Bettors say these markets are more predictive than polls.

This interview is part of Tangle, an independent, ad-free, non-partisan politics newsletter where I answer reader questions from across the country. To subscribe for more for the newsletter and interviews like this, just enter your email below.

In 2016, basically every pundit, polling outfit and prediction market was caught on its heels by the Donald Trump victory.

This time around, people are starting to look at more than just polls to determine the favorites in the election. Endorsements, funding, crowd size at rallies and the electoral college breakdown are all getting a closer look from everyday voters than they ever have before.

But there is another indicator that may be worth considering: the betting markets.

Rapidly growing in popularity, websites like PredictIt and BetFair are giving people the opportunity to gamble on politics. And those gambles provide valuable data for anyone trying to flesh out the odds of a political victory. One of the most popular websites tracking this data is ElectionBettingOdds.com, which takes the bets being placed in the market and publishes each candidates’ odds of winning their race based on those bets.

Tuesday night, I spoke with Maxim Lott, an executive producer for Stossel TV and the creator of ElectionBettingOdds.com. He told me why his model works, where he would put his money on the race and what the story of the election — from the perspective of the bettors — has been so far this year.

This interview has been lightly edited for length and clarity.

ElectionBettingOdds.com chances of winning the election at the time I spoke to Maxim Lott.

Tangle: How did you get involved in this and what was your path toward creating this election betting odds website?

Lott: I work with John Stossel and back at ABC News in 2007 we first covered a prediction market called Intrade. That was kind of the first big political betting site. We covered it then and it was pretty cool and people were betting, but Intrade was based in Ireland. And they were taking American customers. So the U.S. regulators sued Intrade because there were all these special gambling rules in the U.S. and eventually Intrade went out of business. We were kind of annoyed because we had been using their data to predict who would win elections and now they were out of business.

So I found the next biggest site, which was called BetFair, and they were based in the UK and also had millions of dollars being bet on politics. The problem is their website and their odds are unreadable. You have to click through five things on the website and then the odds are like “62,” and you think “what does that mean?” And then you have to divide 1 by 62 and that equals a three percent chance — and it’s not something you want to do. It’s not user-friendly at all, especially for Americans who aren’t used to betting on those sites.

For a while I was just manually converting the odds for the TV show I produced with John Stossel and after a while, I thought, “you know, I should automate this.” Computer coding is a bit of a hobby of mine and so I automated it and then I thought, “what if I automate this and then made it pretty and made a few changes? And that’s basically ElectionBettingOdds.com. The title is good and simple and you can remember it and Google’s search engine picks it up so it works really well. We got more than 8 million hits in the last election cycle in 2016. Right now at this minute, we have 250 people live on the site. I’m very happy with the reaction to it.

Tangle: Could you just tell me what are the election betting odds and why do you guys feel like this is an important thing to include alongside polls?

Lott: On ElectionBettingOdds.com we list who is most likely to win the election based on who people are betting will win the election. Right now, Biden is slightly in the lead — basically the Democratic primary is coming down to Biden and Sanders. Biden is at 31% and Sanders is about 29%. Studies have shown that the betting odds are better predictors than polls and especially compared to listening to pundits because bettors are actually putting their money where their mouth is. They’re not just spouting off their opinions or biases, they are more likely to look really carefully and say “is this poll really reliable?” Or “can I really trust what this person is saying?” Because they are going to lose their money if they’re wrong.

Tangle: So where do you guys get your data? Who are these bettors and how many of them are there?

Lott: We get the data from two sites right now: PredictIt and BetFair. These are very big sites and right now in the markets we’re tracking for the general election there is $12.5 million that has been bet there. And then the Democratic primary is about $5.5 million. And those amounts are going to go up more over the next year. It’s a lot of people betting. I don’t believe there are numbers on the exact number of traders but it’s definitely in the hundreds of thousands. These are just people who are interested in the election. Some of them are people who normally bet on sports, some of them are politics nerds who jumped into it. It’s a very random assortment of people. You have totally normal people who are like, “I thought this guy would win,” and you have people with their excel spreadsheets who have said they have calculated that this candidate has a 39% chance of winning.

