Inside the Wild Stock Market for Politics Where Traders Bet On Our Next President

How PredictIt turned a chaotic election into a haven for political junkies looking to make a buck

Right now someone, somewhere, is trying to make money off the latest news cycle. When the coronavirus pandemic began sweeping the globe, people began flipping hand sanitizer, while fashion brands pivoted to selling masks. When nationwide protests against police brutality and systemic racism erupted in the wake of George Floyd’s murder, some retailers saw an opportunity to sell merch, such as Black Lives Matter–themed wine stoppers and garden gnomes. And when President Trump tweeted on October 2 that he had tested positive for Covid-19, traders looking to earn quick cash on yet another campaign hiccup turned to PredictIt, an online prediction market where people buy and sell shares of what are essentially futures contracts for political events like elections, nominations, and presidential pardons.

I was one of them. Around 1:00 in the morning of October 2, I checked my phone one final time before going to bed and saw the news. While the rest of the waking world lit up social media, I lay in bed frantically refreshing PredictIt. There, prices were rising for bets that Mike Pence or Kamala Harris would win the presidential election — some traders were presumably considering the possibility of Trump having infected Biden at the debate earlier that week and either candidate dropping from the ticket. I followed the spike for a few minutes before jumping in to buy shares of both VP candidates, hoping to sell them off later that day as more traders heard the news and drove a second spike. Instead, the spike was already nearing its peak by the time I invested; I bought high and was stuck with more than 200 shares whose prices plummeted back to reality by the time I woke up.

Political chaos is generally bad for markets. That is, unless the markets are specifically designed to quantify political uncertainty, as PredictIt is. Just as companies like Robinhood and DraftKings have capitalized on the gamification of day trading and dearth of entertainment while millions found themselves stuck at home and in need of a distraction from the pandemic, PredictIt has benefited from an unprecedented election cycle. The site, launched in 2014 as a sort of educational experiment about the predictive power of markets, is a partnership between Victoria University of Wellington, New Zealand, and Aristotle International, a political data company that manages the site. Since then, it has grown into a thriving marketplace, with around 100,000 traders who have funded accounts on the site—a number that has grown as the 2020 election draws closer, with many new traders drawn to markets related to the presidential race.

PredictIt is permitted to run these real-money political prediction markets in part because, as stated in a 2014 Commodity Futures Trading Commission no-action letter that helped distinguish it from illegal online gambling sites, it is a “small-scale, not-for-profit, online market for event contracts in the U.S. for educational purposes.” The data PredictIt collects from users’ trades are made available for free to more than 230 academic partners, which have used the data in research on financial markets, monetary policy, and prediction markets. While Victoria University of Wellington is a nonprofit institution, Aristotle is a for-profit D.C. firm that takes 5% fees on withdrawals and 10% fees on profits to help cover costs for staffing and managing the site. Besides managing PredictIt, Aristotle also offers technology consultancy to PACs and advocacy groups; data and financial management software to political campaigns; and voter, donor, and consumer databases to campaigns and corporations.

Futures trading on political events is like catnip to people who not only love trading and making wagers but also yearn to feel as though their time-consuming habit of following political news and election data has value that can be translated into material benefit.

Because PredictIt tracks the value of shares and the volume of trades in certain markets over time, the site has gained authority as a predictive tool, alongside forecasting models like FiveThirtyEight’s election forecast and RealClearPolitics’ polling averages. For certain events, PredictIt market data has been a more accurate predictor than media consensus: In 2016, when some political commentators and election forecasts were overconfident in Clinton’s chances of winning the presidential race (Huffington Post’s forecast had Clinton at a 98% chance of winning on Election Day), PredictIt’s odds for Clinton based on trading were roughly 70% to 80% in the final month of the campaign.

Now, a week before the 2020 election, as FiveThirtyEight forecasts an 87% chance of a Biden win, PredictIt’s odds for Biden are coming in at around 60%, with Trump hovering around 40%. This could simply be the result of ardent Trump supporters pushing up the price of their preferred candidate, along with pessimistic Democrats, hardened by a surprise upset four years earlier, hedging their bets to avoid another crushing loss on their investments. Or, just as some believe S&P gains can help predict an incumbency reelection, PredictIt’s tighter odds could be another instance of the market flagging something polls can’t see.

Futures trading on political events is like catnip to people who not only love trading and making wagers but also yearn to feel as though their time-consuming habit of following political news and election data has value that can be translated into material benefit. If a primary role of a market is to facilitate trades, most successful online markets can also boast a secondary role of keeping traders feeling busy, entertained, and useful. But while real-world markets often suffer (or cause suffering) from the fallacy of appearing stable and rational, PredictIt’s intentional peg to political uncertainty and amateur prognostication is in some ways a more genuine and authentic interpretation of what markets truly are: a collection of messy individual bets, dependent on luck as much as intuition, that en masse form a parsable, sometimes wiser whole.

It was during the early phases of the U.S. coronavirus lockdown in March that I first discovered PredictIt. Suddenly in need of distractions, I turned to tracking how prices in PredictIt markets tied to the Democratic primary and Trump’s reelection were shifting in response to a national crisis. Like many of the millennials who’ve responded to the siren song of investment apps like Robinhood during the pandemic, I quickly moved past passive observation to taking comfort in buying and selling simplified snapshots of a political system similarly careening into disarray.

