Subscribe
Share
Share
Embed
Stripe cofounder John Collison interviews founders, builders, and leaders over a pint.
[00:00:00.10] Luana Lopes Lara
I brought some Brazilian beers.
[00:00:02.00] John Collison
I'm going to try a Brazilian beer then. Do you want a Brazilian beer or Guinness?
[00:00:05.18] Matt Huang
Well, I think I need to have Guinness.
[00:00:07.03] John Collison
You need to have Guinness, that is true.
[00:00:08.06] John Collison
Ok. We lost Tarek already. It's been like one minute. Tarek Mansour and Luana Lara are cofounders of Kalshi, one of the new prediction market firms that rose to prominence in the November '24 elections. They spent four years pre-launch fighting for regulatory approval to build the first onshore prediction market in the US and now trade more than $10 billion each month in prediction contracts.
[00:00:29.03] Luana Lopes Lara
Cheers.
[00:00:29.11] Tarek Mansour
Cheers.
[00:00:32.10] John Collison
What is this that we're drinking, Luana?
[00:00:33.22] Luana Lopes Lara
It's a Brazilian beer.
[00:00:34.23] John Collison
Ok.
[00:00:35.08] Luana Lopes Lara
It's our most famous Brazilian beer, I would say. Very light. I don't know if you like it.
[00:00:39.23] John Collison
I like it. So what is the split between you guys? Maybe in terms of responsibilities, but more interestingly, in terms of outlook?
[00:00:46.02] Luana Lopes Lara
Well, we actually come from the exact same background. We studied math and CS at MIT, same internships, everything. But I'm a very, very optimistic person. I love taking risks. I think everything's going to work out. He's very paranoid, more on the negative side. So it's always like a very good balance. And I think that, outside of what we do day-to-day, that's what really the difference between us that works out.
[00:01:07.15] Tarek Mansour
I mean, there's a little bit of background. So I was going to be a trader. That was really what I was going to do. And you know, when you're a trader and you probably... I don't know if you've ever met sort of or spent enough time with the persona, but it's like a very—
[00:01:19.11] Matt Huang
John is a secret trader.
[00:01:22.19] John Collison
At heart, yes.
[00:01:23.13] Tarek Mansour
But if you're a trader, you're like an expected-value calculator. I think about these sort of tail, really bad outcomes all the time. And one oftentimes doesn't. And I think this is the thing that actually leads to great outcomes.
[00:01:35.06] John Collison
Ok. So I want to ask about that starting out. This is really interesting. You guys started Kalshi and for several years were not able to operate until you got CFTC approval. And that's interesting where just, one, most companies don't start out that way. And secondly, I feel like the Silicon Valley standard that people sometimes trot out as a criticism is sort of the PayPal, Uber early days model, where you start doing the thing and maybe retroactively a structure is put on top of it, but you do a bit of ask forgiveness rather than permission in the very early days. And so can you just tell a little bit of the story of how you started and that approval process? And then I want to get into whether that generalizes to other companies.
[00:02:20.18] Luana Lopes Lara
Yeah, I think that the approach we took from the start was that financial services or healthcare, I think you can't ask for forgiveness. I think there's a big difference between losing people's money. See what goes wrong with like an FTX example that can go very wrong. With healthcare, there's a lot of other massive examples of it going wrong. And we wanted to do things the right way because also when we look at the market, we thought the biggest question to be answered was not, "Is this going to grow?" It was "Can we do this legally in the US?" And we're like, let's just actually address the biggest problem first and go from there. And I think the strategy for a long time, people looked at it as the wrong strategy. Up until we won the election lawsuit, everyone was saying the folks that went offshore, they're doing a lot better, they're growing a lot more. But I think once we won the election lawsuit and proved that the legal interpretation we had was right and we could do the company as we wanted in the US, I think it just really, really took off.
[00:03:14.09] John Collison
What are the timelines here? So when did you start and when did you win the election lawsuit?
[00:03:17.21] Luana Lopes Lara
Right. So we started the company in 2019. We started in YC 2019, and then it took us three years to be able to get regulated and launch. I think it was 2022 at that point. And then we won the election lawsuit at the end of 2024. And that's when we really started ramping up.
[00:03:35.22] Tarek Mansour
There were a lot of like elongated... I mean, there's some overlap between the timelines, but maybe going back to the question and maybe we can talk a little bit more about the sort of history afterwards, but I think it was like a twofold, or two-step process. One, it was a pragmatic thing, which is we felt like to get proper mainstream adoption and institutional adoption, the elephant in the room was, could we do it in a regulated, credible, and safe way? Because it's a complex marketplace, you're moving people's money, and we're like, we have to solve that problem first. That is the hard problem to solve. And, you know, that will be the road to success. The second thing was a bit more principled. We were... What excited us, like when we created this doc, one page on Google Docs, and we wrote a set of things like why should we do this company and why are we so excited about this? We wanted to build the next generation New York Stock Exchange. We wanted to build a financial market that is in the US, that is credible, that is regulated.
[00:04:40.19] Tarek Mansour
We were not very excited about this idea of building something offshore. And so that was really important because it's like, what kind of company do you want to build and why are you doing this in the first place? There's many paths to success. We just weren't very excited about the other idea. We wanted it to be here.
[00:04:57.06] John Collison
You're the first CFTC-approved prediction markets at any scale.
[00:05:01.10] Tarek Mansour
Yes.
[00:05:02.06] John Collison
And still to this day, all the contracts are individually approved. And so—
[00:05:07.07] Luana Lopes Lara
Yeah, every single contract we file with the CFTC and they have 24 hours to stop it.
[00:05:13.15] John Collison
Ok, so they get kind of a real-time feed of the contract.
[00:05:16.10] Luana Lopes Lara
Exactly.
[00:05:17.18] Tarek Mansour
Yeah. And it was a kind of... It was a very long journey to get to where we are in terms of the contract process and how it works. Because you've got to imagine the first time we walked into the building—actually, the picture is right here. This is the first time we ever walked into the CFTC. You know, you walk in and you're talking about this idea and it's... You've got to imagine the regulator's head starts spinning. You're talking about things that don't have a financial underlying. And then there's this idea of potentially hundreds of tens of contracts, hundreds of contracts a week. And I mean, now we're like... But there's all these things where the model wasn't really set up for this. So a lot of the process was actually like this iterative process where you're trying to figure out how to actually regulate this as you get feedback from the regulators and what can we do to satisfy the concerns. So it was a bit like building a product, but you're not building it for a customer, you're actually—
[00:06:12.05] Luana Lopes Lara
Yeah, it's a regulatory-market fit in a way.
[00:06:14.09] Matt Huang
And so now you've gotten them comfortable with you shipping them unless they say no.
[00:06:21.00] Tarek Mansour
Yes, right.
[00:06:21.05] Matt Huang
Have they said no to anything recently?
[00:06:23.21] Luana Lopes Lara
Well, the biggest they said no to was the elections. That's why we had to end up suing them. They said no for two years. But at this point, I think we've worked with them for so long that we know exactly... They trust us as well as a self-regulated entity to know kind of what we can do and cannot do. So we don't do anything around war, assassination, those things that we don't do. So within the parameters that we've worked with them, it's a lot faster.
[00:06:47.17] John Collison
So the election lawsuit was they were willing to approve contracts generally. They were not willing to approve contracts around who would win the election, which is a pretty popular contract.
[00:06:59.09] Luana Lopes Lara
It's the holy grail.
[00:06:59.18] John Collison
The US presidential election. And so you sued the CFTC?
[00:07:03.04] Luana Lopes Lara
Yeah.
[00:07:03.15] Tarek Mansour
Our own regulator. I mean, and—
[00:07:05.20] John Collison
Which is generally not considered a best practice.
[00:07:07.11] Tarek Mansour
It was. I mean, ok, so we—
[00:07:10.13] Luana Lopes Lara
It worked out.
[00:07:11.02] Tarek Mansour
We started talking about the election market at the end of '21 and we started engaging with policymakers, talking with Congress a bunch and the regulator. And they're like, "Yeah, I think it's a good idea, it's a good idea." But then they weren't moving. We started noticing something is off. By the end of '22, they delayed the approval till after the election, what we call a pocket veto. That was brutal. So that was one of the hardest times in the company where we had to lay off a bunch of people. But the harder part of this is that your team and some of your investors, or majority of investors, kind of stopped—
[00:07:48.03] John Collison
Believing the idea.
[00:07:48.13] Tarek Mansour
Yeah. But believing in the strategy, the idea, and it's a bit like, "This is getting a little bit unhealthy. You guys should do something else. Clearly it's not going to work out." But we could not... We couldn't get ourselves to do something else.
[00:08:05.19] Tarek Mansour
We just couldn't. So we're like, "Ok, we're going to try again." And end of '22, so imagine the team is at an all-time low on morale. They're waiting for a new strategy. A bunch of people left, a bunch of people got laid off because we had to downsize. And our message in that next stand-up was actually, "Guys, here's the '23 strategy, is we're going to try again."
[00:08:25.16] John Collison
We're going to do the same thing, but this time it'll work.
[00:08:28.15] Tarek Mansour
And exactly, but this time it's going to work, even though every inch of evidence was pointed in the other direction. And I will say, a lot of this is her. I wanted this to happen so bad, but my rational brain was like "Got to listen to these people." And Luana is much more dogmatic. So we try again, end of '23 they block it again. And I was really at the point where I'm like—
[00:08:48.13] John Collison
"Ok, this prediction market thing is just not going to work."
[00:08:53.06] Tarek Mansour
It's just... And then Luana was like, "Well, the only thing we can do right now out of the entire range of possibilities is we got to sue the government." And at the beginning it was like, "This is crazy." And we took it to the board and we had Alfred and Michael at the time and they're all like—
[00:09:14.03] John Collison
Alfred Lin and Michael...
[00:09:15.10] Tarek Mansour
Seibel from YC. And I remember that board meeting, it took a few board meetings, it took a few times. At the beginning, I was like, "Well, we have to tell you guys it's a bad idea. These are all the ways it's a bad idea." Because you're a regulator and you're like now a 25-people company. The government can do anything; they can shut you down, take out the license. I mean, it is true. So rarely does it work. And even if you win, you will probably lose. You will end up getting killed in the process. And it took a few times. And I remember very vividly, there was a meeting we had internally before talking to the board.
[00:09:51.00] Tarek Mansour
And this was the night before. We had lined up the lawyers and everything. And I got cold feet. I was like, "Let's just focus on getting a clearinghouse. We could focus on financial products. We can focus on all these other things. We don't need to sort of tank it all on this and really bet the farm on this." And I remember Luana in that call—I forgot the exact wording, but it was something along the lines of like, "Are you fucking kidding me?"
[00:10:15.23] Luana Lopes Lara
That sounds like me.
[00:10:18.05] Tarek Mansour
And I realized, all right, I'm not going to win this fight. `But the other part of me is like, "We got to do this." So we went to the board, and the response was basically, "It's an anti-pattern, it's a bad idea, but a lot of great companies are built by an anti-pattern. There's something off. That is weird. That happens and maybe this is yours."
[00:10:39.01] John Collison
Good way of putting it, that every company is different in some new way, and so yeah, this could be yours. What was the basis for the decision where you won the election lawsuit? Was there any interesting policy angle?
[00:10:54.06] Luana Lopes Lara
Right, so the whole point is that the government cannot stop any type of contract unless it makes a finding that's against public interest and it has to fall within certain categories of war, terrorism, assassination. And the CFTC was taking the stance that they were trying to fit elections into any of these things. They're like, "Oh, elections might be illegal under state law." Because betting on elections, there's this one state that in bucket shop law... They try to find something to stop it. And we knew we were very, very clear on the law, like elections have economic impact. If the elections have economic impact, they need to be allowed to trade on a futures exchange or derivatives exchange. And it was basically, I think what the lawsuit did is it told the CFTC that they couldn't just do whatever they wanted. That kind of like—
[00:11:38.00] John Collison
The categories it prohibited, it needs to actually fall under one of the prohibited categories.
[00:11:43.11] Luana Lopes Lara
Exactly.
[00:11:43.20] John Collison
Which elections did not.
[00:11:44.18] Luana Lopes Lara
Right. Exactly.
[00:11:45.10] Tarek Mansour
And I think that's important because the law... The thing that we always say, the law applies to companies, but it also applies to the government.
[00:11:51.08] John Collison
But you should make your point about maybe suing your—
[00:11:54.16] Matt Huang
Well, I think certainly in crypto and in prediction markets, it's sort of this unique thing of suing the government. But I was sort of surprised to realize that Coinbase has sued their primary regulator. In govtech, SpaceX, Anduril, and Palantir all had to sue for various reasons. So it seems like it's actually more common than Silicon Valley conventional wisdom. So what advice would you have having dealt with the government to people out there trying to build businesses? What sort of situation would be, you think, ripe to actually make that kind of challenge?
[00:12:35.15] Tarek Mansour
I think it's a sort of no other option situation.
[00:12:39.09] Matt Huang
So it's still painful.
[00:12:40.03] John Collison
But did you actually have no other option? Couldn't Kalshi have done fine without elections? I mean, elections are obviously very helpful because they're such a big shiny thing, but I presume elections are not a majority of your contracts today.
[00:12:50.18] Luana Lopes Lara
I think it was just too important and maybe that's the dogmatic or whatever, but it is the holy grail of market. That's the one that you could see the use case of the data the best and you can see the use of the... And you can see the 2024 election, the polls were completely wrong and the markets were so much better at bringing that sort of information. And I think that it's the shining example of why these markets are a force for good and we need to have them in the US and regulated, which other markets just won't have.
[00:13:18.15] Matt Huang
So to John's point on PayPal and Uber and asking for forgiveness, there were other prediction markets operating and showing real usage offshore. And so I'm wondering, how much did that help demonstrate that election markets weren't against the public interest? Did that factor at all into the court case, the fact that people are already doing it?
[00:13:41.23] Tarek Mansour
I don't know, but specifically on the court, it was much more grounded in sort of like—
[00:13:48.22] Matt Huang
The law.
[00:13:49.04] Tarek Mansour
Yeah, the law. Reading our law is the Commodities Exchange Act. So that's one of the financial statutes. The other one is the Securities Exchange Act. And reading it and really interpreting it and is the regulator overstepping? Now, I think for us it was a good way to learn. Because we could not learn about our product because we took this regulatory-first approach, where we will ask for permission first before doing something. And so in some ways it was helpful that you could have some data, some evidence that could guide our decisions over time. And I think it also helped educate some people about the existence of prediction markets. They are here. Here's how you could use them, etc. But I don't think unregulated offshore players really help from a policy perspective.
[00:14:42.03] John Collison
Could Kalshi have been started 10 or 15 years prior, or was there some moment of openness in the CFTC? Was there some tech enablement that was required? Were stablecoins required?
[00:14:54.14] Luana Lopes Lara
I do think that there is a part of it that crypto at the time was augur and there was like some very early prediction markets. I think that the existence of that made the CFTC also be like, "We need a legal regulated alternative to this." Because before you could just say no to things. I do think that played a role, but maybe like 5%, maybe 10%. I don't think it's more than that.
[00:15:16.04] Tarek Mansour
The broader thing is there's always intellectual interest in prediction market. And I think that started in the '50s. It is a better source of signal than most other mechanisms of getting signal. But there wasn't a real pain. I think 10, 15 years ago in the way that there is pain in the last few years. And that pain, I think, is the country is more polarized. I would say the world is more polarized. Social media has really bifurcated social feeds. Clickbait is rampant. Like the incentive structure for most things that we read these days is clickbait, whether it's a lot of traditional news or social media or other. So there was more of a pressing problem, I think, that helped create the wave and this sort of adoption that you're seeing in prediction markets that I don't think happened or would have happened 15 years ago. Because the problem wasn't that—
[00:16:07.08] Luana Lopes Lara
Yeah, and that's because most of our users, like 80% of our users are actually just looking, consuming information. They're just coming in and seeing who's going to win the Texas primary yesterday and seeing, "Ok, the polls are saying they're tied, but they're not tied" and all those things. That consumption of information is way more important and relevant now.
[00:16:25.20] John Collison
So you're saying like algorithmic feeds, Kalshi markets do very well on algorithmic feeds and just maybe people wouldn't have been as interested 10 or 15 years ago.
[00:16:36.17] Tarek Mansour
Yeah, I just think that sort of there's a meaningful and accelerating rise of distrust in traditional sources of information. And so, you need a new one. And this does work, right? Like the incentive structure for a prediction market is truth. It is more volume. It is more liquidity, which translates to better and more accurate forecasts. And it took a few iterations for people to start trusting it. But as you start building a track record, people start trusting it and they're never going to use a better product.
[00:17:12.11] John Collison
Well, to the point of substance and volume, can you guys give us the outline of... It seems like it's growing very quickly.
[00:17:18.12] Tarek Mansour
So volume in February was $10.4 billion.
[00:17:21.09] John Collison
Dollars of contracts traded.
[00:17:26.04] Tarek Mansour
And that's up 11x over 6 months, I think, or a little bit—
[00:17:33.12] John Collison
It's growing so quickly, you don't even bother to go back a year because that's just ancient history.
[00:17:38.14] Tarek Mansour
Yeah.
[00:17:38.22] Luana Lopes Lara
I mean, a year ago, it really is. It's like we just had, for example, one sports market. We only had the Super Bowl.
[00:17:44.16] Tarek Mansour
In February... Yeah. I mean, it's growing very fast.
[00:17:47.06] Matt Huang
Fastest-growing company outside of AI.
[00:17:49.02] Tarek Mansour
Yes, I think so. And we compete with, I think, even some of the top AI companies. I don't know what Cursor and Anthropic's latest numbers are—
[00:18:00.18] John Collison
I think 11x is very quick, even in AI.
[00:18:03.12] Tarek Mansour
It's quick. And I think because it's an... So we are a marketplace. It's a true marketplace that has all the attributes of, it has network effects. So what happens in those situations is that users retain better because there's more diversity and more liquidity. Their participation and volume grows over time, which obviously grows their usage, but it also grows other people's usage because there's more liquidity in the system. And then they share it more with other people because the product is getting better. And so the sort of trifecta of factors is leading to this sort of growth.
[00:18:38.22] Matt Huang
When some of your early growth depended mostly on other brokers, and I think you've evolved that mix today. What's the broker mix and how do you think about that?
[00:18:49.04] John Collison
Sorry, what's a broker in this context?
[00:18:50.22] Matt Huang
Like Robinhood.
[00:18:51.17] John Collison
Ok, like a distribution.
[00:18:53.07] Tarek Mansour
To explain what that is. Which is interesting. I mean, the—
[00:18:56.01] Luana Lopes Lara
Well, I can explain the broker part, but I don't know what we want to show in the numbers part. But basically—
[00:19:01.13] Tarek Mansour
That was a tough one.
[00:19:02.14] John Collison
Ok. We could delete that part.
[00:19:05.02] Luana Lopes Lara
All right. So because we're an exchange and clearinghouse, we basically function like the New York Stock Exchange.
[00:19:10.11] John Collison
Another beer for her.
[00:19:11.07] Matt Huang
Yeah, exactly.
[00:19:13.20] Luana Lopes Lara
The Chicago Mercantile Exchange. So brokers can connect to us. You can go to Robinhood to trade stocks. You can go to Robinhood to trade on Kalshi, same with Coinbase or whatever. And it's always been part of how we think about... We always wanted to be an exchange and a clearinghouse first, and actually connection to a Goldman Sachs or Robinhood is very important for how we thought about this ecosystem as a whole. In the beginning of last year, we launched the first broker partner that we had was actually Robinhood and then Webull. And at the start, actually, when we were starting to ramp up, the brokers were a very, very big part of kind of how we started growing, which was actually very great because the brokers bring so much demand. Then we get all the market makers to come in because they want to trade against retail flow. And then we could kind of buy ourselves time to ramp up the direct product a lot, to where it is now. But basically how you think about it is, well, we really are at the core is an exchange and a clearinghouse, and then you can access us through our app, website, API, but also any broker.
[00:20:09.07] Luana Lopes Lara
We're investing more into institutional now, international brokers, so you can be in Brazil and you can trade on Kalshi, all those things. They're coming soon. But on the numbers, you can take it.
[00:20:18.14] Tarek Mansour
I mean, maybe we won't share numbers, but what we call direct, Kalshi Direct, which is our Kalshi.com, Kalshi app, the consumer business. That has grown. That has sort of dramatically outpaced the rest, our other sort of intermediary or broker business. And I think it's just that the brand has gone mainstream. I think people, when they think about, have a difference of opinion on something, it's sort of becoming synonymous to like, "Oh, let me pull up Kalshi and see the odds," or "Let me sort of place a position on Kalshi." And that's contributing... There's just a lot of organic growth now. And I think that's going to continue over the next few months.
[00:20:58.12] John Collison
You're describing how you grow the individual retail side of the market, whether people are coming through brokers like a Robinhood or people coming directly to the Kalshi website. There is also, when you're an exchange like this, you have to spin up market making. And in the end, the New York Stock Exchange doesn't have to think too much about market making because just the economic incentive is there. And so when something is at large scale, that's not as big of an issue. But I'm curious what that was like in the beginning. Were you guys doing the market making? Did you work with market-making partners? Now, how do you incentivize market makers to participate? I'm just curious what the market making scale-up has looked like.
[00:21:39.03] Luana Lopes Lara
So there's actually two groups of contracts on markets on Kalshi, and they behave very differently, and the market-making incentives are actually pretty different. So you have the long tail of markets, right? The ones like, will One Direction have a reunion or all those things. And they are actually very hard to price. And because there's not necessarily a lot of demand, we actually have to incentivize market makers to come in. And there's like liquidity incentives, all those things for them to come in. And it's actually how we think about how to build our moat long term is how do we get very sustainable, solid liquidity in this long tail of markets so we can get... We have like, I think, 10,000. How do we get to 50,000, 100,000 markets? We've still... But on the other side, you have the more classic like crypto, sports, all of those guys. And on that side is actually a lot easier to market because you have very clear proven demand, is a lot easier to price. So the market-making incentives on this side is actually we don't pay them for it.
