Following the Numbers

Sign up for The Media Today, CJR’s daily newsletter.

Ahead of Election Day this fall, a federal appeals court in DC got America ready for the action, clearing the way for fully regulated political betting. Online election betting has existed for years—Beltway scooplet junkies and armchair pundits can wager on anything from the outcome of the presidential race to how many times Donald Trump will say “God” at a rally—through ventures such as Polymarket (an offshore business with Peter Thiel investment) and PredictIt (affiliated with a university in New Zealand and run by a for-profit company in DC). But a judge’s validation of a startup called Kalshi, over the objections of the Commodity Futures Trading Commission, promises to let the market loose. Some firms will be allowed to wager a hundred million dollars; bets on this year’s presidential race alone (factoring in cryptocurrency) have already surpassed two billion dollars. With so many betting sites competing for attention, a pair of journalists formed their own pseudo–public service offering: Election Betting Odds, which aggregates all the top sites in one place.

Election Betting Odds was started in 2015 by Maxim Lott, who majored in economics and math in college, and his boss, John Stossel, the libertarian TV presenter, now with his own straight-to-social-media network. Lott had been Stossel’s intern years ago, at ABC News, when they started looking into election betting. Both found the data intriguing but hard to parse. “There were these UK-based sites, but their odds were expressed in a really annoying format,” Lott said. “If you go to Betfair.com, it’ll be, like, 4.1 for Trump and 3.2 for Harris. But what does that mean?” Lott did his own math on an Excel sheet, which Stossel found impressively comprehensible. Then Lott figured out a way to automate the work by running a scraper, a piece of code that tracks live betting odds. He figured he’d throw the results up online; Stossel wrote some explanatory text. “It was very duct-tape-and-bubblegum at the start,” Lott said.

Since then, Election Betting Odds has had more than twenty-five million unique views. It’s cleverly named, making it an easy find on Google, and free to access. On the night of the past two presidential elections, surges in traffic have caused the site to crash. “I’ve used my small fame to publicize the site,” Stossel said. “Sometimes I make small suggestions.” Lott does the heavy technical lifting. There are no ads; the whole endeavor remains a passion project. Stossel TV pays the guys’ bills, for the most part. Lott is an executive producer there and has a Substack called Maximum Truth, where he expounds on political polls and betting.

As the years have gone by, Election Betting Odds has tracked the chances of hundreds of candidates and ever greater sums. The proprietors believe that their work is journalistic, in that bettors are impressively accurate election predictors: “Better and more convenient and more useful than listening to polls and pundits,” per Stossel. “When I was at ABC and I interviewed James Surowiecki, who wrote the book The Wisdom of Crowds, we discussed that crowds could be mobs—but how, when it comes to the futures market, their wisdom is usually better than any individual so-called expert.”

Lately, reporters have viewed election betting as an increasingly useful proxy for public opinion—or simply interest—in electoral politics. Donald Trump and Elon Musk have made recent references to the campaign’s market odds, which drove coverage; Nate Silver—the former editor of FiveThirtyEight, who now has a Substack called Silver Bulletin—is an adviser to Polymarket and regularly cites betting-market data. Siladitya Roy, a breaking-news reporter at Forbes, began reporting on political betting after June’s Trump-Biden debate, when the markets moved substantially in ways that suggested Kamala Harris would rise to the top of the Democratic ticket. “Polls don’t have immediate movement, so if you wanted some sort of quick reaction around what the sentiment is,” he said, there it was: “the state of the market’s reaction to it.”

At one point, Lott collected data comparing how the site had fared relative to election outcomes: looking at the odds of some eight hundred candidates on the morning of Election Day from 2016 to 2022, out of all those who were given between a 70 and 80 percent chance of winning, three-quarters ultimately did. As for whether betting markets do a better job of predicting election results than statistical models based on polls and other political factors, according to Rajiv Sethi, an economist at Barnard College who studies this realm, “it’s a close call.” (As of this writing, the markets have Trump leading Harris by a few percentage points.)

Stossel and Lott think more members of the press should take note. “I’m a journalist at heart,” Lott said. “I think it’s important to give people good and accurate information. And of course, I don’t mind the attention.”

Meghnad Bose is a Delacorte fellow at CJR.