expected essentially for the Republican ticket... This was before the ticket was even defined."
He then explained how the key moments in the election have influenced the model. "Biden withdraws. In the very day that he withdrew, you can see the jump here... And then Harris secures the Democratic nomination. And she selects Walls as her running mate. That's August 6... Since then, we're in Democratic victory territory."
Speaking about the impact of the debate on the prediction markets, he said, "The Trump-Harris debate, September 10... We go from September 9, near toss-up, to 400 votes predicted in the Electoral College. An increase of more than 100 votes in one day."
He explained that the debate and then Taylor Swift's endorsement of Harris saw a "double whammy" in the markets. "We go from September 9, near toss-up, to 400 votes predicted in the Electoral College. That's massive."
Speaking more generally about the election, Miller told Newsweek, "I've never seen anything like this. Never, never seen anything like it."
What is Thomas Miller's betting-odds model?
In 2020, Miller used PredictIt, the largest U.S. political betting site, to track the odds in all 56 voting jurisdictions.
He developed a model that converted betting prices into the popular vote and Electoral College projections. His predictions were strikingly accurate: he missed Joe Biden's Electoral College count by 12 votes, and in Georgia's Senate races, he was nearly exact, predicting both Democratic victories.
For the 2024 election, Miller is using PredictIt to gauge the contest between Harris and Trump.
His model, based on data from the last 16 presidential elections, shows a close correlation between the betting odds and the expected popular vote.
As of mid-September, PredictIt showed Harris with a 55 percent chance of winning, which Miller interpreted as 55 percent of the popular vote, translating into a significant lead in the Electoral College.
His model suggested Harris would not only win swing states like Michigan and Pennsylvania but also potentially flip traditionally Republican states.
Miller's forecast is based on the belief that betting markets are a better reflection of voter expectations than polls. If his model proves accurate again, Harris could secure a decisive victory in November.
Newsweek spoke to Eric Zitzewitz, a Professor of Economics at Dartmouth College. He said, "Prediction markets normally forecast the odds of a candidate winning the election, whereas polling provides a forecast of their share of the vote."
"But when people have made them comparable, by converting polling into a probabilistic forecast [like Nate Silver does] or running markets that forecast vote shares [like the Iowa Electronic Markets has been doing since 1988], prediction markets have performed as well or better than polls."
Zitzewitz added, "One advantage of prediction markets is that they are forward-looking. For instance, after Harris won the debate, the markets moved immediately in her direction, correctly expecting that post-debate polls would as well. The same happened, in the opposite direction, after the Trump-Biden debate."
"PredictIt has a market on the margin of victory in the Electoral College. Right now, its prices imply a 4 percent chance of Harris winning by 215 electoral votes or more [i.e., winning about 377 electoral votes, to 161 for Trump]. The prices imply a 2 chance of a victory by this margin or better for Trump. So the chances of even a modest landslide look small. Even in mid-July, before Biden dropped out, the odds of a GOP landslide were only 8 percent. Moreover, historically prediction markets do tend to overprice small probability events like this, so the true probability is probably a bit less than implied by the prices."
"Maybe of more interest is a sweep—a party winning the White House, House and Senate. Polymarket puts the odds of a GOP sweep at 32 percent but a Democratic sweep at 21 percent, despite having Harris as a narrow favorite to win the White House. This is probably because the Senate is likely to flip to the GOP, even if the Democrats win the White House."
According to the latest polls, Harris is currently maintaining her narrow lead over Trump, with polling aggregator FiveThirtyEight putting the vice president 2.9 points ahead in the national polls.
Update 9/18/24, 9:23 a.m. ET: This article was updated with comment from Eric Zitzewitz.
Update 9/20/24, 1:03 p.m. ET: This article was updated to include an interview with Thomas Miller.
Do you have a story Newsweek should be covering? Do you have any questions about this story? Contact LiveNews@newsweek.com.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines