TOPICS > Economy > Making Sen$e

Something better than polls for political predictions? You bet!

February 18, 2016 at 6:30 PM EDT
What’s the best way to predict who will win the 2016 election? The answer may not be polls or pundits, but market forces. Predictit is a website that allows people to buy shares in a given candidate’s support in a given state -- essentially a political stock market, and one that could prove more accurate than traditional polling methods. Economics correspondent Paul Solman reports.


America’s quadrennial campaign horse race to elect a leader is now at full gallop. Just like at the track, one question looms largest: Who’s going to win?

Many of us rely on polls and pundits when it comes to presidential prognostication. But your best bet for accurate and reliable political predictions, it turns out, is actually the marketplace.

Political prediction markets like PredictIt incorporate all available information — including polling data and political analysis. “The markets have really been ahead of the polling and ahead of the pundits,” says David Rothschild, who studies them for Microsoft Research. Rothschild told PBS NewsHour economics correspondent Paul Solman that evidence clearly shows that markets where you can bet on — or “invest in” — political outcomes are “more accurate than any collection of pundits or statistical polling averages — and extremely well calibrated.”

Is it gambling? Is it investing? Call it what you will, but Rothschild says putting your money where your mouth is has a beneficial impact on democracy. “People start paying attention in a way that they never did before,” he explains, “so they make an effort to learn about politics, to understand politics, to engage in politics.”

PredictIt chief John Phillips says his marketplace — no matter how limited — provides a prod to democratic involvement. “Even though it’s only a $1 winner-take-all contract with an $850 limit,” says Phillips, “It still starts to cause you to think about things.”

Read the full transcript below:

JUDY WOODRUFF: But, first, a different take on predicting the outcomes of presidential races.

Our economics correspondent, Paul Solman, looks at how a kind of betting market has fared in recent years when it comes to forecasting winners. It’s part of his weekly series, Making Sense, every Thursday on the NewsHour.

PAUL SOLMAN: And they’re off. The candidates are out of the gate, around the first turn and at full gallop in America’s quadrennial race to elect a president.

But, as at the track, one question overshadows all others: Who’s going to win?

DONALD TRUMP (R), Republican Presidential Candidate: Sixty percent.

PAUL SOLMAN: Like every other candidate, if somewhat more loudly, Donald Trump keeps pointing to the polls.

DONALD TRUMP: Here’s an important one, best chance of winning in November. Trump, number one, 47 percent.


PAUL SOLMAN: But when it comes to presidential prognostication, what source is your best bet? A poll? A pundit? A real-time trending Google search? Turns out it’s none of the above.

DAVID ROTHSCHILD, Economist, Microsoft Research: I think that the markets have really been ahead of the polling and ahead of the pundits.

PAUL SOLMAN: Because, says David Rothschild, who studies them, political prediction markets incorporate all available information.

DAVID ROTHSCHILD: This is where markets have a huge advantage over the smartest people just dealing with polling or historical data. We looked at the 2012 elections. We looked at all the primaries. We looked at all the senatorial, gubernatorial, presidential elections.

And you see something that is more accurate than any collection of pundits or statistical polling averages, and extremely well-calibrated.

NATALIE JACKSON, Senior Polling Editor, The Huffington Post: Polls show what people are thinking now, whereas the prediction markets are what people think will happen in the future.

PAUL SOLMAN: Moreover, says Natalie Jackson, senior polling editor for The Huffington Post, technology has made polling more dicey.

NATALIE JACKSON: We’re seeing a huge drop-off in landline telephones, which was pollsters’ bread and butter for a long time.

PAUL SOLMAN: John Phillips heads a nonpartisan political technology company called Aristotle. His firm built and operates PredictIt, one of just two places Americans can legally bet on or, Phillips would prefer I say, invest in political predictions, though no more than $850.

Still, as at the race track or stock market, you’re putting your money where your mouth is.

JOHN PHILLIPS, CEO, Aristotle: And that’s the beauty of putting just a little bit of money on it. It all of a sudden becomes a matter of pride, but it also becomes a matter of focus. You want to see how your investment’s doing.