Tangle: I’d be curious to hear what your take is on the Democratic primary so far from the perspective of the betting markets. Who has risen and fallen? Who is rising now? Whose chances seem stable?

Lott: There has been a lot of shifts there. At first, the bettors thought Kamala Harris had a good shot. But she kind of got knocked out after a few debates. Then Warren went over 50% in the Democratic primary odds. And that was when everyone in the media was saying she was taking the mantle from Sanders and Biden keeps having slips of the tongue. She was slightly over 50% for a bit. Now she’s completely crashed down to earth and is at 11%. I think the bettors are thinking Sanders and her are kind of splitting the left-wing vote and Sanders seems to have all the momentum going into Iowa — at least as of right now.

Tangle: Do you guys expect that after Iowa and New Hampshire the markets will make a move?

Yes. Obviously it’s possible they could be in the exact same place they are right now but Iowa and New Hampshire will provide a lot of new information. The polls are one thing but the actual votes are kind of the only real polls. Who really has the ground game? Who really has the supporters who will bother to go to the polls?

Tangle: It seems like the president’s numbers have stayed relatively solid through. If you were a betting man yourself, would you put your money on Trump for re-election?

Lott: I think the betting odds are about accurate, but if I had to bet I would bet against Trump winning re-election. Me personally. But I may be influenced by the New York City bubble I live in [laughs]. But betting has it around a coin flip and as far as I know, that’s about right. There have been some trends in that when the Ukraine impeachment news was all breaking Trump went down from 50% to 40%. As that process concluded and he didn’t take any real hit in the polls and the economy and stock market continued to do really well, he climbed back to 53.9%.

What really seems to move the markets or cause the most cataclysmic shifts that you sometimes see?

It’s breaking news that definitely moves the odds the most. That’s the most interesting thing. For polls, you always have to at least wait at least three days [for news to impact the polls], whereas the betting — if there is some scandal like Ukraine impeachment — the odds will move right away. They also take into account momentum. Right now, in the polls, Bernie is only slightly ahead of Warren. But in the betting, he is way ahead, something like 28% to 11%, and they’re seeing that he has momentum going into Iowa. Warren peaked in the polls two months and if the primary was held then maybe she would have won.

Tangle: I’ve noticed some famous pollsters like Nate Silver have said repeatedly that he firmly believes that polls are the most predictive way to guess who is going to win a primary or general election. I’m curious about what your response is to him or people interested in elections and why they should trust your model more?

The first thing to note is that Nate Silver also thinks that betting markets have predictive power and he has actually linked to our site many times. So his model is based on the polls but I think no one has done a strong analysis yet on which methods have the best predictive power historically. Silver has had very good predictions, but the betting markets have had good predictions too. It would be interesting to do a rigorous analysis on which one has had a better track record. But on the track record section of our website, we have a graph where you can see how closely the betting tracks reality (Note: Below are the two graphs on Lott’s site illustrating the correlation between the betting market odds and “reality”)

Tangle: Since you referenced 2016, I’m wondering how did this model prove out in what was a shock election?

Lott: Well, we did not predict the result correctly along with pretty much everyone else. But it was closer than all the models except Nate Silver’s regarding the main result of the election, the presidential race. Silver had it about roughly a 30% chance of Trump winning. We had something like 18%. And then a lot of these models were giving Trump 8%, 4%, the Princeton Election Consortium gave Trump something like less than .1% chance. It was ridiculous on its face, your model is off, it’s not controlling for the natural variability in the world.

But many of the serious models had Trump at 1% or 2% chance and the betting odds had him in the teens. Obviously he won. The betting odds reacted very quickly as the data came in, much faster than the TV pundits did, and Trump crossed the 50% threshold somewhere around 9pm EST, which was much earlier in the night than most people were handing it to him.

Overall, it was not perfect, but it was better than most of the respected statistical models.

Tangle: Well, I’ll let you go, I know you’re busy, but it seems like the appropriate way to end this is to ask for your best bet on the Democratic primary. I know you said you probably would put your money against Trump in the general, but I’d love to hear your thoughts on how you think the next couple of months are going to go?

Lott: My money would be on either Biden or Sanders, but I probably would give the edge to Sanders.