Given Miller’s excellent record in the 2020 cycle, it’s highly instructive to examine his outlook for this year’s race. And as in the case of the Georgia runoffs, his view is shockingly contrarian: While most pundits and prognosticators see a “dead heat” or “toss up,” Miller’s numbers show the Harris-Walz ticket far in the lead and the wide Democrat advantage settling into a remarkably stable pattern.

The Miller formula translates the betting odds into electoral votes

Miller’s model centers on the “prices” posted on America’s most active political betting site, PredictIT. It’s a highly-liquid market that trades, on average, 37,000 shares per day. His ruling principle: The odds for the two candidates, after eliminating bets on 3rd party contenders, closely reflect their share of the popular vote. If a candidate’s price on PredictIT is 54 cents, that 54 cents can be used to estimate vote shares, which will be lower, but still still track those odds. Miller also concludes from assessing the results of all presidential elections since 1960 that the popular vote shares reliably predict the percentage of the electoral tally each candidate will receive. “The core of this model is 60 years of presidential election history,” Miller avows. He stresses his respect for the work of Allan Lichtman of American University, who also bases his widely-followed predictions on the history of presidential elections.

“The polls rely on much smaller samples than on PredictIT,” says Miller. “They typically survey only 500 to 1,500. There’s no consistency in the people who are polled. The pollsters are all using different groups of likely voters. They’re are also using different methods so there’s no single way of analyzing their data. By contrast, tens of thousands are voting with their dollars on PredictIT. If they’re right, they turn 57 cents into a dollar. It’s a substantial return on investment. The participants are motivated to stay in the market until the election’s over to get the payoff, so there’s a consistency to the group of investors who are setting the prices.”

That PredictIT’s open 24-7 is another plus. “Hence, at any time you can sell your shares,” Miller continues. “If things aren’t going well for the Democrats, you sell the Democrat shares and buy Republican shares. The investors are always asking, ‘I’ve got a take on what will to happen on November 5, and I want to make money on it. Tens of thousands of participants are expressing their opinion for the outcome with their bets.'” The bettors gather to form a political marketplace that resembles an exchange where investors used their best information to predict future prices for stocks, bonds or commodities.

From now until Election Day, Miller is posting the projected electoral vote (EV) totals for the two candidates in real time; he updates the numbers shortly after the PredictIT prices change, on his website The Virtual Tout. As of 11 AM on October 2, Miller has Harris-Walz leading Trump-Vance by 302 to 236, a substantial bulge of 66 electoral votes. The ill tidings for Republicans: The Harris ticket has been well ahead ever since the Harris-Trump debate three week ago, and bettors are displaying a low tendency to change their minds.

The VP debate moved the markets only slightly

By Miller’s metrics, the vice presidential face-off on October 1 did relatively little to change the trajectory of a race that appears to have reached something of a steady state. “During the entire 90 minute debate, there were no changes in the forecasts. But at midnight on October, a slight change moved in favor of the Republican ticket, suggesting that Vance’s performance was a shade better than Walz’s,” Miller told Fortune. “In this case, only six votes moved from the Democrats to the Republicans, so it was a minor change showing a slender edge towards Vance.”

Miller recently made a key adjustment to his model. “I found that the model doesn’t work well when the odds are high in one candidate’s favor,” he says. “That’s because when people perceive one candidate’s extremely likely to win, they bet only on that candidate, skewing the prices too far in their favor. At the far margins, the initial model distorted the estimation of vote shares. That’s been corrected since September 22.”

Miller found that the campaign’s turning point was Trump’s appearance at the National Association of Black Journalists convention on July 31, when he suggested that Harris had mislead voters about her race. Prior to that gaffe, Trump was running at around 290 electoral votes, 20 more than the number needed to win. The day following the NABJ event, Harris bolted into the lead, and she’s stayed in front ever since. Trump briefly pulled nearly even just before their debate on September 10. Days earlier, Trump got a lift, courtesy of a New York State judge’s decision on to delay sentencing in the GOP standard-bearer’s hush money case until after the election. “On September 10, Trump stood at 261, only 9 short of a win, and Harris had 277,” says Miller. “By the end of the day, Harris had added 35 to get to 312.” Miller also credits Taylor Swift’s endorsement, delivered moments after the candidates left the stage, for giving Harris an added boost.

Harris’ electoral count soared over the next two weeks, climbing to a high point of 337 electoral votes by September 20.

Since Harris hit a post-debate peak, her margin’s been drifting downwards again. Is this the start of a resurgence for Donald Trump?

Since Harris achieved that summit, her electoral count has fallen from nearly 340 to the current 302, and Trump’s rebounded from just 201 to 236. But for Miller, that shift doesn’t indicate that Trump’s staging a comeback. He notes that for the past week that spans the VP debate, Harris’ position’s been consistent at just over 300 electoral votes. Most of all, he says, the bettors’ views of who will prevail seem locked in place. And for Trump to regain the White House, a huge share of wagerers must move to his camp.

“I don’t regard Harris’ drop as significant,” says Miller. “Investors in the predictions markets sometimes change their minds, but not many of them are doing that.” He points to the low volumes on PredictIT. “The last time we saw a spike was at the moment of the debate and Swift endorsement,” he adds. “There was a little uptick for Trump after the VP debate, but the last significant event for moving the odds was the presidential debate. Now, most people aren’t changing their bets, they’re keeping their contracts in place.”

The electoral numbers show that Harris is, as of now, on track to mostly sweep the swing states

By basic electoral math, says Miller, Harris as of today holds a lock on most of the big swing states, looking like a sure winner in Wisconsin, Minnesota, Pennsylvania, and Michigan. He adds that Harris also had Georgia in her column when she reached 337 EVs at the apex of her post-debate climb on September 20. Now, he says the modest Trump bump has put the Peach State back in play, and that Trump also stands a good chance in North Carolina. But for now, the states that have loomed as most pivotal since the race began remain beyond his reach.

The world’s gotten a lot more dangerous in the last few days as the dock workers’ strike threatens to hike the likes of grocery prices and the mideast teeters on the brink of widespread warfare. Big events could still change the course of this election. But for now, Harris is in command, says Miller. The betting sites get it, and predictably, the polls lag far behind the curve. According to the bettors, Trump blew a nice lead by stumbling before the NABJ and botched a budding comeback by floundering at the debate. He may need a watershed moment or a Harris screwup to turn the race around. It could happen. But the bettors doubt it will happen, and as Miller’s shown, the bettors are best at cutting through the fog of polling and pundits, and getting elections right.