PAUL SOLMAN: Phillips gave me a tutorial on how PredictIt works. I’m a guy who loves to play long shots. My father worked at the racetrack for many years when I was a kid. So, I would go with John Kasich at 2 cents. All he has to do is go up to 4 cents, and I have doubled my money, right?

JOHN PHILLIPS: Well, yes, if he goes up to 4 cents or…

PAUL SOLMAN: So, what the heck? I bought 25 shares.

JOHN PHILLIPS: Here you got — here’s your 50 cents up here. You are now the proud owner.

PAUL SOLMAN: Of John Kasich at 2 cents to win the Nevada primary, that is.

In addition to the thrill of victory, the agony of defeat, and the chance to make money, PredictIt provides a passel of information about the race.

JOHN PHILLIPS: Now, there are 84 comments on this particular market right now.

PAUL SOLMAN: Oh, look, Clinton 36 cents.

This was the Democratic primary market in Colorado.

So, the market has suddenly collapsed. So I think I will go with that. And now, if I want to buy Bernie Sanders in another market, to level the playing field, and so Oklahoma, Sanders is down to 22 cents in Oklahoma, so I’m going to buy Sanders. Offer matched. Close. I’m in.

So, I’m rooting for Bernie Sanders in Oklahoma, Hillary Clinton in Colorado, and John Kasich in Nevada.

Other larger online political markets, like Ireland’s Paddy Power and Britain’s Betfair, echo the odds on PredictIt and take as much as you want to bet. But they’re illegal for U.S. citizens to play. PredictIt got government approval by limiting the amount to $850 per market and casting itself as both an investment marketplace and academic research tool, since it began as a project at New Zealand’s Victoria University.

JOHN PHILLIPS: We provide the technology. Once the data is collected, it’s anonymized, and then there are research agreements with almost two dozen U.S. universities and Oxford University.

PAUL SOLMAN: But you’re still betting real money.

You keep saying investing, but it is betting, isn’t it?

JOHN PHILLIPS: I see it as a stock market. A funny thing happens when people try to figure out what is going to happen tomorrow or a week down the road. And if I can put a little bit of money in the outcome that I’m expecting, I think that’s a stock market.

PAUL SOLMAN: Can we compromise on agreeing that it’s gambling?

JOHN PHILLIPS: I think you can call it anything you want.

PAUL SOLMAN: But whatever you call it, says researcher Rothschild:

DAVID ROTHSCHILD: The people who are playing the market are taking the pulse of the general population, and they have done so accurately for long enough that we can trust that they’re understanding how things are impacting very quickly.

PAUL SOLMAN: But PredictIt chief John Phillips says his market provides something more: a prod to democratic involvement.

JOHN PHILLIPS: These traders not only pay more attention to politics, but they also are more likely to vote. Even though it’s only a $1 winner take-all-contract with an $850 limit, it still starts to cause you to think about things differently, and that’s the magic here. It may be that it is an engagement tool for the 20-year-olds.

PAUL SOLMAN: Researcher Rothschild agrees.

DAVID ROTHSCHILD: People will start paying attention in a way that they never did before, and so they will make an effort to learn about politics, to understand politics, to engage in politics, in a way that is very, very beneficial to our society.

PAUL SOLMAN: My session with PredictIt’s John Phillips was almost over, but I couldn’t resist checking in on my 2-cents-a-share Kasich investment.

Oh, my goodness. Look at that.

It had actually gone down.

Down to 1 cent. This is just 50 cents, so its impact on my financial future is minimal. But I can feel myself rooting for Kasich to go up. It’s just preposterous, but, I mean, I can feel it. I can feel it.

So, what happened? Well, as of today on PredictIt, John Kasich has bottomed out in Nevada, where Donald Trump is now heavily favored. In the Democratic primary in Colorado, Sanders is the heavy favorite, and I just sold Hillary Clinton at a loss.

But I have done more than OK in Oklahoma. Clinton is still favored, but I sold surging Sanders at nearly double what I paid for him. Bottom line, a $1.33 profit, or a better than 13 percent return on investment in less than a week.

For the “PBS NewsHour” and the power of markets, this is economics correspondent Paul Solman, wishing the rest of you the best of luck.