One of the things I soon learned was how easy it can be to mistake reading a lot about politics for having any honed ability to predict political outcomes. A couple of my early trades reinforced this fallacy: I made just over $3 wagering if Montana Governor Steve Bullock would file to run for Senate by March 9 and flipping shares of Stacey Abrams as a possible VP pick to take advantage of Biden’s sudden sweep of the Democratic primary, leaving me feeling utterly triumphant. But a series of losses quickly negated those profits: I messed up predicting if there would be a Biden-Sanders debate in April and the degree to which Trump’s job approval rating would rise during the crisis. When your own money is on the line, you suddenly realize how often you’re wrong compared to how often you convince yourself that you were correct from the start. If I’d been forced to wager on every political event I confidently prognosticated over the past few years, I could easily imagine being deep, deep in the red.

More than 100 million shares have been traded so far in the “Who will win the 2020 U.S. presidential election?” market, while a smaller, more niche market like “Will Trump grant clemency to Paul Manafort in his first term?” has had fewer than 50,000 trades. Markets can range from broad and national (“What will be the popular vote margin in the 2020 presidential election?”) to minute and local (“Will California Proposition 15 be approved in 2020?”) to global (“Will Kim Jong-Un be Supreme Leader of North Korea on Dec. 31?”). This summer, PredictIt even launched several markets where traders bet on whether Kanye West would make it onto the general election ballot in certain states. However, there are limits on the types of political markets PredictIt can run. For instance, it can’t facilitate trades made on questions about “death, disease, or terrorism,” according to PredictIt spokesperson Will Jennings, so you won’t find any markets directly addressing the president’s Covid-19 diagnosis.

Each share in a PredictIt market is priced between $0.01 and $0.99 and ultimately pays out at $1 if the prediction comes true and nothing if the prediction is wrong, but traders can also buy and sell shares anytime before a topic closes. Each share represents either a “yes” or “no” position, meaning you are always betting that something will or won’t happen. Some markets have numerous different “contracts,” each with their own “yes” or “no” share options; the market titled “What will be the net change in Senate seats, by party?” has 12 different contracts, where you can buy “yes” or “no” in options like “Democrats +2”, “GOP +1”, or “No Change.”

Despite its academic applications, the site has some noticeably gamified features to encourage betting: In return for selling greater numbers of shares for profit, traders are awarded higher-level numbers and new titles in the vein of a role-playing game like Dungeons & Dragons or World of Warcraft that awards experience points and levels up characters for completing missions. (I am Level 3, a “Diviner.”) PredictIt also lists its traders on an esports-style leaderboard, ranked by the highest returns on investment. Traders can earn “trophy icons” on their leaderboard listings for achieving a certain number of “wins,” which is the act of selling a share at a profit.

Each market also has its own comment section where traders express frustration at certain prices being too low or high, explain why their prediction will ultimately prevail, or do their best to convince other traders to make investments that likely benefit their own positions in the market, using persuasive arguments, links to polls and articles, and, sometimes, completely fabricated information. “BREAKING: Rudy Giuliani is covid positive,” one user posted the morning of October 9, possibly with the goal of shorting Trump’s presidential bid. (As of writing this article, there has been no news of Giuliani testing positive for Covid-19.)

I bought 22 “yes” shares in Kamala Harris in May at $0.42 and held them until the market closed, walking away $12.76 in profit (before $1.28 in fees).

One reason these seemingly absurd gambits (which, admittedly, could also just be jokes) could influence traders in PredictIt’s comment sections is that unless you know someone’s username, you have no idea who’s making the comment. Almost any U.S. citizen or resident over 18 can trade on the site, and unlike insider trading protocols around stock, a campaign staffer who also trades on PredictIt could decide to bet on the outcome of their boss’s chances of winning an election. Jennings points to PredictIt’s investment cap (traders can only invest up to $850 on any single contract) as a measure that helps prevent any one trader from taking advantage of the system.

If the chance that a politician is trading in their own market feels unfair, it hasn’t stopped me from trading, though I exclusively deal in small sums of money. I’ll usually risk only between $5 and $15 on a single contract, but some traders have thousands invested across dozens of markets. According to Jennings, around 500 to 1,000 of PredictIt’s traders may be trading full-time. In 2018, CNBC reported that one PredictIt trader, Tom Gill, was making 5,400 transactions and $25,000 in earnings in just one month.

The only time I’ve felt an urge to trade in larger volumes was during the height of this year’s Democratic VP nomination market, which offered “yes” and “no” shares on 34 politicians ranging from more-likely options like Elizabeth Warren and Gretchen Whitmer to long shots like Andrew Cuomo and Hillary Clinton. This was a particularly raucous market with numerous spikes and dips; more than 89,000,000 trades were made and 180,000 comments were left by the time it closed on August 20. I bought 22 “yes” shares in Kamala Harris in May at $0.42 and held them until the market closed, walking away $12.76 in profit (before $1.28 in fees). But I also tried to buy low and sell high on Tammy Duckworth and Amy Klobuchar well before the market closed and failed to time up my sales with any media-driven hype. I was convinced that CBS News reports that the Biden campaign had begun formally vetting Klobuchar would drive her price up; instead, I bought high, like every other trader reading the exact same article, trying to pull off the exact same trick.