[00:22:32.07] Luana Lopes Lara
We just rebate fees. But they have very, very, very hard conditions to meet. They need to have uptime of a certain amount, spreads, and top-of-book size and all of those things. Because we see it more as like incentivizing stability of the book than it is incentivizing them being there.
[00:22:47.15] John Collison
What does incentivizing stability of the book mean?
[00:22:50.03] Luana Lopes Lara
For example, if you think of a live game or like if you're trading an hourly crypto—
[00:22:56.00] John Collison
You actually don't want the price flying around a whole bunch if there's no new information.
[00:23:01.07] Luana Lopes Lara
Right, exactly. Or even if there is, if someone is about to sort of touch down, you don't want the book to just have like no liquidity whatsoever. You want it maybe to go a little bit wider, but you want people to be able to trade. And actually, when we go into the intermediated model, the brokers come with expectations that they have from traditional markets. So they are expecting, "We want this spread and this size at any point in time. It doesn't matter." So we need to go to the market makers and we're like, how do we incentivize this? Even though if you think about, you should just let the markets do whatever they want to do. And if they're going to go way wider because they need to, it is. But we have to kind of play with incentives in a way to... For all of our users, including the brokers.
[00:23:37.19] Matt Huang
So during those moments when the spreads would normally blow out wide, are market makers losing money then and they're cross-subsidizing to the other parts where like it's more stable?
[00:23:49.05] Luana Lopes Lara
Well, now there's so much demand that I don't think they're... You can make money on spread even if the spread is a little lower. But that's the point of the program. You have to think about all the benefits you get in this program. And then even if you're losing a little bit in this time, having the benefits is worth it.
[00:24:04.21] John Collison
So you want tight spreads all the time for the major markets.
[00:24:08.21] Tarek Mansour
Yes.
[00:24:09.01] John Collison
That's what market stability is. And that actually takes work to engineer.
[00:24:12.14] Tarek Mansour
It's hard to get there. You have to... But there's more to it, which is... So I think the magic and the uniqueness of, I would say, or the special thing about prediction markets is that a lot of the liquidity is not what you consider a market maker.
[00:24:28.12] Luana Lopes Lara
Right, right.
[00:24:29.03] Tarek Mansour
It's people.
[00:24:31.16] Tarek Mansour
And so this goes back to the whole point. So maybe let's just go back to first principles. There was the regulatory thing that we figured out, but then there's a liquidity problem, which is historically it's a bit like we said, the New York Stock Exchange or the CME. They're like, "We're going to create a green future. We're going to take two years to figure out what it looks like. And we're going to call all of our buddies, like the 50 market makers that we all know, we all hang out at Christmas parties together, etc. We're going to get them ready and they're going to start helping us get this product launched. And then we're going to market it for the next three years." And then it's just the same thing. The liquidity is there. But prediction markets is really different because now you have to create liquidity in these products on like a weekly, daily, maybe even hourly basis. How do you do that? It's much more dynamic. There's new things all the time.
[00:25:20.23] John Collison
I think it's counterintuitive to people that you have to incentivize market makers to create liquidity because in the stock exchange, you don't have to incentivize high-frequency trading firms to create sub-second liquidity. They are very excited to take on that project themselves and build the high-speed interconnect between New York and Chicago to accomplish and everything like that. And so is this just the stage that prediction markets are at, or is there something fundamentally different?
[00:25:47.09] Tarek Mansour
I think this is where I was talking about, which is this idea that you need... So, so maybe finishing that sort of line of thought and then like I'll get to the answer there. You now have a model where you need liquidity built on the fly much faster, much more dynamically. And the market makers, the traditional Wall Street market makers are not geared up for that. It's not like they can spin up a new desk to price politics or price culture in an hour. And so, but this is the part that gets really interesting, and this goes back to the foundational principles around prediction markets, is like a lot of these markets, the people that will price them the best may not actually be the experts or the authority figures that you usually would think about. It's actually random people that live—
[00:26:36.14] Matt Huang
Internet anons.
[00:26:37.20] Tarek Mansour
Exactly. The super forecasters, those are... It's like extremely dispersed. You cannot find like a clearly defined demographic who they are. And I think that the thing where we got to now, and that took a very long time and you had to incentivize, is we have the community, a strong community of superforecasters that are on Kalshi that can help price these things extremely effectively and fast. Where you don't have... But it took a while to kind of get them incentivized and come in and spend the time and resources to take it from a hobby, which is—it was a hobby to for a part-time job. Now it's a full-time job because the pie is so big.
[00:27:12.18] Luana Lopes Lara
And a metric that we can share is actually the... When you think about traditional market makers, the biggest percentage on a platform of a traditional market maker is less than 5% of the maker orders in that market, of the liquidity.
[00:27:26.10] John Collison
Sorry, say that stat again?
[00:27:27.10] Luana Lopes Lara
Yeah, so less than 5% of the order—people come and make orders. Less than 5% of the ones that match actually come from the big institutional market makers you'd think about.
[00:27:37.01] John Collison
I see.
[00:27:37.07] Luana Lopes Lara
Over 95% are just like—
[00:27:38.15] John Collison
Peer-to-peer.
[00:27:38.23] Luana Lopes Lara
Peer-to-peer or funds that have like two people that just got set up.
[00:27:43.00] Tarek Mansour
Which is unusual for—
[00:27:44.00] Matt Huang
How many of those small full-time little shops are there?
[00:27:48.02] Luana Lopes Lara
There's over 2,000 people that market-making.
[00:27:50.19] Tarek Mansour
People/small shops.
