His methodology is very different from pollsters.
Market Forces
A data scientist who correctly predicted a series of recent elections is back to throw his analytical hat in the ring — and he’s betting that Kamala Harris is going to win big in November.
As Fortune reports, Northwestern University’s Thomas Miller startled pundits and pollsters when he called not only the presidential race for Joe Biden, but also two close Senate races in Georgia. When doing so, he not only got the winners right, but also the amount they’d win by, and in all three contests came within 10 points of the final tally.
Using his same complex number-crunching models, Miller is now forecasting not only that Harris will win against her rival Donald Trump, but that she’ll do so with 449 of 573 electoral college votes, per his Virtual Tout modeling website.
The Northwestern data scientist’s approach relies not on canvassing likely voters, as traditional pollsters do, but on election betting markets. To Miller’s mind, polls are a “snapshot of the recent past” and use sample sizes too small — generally between 500 and 1500 people — to accurately reflect the electorate.
Betting markets are, per the scientist, the “best at predicting the wisdom of the crowd.”
Predict This
Rather than being based on political ideals, these predictive markets are built upon cold hard cash, and because people are placing bets hoping to profit, they’re guessing who they expect to win rather than who they want to vote for.
In both the 2020 presidential race and this year’s election, Miller used data from Predictit, the biggest election betting site. By breaking the betting tallies into jurisdictions and tracking hourly changes, the scientist is able to then feed the information into his bespoke model that spits out daily odds.
As Miller explains on The Virtual Tout, the bespoke model that correctly forecasted the 2020 presidential race has suggested since September 10 — the day of the Harris-Trump debate — that the vice president would usher in a “Democratic landslide” in November.
“Both parties’ election prospects changed dramatically on September 10,” the site explains, in the wake “two critical events”: the debate, and pop megastar Taylor Swift’s subsequent endorsement of Harris. Before that, it was “practically a tossup,” he said.
While only time will tell whether Miller is going to be right again, he seems correct in pointing out one of the key themes in this year’s horserace: that “it’s gone from a drastic landslide in Trump’s direction to a drastic landslide for Harris” in the wake of Biden’s exit from the race.
A reversal of fortunes on this level has, as Miller told the magazine, never been seen before — but then again, neither has anything else that’s occurred in recent political history.
More on the 2024 elections: Stranded Astronauts Say They’ll Vote From Space