In August, Donald Trump tried to assure mainstream Republicans that he’s not running a nakedly ethnocentric campaign by “reaching out” to the black community in a series of speeches and photo-ops. His pitch to African Americans, first delivered in a lily-white suburb of Lansing, Michigan, was, “What do you have to lose by trying something new, like Trump? You’re living in poverty, your schools are no good, you have no jobs.… What the hell do you have to lose?” (He then vowed that he would win 95 percent of the black vote, a promise that was met with loud approval by his supporters.)

Trump wasn’t just making a show of appealing to black voters with this line. His entire pitch to the American people is centered on loss. The country, led by feckless, corrupt, and incompetent elites, is going to hell, he tells us. Americans have been sold out and are being screwed over by foreigners and immigrants, and he’ll make it great again.

“What do you have to lose?” is a powerful question. It goes a long way toward explaining why Trump’s supporters seem to be impervious to their candidate’s constant reversals and frequent steps across various “third rails” of American politics.

And in the end, it’s also a question that will likely doom Trump’s campaign to failure.

At times, the 2016 campaign has proved confounding to political analysts, and that’s largely because it’s a race for which there’s no precedent—the two major parties have never before nominated a woman or a popular celebrity with zero political experience. But we can draw some insight from behavioral economics, a field where “loss aversion” is a well-established aspect of human nature. First proposed by Princeton psychologist Daniel Kahneman and his colleagues, loss aversion is the principle that we feel far more pain when we lose something we already have—twice as much, according to some studies—than we experience joy when we gain something of equal value. Trump’s voters are motivated to take a chance on their erratic candidate by the fact that they’ve lost—or at least they feel that they’ve lost—a lot.

Kahneman regularly asks his students how much they might need to win on a flip of a coin to risk losing $10. In strictly rational terms, the potential to win $11 on a 50-50 chance would make it a solid bet, but Kahneman consistently finds that they won’t risk it until they’re offered $20 or more. The same principle applies to high fliers. In 2013, Kahneman told The New York Times that he’d been conducting the same experiment “with executives or very rich people, asking about tossing a coin and losing $10,000 if it’s tails. And they want $20,000 before they’ll take the gamble.” Loss aversion is a critical driver of consumers’ choices—it explains why people are far more likely to avoid a $5 surcharge than go out of their way for a $5 discount.

A growing body of research suggests that loss aversion also exerts a powerful influence on our political choices. The principle goes a long way toward explaining why Trump’s message is resonating, or failing to resonate, with different groups of voters. The fear of losing what you have “matters the more you have to lose,” explains Bruce Miroff, a political scientist at the University of Albany. That may explain why Donald Trump’s support is strongest among non-college-educated white voters. Those voters may be making a desperate attempt to regain what they had. (Loss aversion explains why gamblers “chase” their losses—bet irrationally trying to recoup what the house took from them.)

It also helps explain why Trump is poised to become the first Republican in 60 years to lose college-educated white voters, who tend to be doing better in this economy and have more to lose. “I would think that loss aversion would imply that traditional, more affluent, better-educated Republicans—the classical stereotype of a country-club Republican—would be fearful of a Trump presidency,” says Miroff.

A common thread throughout this cycle is that Trump’s base feels that they’ve suffered profound economic and cultural losses. Alexander Zaitchik, who traveled through some of the most depressed areas of the country talking to Trump supporters for his book, The Gilded Rage: A Wild Ride Through Donald Trump’s America, told me that, “for a lot of these people, it’s a roll of the dice—they’re feeling out of options and they’re like, ‘Let’s see what happens. It’s worth a shot. Look where we are.’ They’ve seen a steady deterioration for decades, and ‘What do you have to lose?’ does come into play with a lot of these people.”

Since launching his campaign by invoking the specter of Mexican rapists, Trump has appealed to white Americans, especially men, who feel that they’ve lost their privileged position in society. According to an ABC News/Washington Post poll released last week, “controlling for demographics, partisanship, ideology and presidential approval, seeing too little influence for whites and men and too much influence for minorities and women independently predicts support for Trump.” These supporters feel that they’ve lost the economic security, social status, and political dominance their parents took for granted—and are chasing those losses with a toxic reality-TV star. Their sense of loss is so profound that they don’t care whether Trump knows what he’s talking about or is a misogynist or a grifter or has the temperament to serve as commander in chief—they’re the people Trump had in mind when he said he could shoot somebody on Fifth Avenue and not lose their support.

But loss aversion is a double-edged sword, and it’s likely to sink Trump’s hopes of ending up in the White House. Our aversion to political losses has a greater influence on our decision-making as an election nears—that’s why support for third-party candidates tends to fall off at the end of a race. In August or September, it’s easy to say that you’ll take a flier on a third-party candidate or just stay home—especially if your party’s candidate is comfortably ahead in the polls—but come November, the prospect of losing becomes more immediate and concrete, and marginally attached voters tend to come back into the fold.

While the media amplify every twist and turn in the race, the reality is that it has remained structurally stable since the beginning. In The Huffington Post’s average of national polls, Trump’s support has ranged between 38 and 44 percent, while Clinton’s bounced around between 45 and 49 percent. He’s never held a lead. And the only chance he has of becoming president rests on the fact that an unusually large number of voters—including millennials and other key parts of the Obama coalition—remain undecided, are unenthusiastic about voting, or say they’ll support a third-party candidate. Trump needs these groups to sit this one out or vote for Jill Stein or Gary Johnson to have any chance of winning.

And while some Obama voters are just as unsatisfied as any other group with the status quo, they tend to blame the Bush administration for the economic crisis, and don’t share the feeling that they’ve “lost” their country that animates so much of Trump’s. The recovery has been slow, and Washington remains an unholy mess, but they’ve made at least modest gains during Obama’s presidency. Until recently, the prospect of a Trump presidency has been an abstraction, especially when Clinton had an eight-point lead in the national polls, but as we come down to the wire in a relatively tight race, the potential for a Trump victory will come into sharper focus.

And it’s very likely that fear of losing power to Trump’s coalition will drive a lot of those Democrats and Democratic-leaners who aren’t enthusiastic about supporting Hillary Clinton into the voting booths in an attempt to keep a firm grip on what they have. At that point, “What do you have to lose?” becomes the Democrats’ trump card.