[00:27:52.06] Luana Lopes Lara
Yeah, on like a specific—
[00:27:53.12] John Collison
I think what Matt's getting at is like, who is a market maker on Kalshi? There's all these Jane Street conspiracy theory memes. Is it like—
[00:28:01.01] Luana Lopes Lara
That's my Twitter feed.
[00:28:01.18] John Collison
Or is it like some guy in his garage, you know, drinking Red Bull at 3 AM, market-making by himself?
[00:28:07.16] Tarek Mansour
The guys in the garage are the most crucial.
[00:28:09.09] Matt Huang
And you're saying those are 95% of the flow?
[00:28:12.20] Tarek Mansour
They're extremely crucial to the ecosystem because they price fast. They're monitoring the situation all the time. They're the original situation monitors.
[00:28:22.17] John Collison
Yeah, I see. Kalshi is built on people who are monitoring the situation.
[00:28:26.15] Tarek Mansour
And so one example I'll give, and I've given this in the past. The best inflation forecaster on Kalshi over the last few years is not— none of the institutions or the big-name hedge funds. It's this guy who lives in Kansas, never traded financial markets before, just likes to read the news and just knows how to predict inflation. He can feel it. And you have so many of these people. Now you have a few thousand that are formally committed. But there's tens of thousands of these people that know a bunch about a bunch of different topics. And they're sort of actively pricing these things. And they do it as a full-time job. And they get rewarded for that.
[00:29:10.18] Luana Lopes Lara
You need to talk about my favorite user.
[00:29:12.13] Tarek Mansour
Oh, yeah.
[00:29:15.02] Luana Lopes Lara
Or—
[00:29:15.06] Tarek Mansour
Well, I have a new favorite user, by the way.
[00:29:16.22] John Collison
Ok, each of you can tell us your favorite user.
[00:29:19.08] Tarek Mansour
I was thinking about it this morning.
[00:29:20.03] Luana Lopes Lara
What's your new favorite user?
[00:29:21.09] Tarek Mansour
The Washington Journal article tax guy.
[00:29:23.02] Luana Lopes Lara
Oh, yeah, that's true. He's a good candidate. But no, my favorite user is this Ariana Grande superfan. And he found Kalshi during the election season, and he's like, "I don't like the elections, whatever." Then he found our Billboard ranking markets of like charts.
[00:29:37.05] John Collison
He's going to work on important markets.
[00:29:39.04] Luana Lopes Lara
Important markets. To me, very important. And he's made over $150,000. He's getting every single thing. He paid back student loans. He put himself through a master's degree, bought a car. And all those things. And he just like loves these markets. And he's never really traded, never done anything like that before. But it's the first time that he actually has a way to monetize this very compulsive hobby that he had on music charts. And he's able to do it. And he's also very, very nice to us on Twitter. So I like him.
[00:30:07.13] Tarek Mansour
So I had many over the years, but maybe I shift a lot. But like the—
[00:30:12.14] Luana Lopes Lara
He's not loyal. See, I'm loyal to my guy.
[00:30:14.22] Tarek Mansour
Well, you know, I love all of our users, But there was an article last week in the Washington Journal about a tax accountant who was very active on Kalshi, Alan. And when DOGE came around and like there was a lot of talk about how much they could cut, he actually read a bunch of tax codes and a bunch of statutes, just dug extremely deep, and then realized there is no way they could hit the targets. And even if... He really deterministically realized. And then he basically talked to his wife and he's like, "I have extremely high conviction in this trade. I know I..." It's a bit like Michael Burry with The Big Short. So this guy put a big short but on DOGE this time. And it was big, he really went all in. And he won. And it's just like one of those amazing sort of showcases of what this can do. Now you have a market that if you have that sort of knowledge, which maybe oftentimes is esoteric, like I'm assuming none of us have read all these tax codes, you can actually go out in the world, do research, get smarter about the world, and then get rewarded for that.
[00:31:33.23] Tarek Mansour
And that's awesome.
[00:31:35.11] John Collison
You know, an early field of AI was poker bots. Are you seeing any good AI market makers? When you say no one's read all these tax codes, I mean, no one except Claude.
[00:31:47.19] Luana Lopes Lara
That's fair. We should ask.
[00:31:48.16] Tarek Mansour
We are seeing more, increasingly more people using agents to trade. So that's definitely—
[00:31:53.05] Luana Lopes Lara
Especially on the API side.
[00:31:54.05] Tarek Mansour
On the API side, it's very big. And, you know, like—
[00:31:56.21] John Collison
But do you have users who are successfully running market-making businesses that are mostly agentic?
[00:32:01.13] Tarek Mansour
Users don't exactly tell us their strategy.
[00:32:03.16] John Collison
But you talk to them, you know, you just—
[00:32:05.15] Tarek Mansour
Generally, yes. But the way I think about it is like, do we think RenTech back in the days was using agentic models? I'm talking about Renaissance Technologies to trade. Like, yeah, the early versions of them. And so I think they're just evolving and they're getting better. And like most of our traders in their stack have some sort of summary and synthesis module that's AI-driven.
[00:32:32.21] John Collison
I guess what I'm curious about is, fully autonomous, no human in the loop, consuming information and providing a market based on that? That feels like it's coming quite soon, if it's not like your Claw making a market.
[00:32:46.06] Luana Lopes Lara
Yeah, I don't know if there's a full... I know that, for example, there's a lot for international elections, just on translating all documents posted in English to do all of that. But I don't know if it's all fully—
[00:32:58.04] Tarek Mansour
We're doing... So we don't know if the models are there yet. So we launched Kalshi Research recently. And one of the threads that we want to work on is we're talking to some of the research labs to create a new benchmark around which models actually predict the future better, which could be a unique benchmark around are these models developing some understanding of the world that goes beyond memorizing old patterns? And I'm honestly excited to see how it goes.
