00:00This is a story about telling the future, or at least discounting it by paying attention to a whole lot of other people telling the future.
00:08The business of making a market and futures contracts tied to events is exploding.
00:13They may be fun, they may make some people some money, but do they tell us anything about what is around the corner?
00:21I think our goal with prediction markets is really to allow every person that has an opinion on something
00:26to find their market and that big event that they care about to be able to trade on.
00:32Luana Lopez Lara is the co-founder and COO of Calci, the first events contract exchange regulated by the Commodity Futures and Trading Commission.
00:40The company is currently raising funds at a valuation of $5 billion,
00:45even as the Intercontinental Exchange seeks to buy Calci's rival, Polymarket, for closer to $10 billion.
00:52While we went to MIT, the one thing we realized is that most trading happens when you have an opinion
00:57about what's going to happen in the future and you find a way to put that in the market.
01:01So you think this country is going to go into a recession or you think this person is going to win an election
01:05or Brexit is going to happen and you find a way to express that view in the market to hedge some risks that you have
01:12or also to get exposure to something.
01:15And what we realized was there should be a better way to do it.
01:17For us, it was, why is there no direct exchange, that you can just trade directly on what you think is going to happen.
01:23Former Treasury Secretary Larry Summers sees real value in prediction markets like the ones Calci and Polymarket have created,
01:30not only for the people betting on events, but also for those hoping to learn about the outcomes.
01:36Polymarket is probably the largest of the various prediction markets that are out there.
01:45There are a number of them.
01:46I think these are a very useful tool because what markets do, markets where people can place a bet, take a position,
01:58is they enable you to get access to a kind of consensus opinion where people aren't just talking,
02:07but they have to put their money where their mouths are and they're taking a stake in something.
02:12So let me give you an example.
02:14I don't live in New York.
02:17It's not my primary interest, but I do follow the New York City mayoralty election.
02:24There was a development.
02:25Mayor Adams withdrew.
02:28How much did that change the likely prospects of the election?
02:32One could have formed a judgment by trying to think about it.
02:35One could have formed a judgment by reading a lot of political analysts.
02:40In fact, you could just look at the prediction markets.
02:44That, I think, is the value.
02:45It helps us all be more informed, more informed about, even when there isn't much information available,
02:54about what's likely to happen in politics, what's likely to happen in foreign policy,
03:01what are the prospects of an agreement between Russia and Ukraine.
03:06I don't think anybody sensible would pretend that these estimates are exactly right or these estimates are estimates that can be completely relied on.
03:19But I think there's a great deal of evidence from elections, from sports judgments, from other efforts at forecasting,
03:33that looking at these prediction markets provides what's a better and more reliable sense of which way things are going to go than talking to a single favorite expert.
03:48Do prediction markets like polymarkets potentially replace polls?
03:52I mean, polls have come under some siege.
03:54It's gotten harder and harder to do them.
03:56And by the way, are they more susceptible of being gamed?
03:59David, I don't think they replace polls any more than the stock market replaces earnings forecasts.
04:07The stock market reflects earnings forecasts and helps people process what a company is worth, given divergent discussions.
04:18But it wouldn't be able to function if there weren't data and there weren't analyses and projections of future earnings.
04:29So, no, I don't think that prediction markets are a substitute for polls.
04:34They are a way of processing the information in polls, if you're not yourself a pollster, in a much better way.
04:47Scott Rasmussen is a traditional pollster and editor-at-large of Ballotpedia.
04:53And as challenging as the business of polling has been in recent years, it's far from clear that prediction markets can replace it.
05:00The definition of a good poll for most people is one that confirms what they want to see.
05:06And a lot of the criticism about polling is because of perceived inaccuracies or perceived biases.
05:13And a lot of that has to do with the difference between conducting a poll and interpreting the poll, between the polling and the analysis.
05:212016 is a year that I constantly hear people say the polling was wrong.
05:27Well, actually, the national polls in 2016 showed that Hillary Clinton would win the popular vote by three points.
05:36She won by two.
05:37On the state-by-state polls, there were 47 states where there were no surprises whatsoever.
05:43There were three states that were a shock, the big blue wall states.
05:46My point to all of this is that the analysis, the way you look at the polls and the way you read them,
05:52has as much to do with the polling as or with the interpretation of the polls as anything else.
05:58Scott, how has technology changed your business, maybe made it harder,
06:02particularly in getting representative samples in, things like cell phones as opposed to landlines?
06:09We do almost all of our work now online through different panels and text approaches and other things.
06:14Every one of those, every polling methodology you use, and this has been true forever,
06:20has certain biases that you have to correct for.
06:23When we were doing phone polling exclusively, if you had to call into an urban area,
06:29you didn't get as many people answering the phones because they were out doing other things.
06:34So we would have to consciously work to place more calls into those areas.
06:39Today, I believe that pollsters can get a pretty good sample if you're careful with your methodology,
06:47if you're talking about a survey of registered voters or a survey of all adults,
06:56because we have census data.
06:58We have good models for what that should look like.
07:01Where pollsters have struggled in recent years and will continue to struggle is in the question of likely voters.
07:08Who's actually going to show up on election day?
07:11And as we all know, in close races, that's decisive.
07:15And now there's a new technology in town called artificial intelligence.
07:18How is that changing polling? Is it making it better?
07:21The thing that first made polling better, by the way, was having a lot more competition and a lot more pollsters
07:26because we actually had to improve our game.
07:30Right now, artificial intelligence is reshaping all kinds of industries in ways that we can't imagine.
07:37I'm working on a project with Jigsaw, which is a tech incubator inside of Google.
07:42We're going to have a national conversation with five people from every congressional district.
07:48The difference in this poll, this is not to determine who's going to win the midterms or the 2028 presidential election.
07:57We're focused on what does it mean to be an American in the 21st century.
08:01This gets to one of the core problems in polling, and this is where AI can help.
08:08We don't all speak the same language anymore.
08:10So what we're doing with this project, we're asking questions, but we're letting respondents answer in their own words.
08:18We also are seeing a rapid uptake in so-called prediction markets, or these futures markets tied to specific events contracts.
08:27How do those fit into the world of polling, if at all?
08:30They don't fit into the world of polling, but I will tell you that on election night,
08:35when I'm doing, whether I'm doing analysis or if I'm sitting at home,
08:39I watch the prediction markets because you've got a whole crowdsourced army of people out there
08:45who are looking for the latest news, and if I see a candidate going from a 70% bet to a 40% bet quickly,
08:54I say, oh, something has gone on here.
08:56I need to investigate and find out what it was.
08:59Leading up to election night, though, prediction markets are heavily dependent on polling.
09:07If you don't have polling, the prediction markets wouldn't be nearly as good as they are right now.
09:13What prediction markets have done is they're, in effect, another analyst.
09:19They're looking at all of the data.
09:21They're looking at the polls.
09:22They're also looking maybe at who's voting early.
09:26They may be looking at some other factors going on in the campaign,
09:30and they assemble that in a way that an analyst can't.
09:35No individual analyst could combine all of that information.
09:38So I like the crowdsourcing aspect of it, but, again, it's not really a replacement for polling.
09:44There's all kinds of data that's out there today that we can look at and analyze and think about,
09:50and prediction markets are another important tool.
09:52There's all kinds of data that aren't helpful.
10:12If you don't know about testing, go watch them out there soon.
Be the first to comment