While novices like myself largely tracked “yes” prices up and down, more serious traders often substantially hedged their VP bets to prepare for any outcome. “My strategy in the VP market was to mostly buy ‘no’ shares across the board,” says Jon Kimball, an active PredictIt trader and co-host of Election Profit Makers, a politics podcast centered around PredictIt. (Beyond PredictIt’s own comment sections, there is a small but active online ecosystem for PredictIt traders stitched together from Twitter threads, podcasts, Discord servers, and an invite-only subreddit.) Kimball bought the short position anytime a new candidate got a bump in the news; by the end, he was maxed out on bets against almost everyone except Harris and Rice. “I made about $1,000 in this market,” he says, “which isn’t that great considering I had about $3,000 tied up in it.”

Despite its use in research and election coverage, a heavy trader like Kimball doesn’t place much weight on PredictIt’s use as a forecasting tool. “The fact that Trump is still trading at $0.40 to win the presidency is insane in my opinion,” he says, referring to market shifts following Trump’s Covid-19 diagnosis. During the same time period, FiveThirtyEight was giving Trump a 13% chance of winning. “Trump can certainly still win, but his odds are more like 10% to 20% if the election were held today, so I don’t think the market is rational there.”

The gap between price and prediction may be especially concentrated in the most popular markets, like “What will be the Electoral College margin in the 2020 presidential election?” where novice traders bet more with their convictions and emotions. “Where I might use PredictIt as a forecasting tool is in the more esoteric markets, like a party primary race, where inside knowledge carries a lot of weight,” Kimball says. “But most markets on PredictIt are just reflecting the conventional wisdom at the moment.”

If anything, PredictIt can sometimes feel almost cautious in its reversion to the historical mean. In congressional race contracts in states with strong partisan leans, prices lean heavy on party incumbents. Social media and news media have, for more than a year, covered Amy McGrath’s record-setting fundraising in her Senate race against Mitch McConnell in Kentucky, often framing McGrath as a serious contender to flip the seat blue. The reality of this is very unlikely; FiveThirtyEight’s Senate forecast gives McConnell a 96% chance of winning reelection. The shares of a McConnell victory are currently priced at $0.94 on PredictIt — perhaps a very slight deal for those hoping to earn a profit on the margins, but far from being unduly influenced by speculation alone.

Just as I tell my friends they’re better off investing in a boring mutual fund instead of gambling on individual stocks, I would probably earn better returns if I treated PredictIt more like a spreadsheet than a video game.

This correlation speaks less to PredictIt having a unique ability to accurately predict events than it does to PredictIt traders’ ability to read the polls and visit other forecast models. In a world where no methodological political forecasting or analysis existed, PredictIt traders would likely be more heavily influenced by their own personal preferences and assumptions, which could result in a market that acts more like a poll of a certain subset of Americans who are more engaged with politics than the average voter. Even with the aid of other prediction markets and models for reference, some of that emotional betting still persists on PredictIt — what Kimball refers to as “retail gamblers” who make bets “with their hearts.”

This mix of quasi-methodological, quasi-personal trading is nothing new to markets. According to Bloomberg, 21,000 people with $40 million in assets are copying the stock market trades of amateur English investor Jay Smith through EToro, a “social trading” network. Smith gained his following through impressive gains — his portfolio is up 62% this year — but the fact that he is a real-life amateur makes following his trades and advice feel more personal, fun, and aspirational than if he were a brokerage or fund. So too is it more fun to make trades based on a mix of genuine research alongside the emotions and gut intuitions that, correct or not, are largely impossible to separate from any perspective into politics.

Just as I tell my friends they’re better off investing in a boring mutual fund instead of gambling on individual stocks, I would probably earn better returns if I treated PredictIt more like a spreadsheet than a video game. And if I was a full-time trader, waiting for “no” prices to dip a few cents so I could hedge my bets across a dozen contracts, that’s probably what I’d have to do. But the lure of PredictIt for me still lies in the hypothetical rather than the practical, even if I still occasionally feel a certain moral uncertainty in making trades based on news of events and decisions that affect — often profoundly, and negatively — the lives of real people. I am no less bothered by it, however, than I am by the knowledge that, right now, traders are buying and selling shares of companies working toward coronavirus vaccines.

If the 2016 election drew more Americans than ever into regularly consuming news, the pandemic has required many of us to crank that habit up to 11 as we’ve searched for ever-updating health expertise and advising, economic indicators and progress reports, and reports of national and local responses to the crisis. And if I feel a self-imposed obligation to consume this information, I reserve the right to feel 1% less powerless in the face of an unending tide of disasters by placing bets on how certain events may turn out. If I earn enough money to buy lunch one day, that isn’t half bad, either.

Bylines in Vox, VICE, The Paris Review, BuzzFeed, and more. Contributor to The Onion. Check out my work here:

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store