When we look at presidential candidates, we pay a lot of attention to their experience. But choosing a president based on past performance is hard. It’s like trying to figure out which kid can run fastest after watching one shoot a basketball and the other kick a soccer ball. The candidates’ track records aren’t particularly useful indicators—they’ve all been doing different jobs with limited relevance to the presidency.

So how should we evaluate who is best fit to hold the nation’s highest office?

Great presidents can glimpse what’s in the future and shape a vision around it. So rather than judge candidates solely on their past, we should examine whether they have the key traits that make for effective forecasters.

The American presidency has been defined by unexpected events. Herbert Hoover didn’t see the crash of 1929 coming—he thought it would just be a downturn. Franklin Delano Roosevelt didn’t expect the Pearl Harbor attack. Harry Truman didn’t anticipate the Korean War. John F. Kennedy was caught short-handed by the Cuban Missile Crisis. Jimmy Carter was blindsided by the Iranian Revolution. George W. Bush was surprised by the 9/11 attacks and the financial crisis of 2007.

Although the specifics of an event may always come as a shock, presidents will be better equipped to lead us through crises if they have already prepared for the possibility of similar scenarios. And with a more reliable crystal ball, presidents might even be able to prevent some of those crises in the first place.

To find out what makes someone a good forecaster, the political psychologist Phil Tetlock spent two decades tracking the accuracy of predictions made by experts (and non-experts) about major geopolitical events. And his research revealed some of the underlying attributes that made certain people far more accurate than others.

It wasn’t ideology that mattered. Tetlock found that “how you think matters more than what you think.” Among both liberals and conservatives, there were foxes and hedgehogs—a concept pioneered by Isaiah Berlin. Foxes are interested in many ideas and see the world in shades of gray; they’re constantly doubting what they know. Hedgehogs zoom in one big idea and see it in black and white, clinging with certainty to their core beliefs.

In Tetlock’s research, hedgehogs were consistently the poorest forecasters, because they couldn’t resist the temptation to oversimplify a complex world. Foxes did much better. They tended to come off as less confident in their predictions, but that openness to the variety of possibilities is precisely what made for stronger forecasting.

Here’s the catch: Although foxes are better at predicting the future, voters tend to prefer hedgehogs, because we make the mistake of confusing confidence for competence. Evidence shows that the more assertive someone appears, the more capable we think he or she is, even though the two are unrelated. As Quiet author Susan Cain explains, “There’s zero correlation between who’s the best talker and who has the best ideas.”

To elect a president who has the best grasp of the future, we need to focus less on what candidates say and more on how they think. Here’s my proposal: Presidential candidates should participate in forecasting tournaments. Over the four years prior to the 2020 election, let’s invite anyone who is considering running to register predictions about dozens of major world events that will be shared with the public during the primary season.

Will Putin still be in office? Will North Korea detonate a nuclear weapon? Will any countries have more cars that are self-driving than human-operated? Along with making their guesses, candidates will enter probability estimates.

I’ll vote for the candidate who excels on two metrics. The first is the accuracy of his or her predictions. But the second, which is equally important, is how much those predictions improve over time. To paraphrase Václav Havel, the playwright who went on to become president of Czechoslovakia, I would rather elect a president who’s looking for the truth than one who thinks he has found it.

Won’t candidates game the system by recruiting teams of unusually skilled forecasters? Let them! We should elect a candidate who has the wisdom to assemble the best team. That may well mean reaching across party lines, like Republican Abraham Lincoln did when he invited Democrat Edwin Stanton to join his team of rivals.

The earlier that presidential candidates start working with their forecasting teams, the better suited they’ll be to succeed in office. Evidence shows that with more experience collaborating together, cardiac surgeons have lower patient mortality rates, software developers are more likely to deliver high-quality projects and finish them on schedule, NBA basketball teams win more games, and flight crews fatigued after a few days of flying together make fewer errors than rested ones who have never flown together.

It’s true in forecasting too. Groups of skilled forecasters who gained experience together were significantly better at making predictions than groups of equally well trained forecasters who hadn’t worked together in the past. A history of shared experience helps teams learn to leverage strengths and compensate for weaknesses, coordinate more effectively and develop productive routines.

The best results don’t belong to the teams with the most experienced members, but to the teams with the most experience working together. A forecasting tournament would incentivize candidates to get that experience early on with their core advisers.

Elections should be a skill contest—and that requires reducing the roles that popularity and ideology currently play in picking our presidents. Finding the best leader means elevating candidates who have the wisdom to know what they don’t know. Who are confident enough in their forecasting skills to put them to the test. Who get the right cabinet on their bus and spend meaningful time driving together. Do all that, and we’ll have presidents who are not only better prepared to respond to the future, but who are better positioned to shape it.

Adam Grant is a professor of management and psychology at the Wharton School of the University of Pennsylvania, and the author of Originals: How Non-Conformists Move the World