Miller, a professor at Northwestern University, bases his predictions not on polls, the views of experts, or the views of “fundamental” analysts like “Nostradamus” Allan Lichtman of American University, but on betting prices. His sources are the odds published on the platform he considers the most trustworthy, PredictIt. He uses a proprietary model that converts PredicIt odds into EVs; It is a system that has proven to be extremely accurate in determining the results and margins of the 2020 presidential race and the two subsequent Georgia Senate runoffs, both of which were decided by Democrats.

Before Miller unveiled virtualtout.io a few days ago, he updated his betting forecast every day at midnight. Since late September, the trajectory of the race that these postings have been following has fluctuated wildly. On September 20, shortly after her big debate win, Harris boasted a gigantic lead of around 200 EVs. Trump rebounded early last month and by Oct. 26 had built a lead nearly equal to Harris' lead five weeks earlier. With nine days to go, it seemed as if Trump was headed for an easy victory.

Harris' fortunes improved dramatically after Trump's six-hour rally at Madison Square Garden, a stunning spectacle that bettors believed would anger female voters. On November 2nd, Harris retook the lead with 301 EVs, a lead of 57.

But buckle up and grab your hats: The roller coaster ride took yet another bumpy turn: At the end of November 4th, Trump jumped back to the top, albeit with a super-slim 12 EVs.

On the morning of November 5th, the data scientist's odds show that Trump will win

On the eve of Election Day, the website virtualtout.io – it updates every minute – showed the EV numbers for Trump and Harris swinging back and forth around the 270 needed to win. Put simply, bettors called the race what the experts said and polls had shown for weeks: a coin toss. But the next morning, as Americans went to the polls, the outlook changed again. Trump has been consistently ahead so far today. But his lead varied from large to narrow. Miller warns that the competition remains very volatile in the final hours and could take yet another unexpected turn – this one towards Harris. He points out that November 4th was the most active trading day of the campaign on PredictIt; 518,000 shares changed hands, more than ten times the average for the last few months. And Miller expects trading to be even higher today. This frantic activity, in his opinion, illustrates that hordes of people already betting on PredictIT are changing their minds about which competitor to bet their money on, and that numerous new players are populating the site and placing bets that could spill over into waves, what The choice changes the course of the bets.

He offers five reasons why Democrats should still have high hopes. First, as of midnight on the 4th, the numbers were extremely close, with Trump only a dozen electric cars ahead. separated, in Miller's words, by “just one or two states.” Second, Miller emphasizes that the betting markets, including PredictIt, are biased toward Republicans. This is because the majority of participants are men who frequently bet on sports. “There is a big gender gap favoring Democrats in the election, and also a gender gap in the prediction markets,” Miller says. “It’s not as great as it is for the electorate, but it exists. Even though the know-it-alls support those they expect to win, there is a slight bias towards the person they want to win want win, and that also improves Republicans' chances.” Miller corrects for Republican bias using the same methodology he used in the 2020 races. But the data scientist isn't sure he's made a big enough rebalancing toward the Democratic side. He acknowledges that if his model doesn't adequately account for GOP leaning, the margin Trump enjoys over Harris on virtualtout.io heading into Election Day could be smaller than the modest and fluctuating margin shown on the site , or could even be zero .

Third, Miller argues that while the “technicalities” represented by the betting odds and, to some extent, the polls lean toward Trump, the “fundamentalities” should increase Harris' chances. Among the campaign priorities she supports that usually win, he says, is a centrist, hopeful and optimistic message, in contrast to Trump's extreme positions and negative messages on immigrants and an economy that actually has good numbers overall. The dark, dour approach is usually a loser, says Miller.

Fourth, most of the movement over the past week has clearly been in Harris' direction. Miller says the longer-term trend could be more significant than the Trump spike on November 5th.

Fifth, Harris' closing message sounds like a winner to Miller. It underscores her willingness to work across the political spectrum, adopting the best ideas of progressives like AOC and inviting a Republican to her Cabinet. “The American people want both sides to work together, and that's what she's advocating for by appealing to Republican voters,” he explains.

“The better ones believe that Trump will win,” admits Miller. Although his model was right in 2020, he's much less sure it's the North Star this time around, especially because GOP leanings on PredictIt are so difficult to gauge. “The ticker will show all the twists and turns throughout the night in which direction the race is heading, long before the broadcasters announce the crucial states,” he concludes. “It could get wild. I don't think a definitive, set pattern will be established until profits come in. Bettors will change their minds if they see Pennsylvania or North Carolina turn blue. Based on the data we have, the market could be dominated by day traders.

But in this competition, the only certainty is that nothing was predictable, and for Miller, even the people who risk their own money and are usually right with his predictions can potentially be wrong.