[00:33:28.11] John Collison
And what's the eval for that?
[00:33:30.11] Tarek Mansour
We don't know yet, but I think you could roughly let the models run for, make predictions on same set of markets for a month or two and see which ones perform that percentage of predictions that were correct, P&L over time, etc.
[00:33:43.15] Luana Lopes Lara
Right.
[00:33:45.02] John Collison
Ok. Another market-making question. So sports bookies have this need to crack down on what in their industry is called sharps, you know, people who are too good. Where, I think people don't think so much about this dynamic, but for a sports bookie, the best possible punter is someone who is kind of unsophisticated, bets on, like, their home soccer team's game.
[00:34:13.01] Matt Huang
Irrespective of the odds.
[00:34:15.05] John Collison
Exactly, and just wants the home team to win or whatever. And the worst kind is someone who's super sophisticated, finding the narrow markets, because for a bookie, maybe they're making odds on 10,000 different markets. They only need to be wrong once or twice for people get to choose which they play on. And so they can't be right on all the odds that they're offering. And these sophisticated people go find those. And so what happens is they basically use behavioral signals to identify if you are just signed up and you're betting on your home sports team's game, that's good. And if you appear to be really sophisticated with all the signals that they would use, they shut them down. But it's interesting, right? You think, I'm just booking on the bets that, on the odds that you're offering. But if you're too good, they'll shut you down. It's maybe like card counting in Vegas. Do you have this dynamic with sharps? I would have thought no, that you just are fine with it. But do the market makers worry about too sophisticated counter-parties on the other side?
[00:35:16.23] Tarek Mansour
The sharps are the market.
[00:35:18.01] Luana Lopes Lara
I mean, to be clear, we don't limit any winners. We don't have any of the... We want all the winners.
[00:35:25.22] Tarek Mansour
We need the sharps because how do you get market accuracy without the sharps? This is the difference of the—
[00:35:31.21] John Collison
Well, yes and no, right? Because what you want is different because the sharps can snipe. They can just turn up once when the odds are wrong, grab a big win, and then disappear. Whereas what you're describing is you want during the game or during the election, You want narrow spreads all the time. And so I feel like providing good market-making is different than being right.
[00:35:53.15] Tarek Mansour
But a lot of the sharps can actually do better if they provide market making and become part of the liquidity. So this is the main difference, which is very important, the maybe disclaimer—I don't gamble, I trade, which I've always found the difference. And I think gambling is this idea where the business model is you are the house and your revenue is your customer's losses. So a lot of dynamic that you describe has to be true because your incentive is like, well, if somebody's making money, I got to stop them because they're making me lose. That's just going straight from my bottom line. And the opposite is true. If somebody is losing money, I got to figure out how to bring them back. That's a very different model from traditional financial markets where the structure is you have to incentivize fairness and transparency. That's the structure. You want to create fair rules of the game for people to participate. Maybe Matt is better than Luana, and maybe Luana is better than Matt, and they can battle it out. They can figure it out.
[00:36:49.03] John Collison
But you think the incentive system is very different, where you do not, like a casino or something, you do not monetize on some zero-sum other person losing. You monetize just on transaction fees.
[00:36:58.23] Luana Lopes Lara
The best outcome for us is that people are like, "This is fair. They have good prices. They have stable liquidity. I'm going to go there." But of course, for us to get there, we also need to incentivize different players differently. So that's why, for example, a lot of the liquidity programs come into effect. They're like, for providing liquidity, you're taking a lot more risks because you're putting yourself out to be sniped. Then we're going to lower your fees. But if you're taking and you're going to snipe, you're going to have higher fees so you can pay for that activity. So in a lot of ways, you use fees to incentivize pro-social behavior. Exactly. And I think that that's actually how we see a lot of how financial markets actually do the same.
[00:37:28.06] Tarek Mansour
Financial markets are the same.
[00:37:29.03] Luana Lopes Lara
The same thing.
[00:37:30.06] Tarek Mansour
But it's more balancing out the marketplace so that people are providing value to the marketplace. Have a little bit more tilt. And then people that are taking away value have a little bit less.
[00:37:39.03] John Collison
What behavior is pro-social and what behavior is antisocial?
[00:37:42.13] Tarek Mansour
Well, insider trading is antisocial.
[00:37:44.11] Luana Lopes Lara
Right. That's a big one.
[00:37:45.17] John Collison
Yeah.
[00:37:46.10] Tarek Mansour
And illegal. But it's... And look, sniping is part of it, right? You need people that all of a sudden have gained some information edge and they do it in traditional financial markets all the time. But to have liquidity and make sure that people are there and they're investing the resources, they have to get, as you said, some incentives. But the interesting point, and I think this is part of why prediction markets are being adopted so much, is people like this idea that if your edge is proportional to your research, how informed you are, how much time and energy you put into this. And I think that only exists in prediction markets or traditional financial markets, except that for a lot of people, traditional financial markets, they're just less interesting. Like in here, you're researching about DOGE and what's going to happen or what the election and how people think about elections and how they vote. That, at least to me, feels a little bit more interesting than, let me think about IBM's quarterly earnings every quarter.
[00:38:50.16] John Collison
Kalshi has built a new kind of marketplace where real-world outcomes are traded, like whether the US will confirm whether aliens exist before 2027. You have thousands of participants opening, transferring, settling their positions all in real time. And underneath it, as you can imagine, there's a complex multi-party flow of funds. That choreography on Kalshi is powered by Stripe Connect—onboarding participants, processing payments, routing funds, managing payouts. When money movement becomes programmable, new products or even new market structures become possible. So, if you're building something new with complex money flows, Stripe Connect was built for you.
[00:39:29.13] Tarek Mansour
Sorry, guys.
[00:39:30.12] John Collison
No, you're good.
[00:39:35.16] Matt Huang
Let's talk different market verticals. And I think today everyone gets elections, they get sports, they get economic indicators. But I think you can look at prediction markets as this kind of search function across the set of interesting markets humanity wants to trade. And it's kind of a weird artifact that like the CME used to greenlight like wheat and oil and corn. But now you get to greenlight 1,000 markets a day. So what do you think we're going to find as we do that?
[00:40:08.02] Luana Lopes Lara
One thing that we're very excited for, we're actually starting to go in the direction of, for example, things like watches and bags and all of those things are more going to the collectible side. They're actually able to do derivatives on those things. One thing that you should talk about is the—
[00:40:22.12] Matt Huang
Yeah, I feel like compute could be a huge one.
[00:40:25.02] Luana Lopes Lara
And I think the compute that... What we're thinking a lot about is that there's a lot of these types of things that they function better as a more traditional future. So things that don't have a binary, "Will it be at this price? Yes/no." But it's more like an actual future. You can have margin, you can have more like institutional grade liquidity and all of that. And I think that that's a great example of like when we start going more in outside of binary markets and more into the traditional ones, then what we're doing is expanding kind of that from grain to compute.
[00:40:56.01] Matt Huang
Because it strikes me that obviously the futures markets that have worked best are these large commodity categories. But we're sort of in an era where humanity is spending more money than it's ever spent before on a specific new kind of commodity. And so, and the other traditional markets don't seem to be attacking compute.
[00:41:17.11] Luana Lopes Lara
So the way that we think about—we want to be the biggest derivatives exchange in the world. And for that, when we think about product roadmap, there's four things that matter. The first one is breadth of topics of markets. So we think about compute, we think about sports, think about elections, think about securities, we think about all of that. The second bucket is really market structure. So right now we only have the binary yes/no. We want to have things like futures, like swaps, options, all of that. The third one is really margining systems. Right now it's very bad. You have to put all the money up front—
[00:41:52.00] John Collison
You have to tie up all the capital.
[00:41:53.14] Luana Lopes Lara
Which makes a lot of, for example, will a hurricane happen this year, very, very bad for you to be actually like market making or selling those contracts. Doesn't make almost any sense, like capital-wise. And then the last one is liquidity. And when we think about it, it's like if we win these, we have the broadest set of markets, we have the broadest set of market structures, we have great and very... I don't want to say cheap margin, but in a way around that and then good liquidity, we're going to win on everything that we do.
[00:42:19.05] Luana Lopes Lara
So everything that we do in the company is like, it needs to be in one of these four buckets. And I think a lot of the topic side is like, how do we actually match the right topic with the right market structure, the right margining, and how do we make sure that it's all coming together? But you're completely right. I think that being able to build all the margining systems, all those things from scratch, we're going to be able to do margin models a lot faster and list a lot of these kind of new markets a lot faster.
[00:42:45.12]
[00:42:50.16]
[00:43:39.13] Matt Huang
Because of your direct mobile app interface and the fact that you target a lot of retail, do you worry that sort of the markets you're going to gravitate towards are the ones that are most interesting to just retail? And how much do you think about—
[00:43:55.01] John Collison
As opposed to like the pro markets?
[00:43:55.22] Matt Huang
Yes, the kind of institutional markets. Like I think of compute as much more of an institution-to-institution market. How do you think about building liquidity and interest upmarket?
[00:44:04.16] Luana Lopes Lara
Yeah, we almost divide the company again, in a way. We divide the markets and like sports, crypto and everything else. But in how do we make what we have great but very new things? Because I think what got Kalshi here was not the regulatory side, was not... It was really that we're just really pushing what is the next thing. It was elections and then after election, sports. And for us it's like, we need to be pushing what the next thing is and doing that very well. And I think that if we stop doing that, we're not going to win. The company is kind of structured that we have the market operations, we have the engineering side, all of that that's set up for maintenance and improvement of what we have. And then the new teams like institutional, the margin team, international that are kind of pushing forward. And we just try to kind of find a balance on those teams and then a platform layer that is like the core exchange and compliance and all of that.
[00:45:00.14] Luana Lopes Lara
But it is tricky because we're still like 120 people to do it.
[00:45:05.07] Matt Huang
Are you seeing pull already on the institutional side and certain topics?
[00:45:09.17] Matt Huang
Yeah, for sure.
[00:45:10.18] Luana Lopes Lara
And I think that we are...We actually just launched a week ago this thing called block trades. I don't know if you know, it's a very institutional way to do it that I can call you and negotiate a trade. And then we go and put it on the exchange versus trying to do everything that way. So we're trying to build a lot of features to start getting more institutional side.
[00:45:27.18] Matt Huang
Are they trading the same things that are on the Kalshi retail or are you offering new types of products?
[00:45:32.19] Luana Lopes Lara
Yes and no. So whenever they're interested in something like the... There was a lot of interest on the tariff situation. Is there going to be a tariff or not? A lot now with the petroleum, like the reserve and kind of how that's going to go. So whenever we hear, "We want to trade this market." We just list it directly and then it's accessible to everyone. But I do think there's going to be a very big gap on what the institutions are going to end up trading versus not. But we just list it to everyone. It's very cheap for us at this point to list new markets.
[00:46:00.02] John Collison
In the early days of Uber, it wasn't bad for the taxi business because it was just excess capacity and, you know, it was serving unmet needs. But then after a while, it was bad for the taxi business. Are there existing businesses that will feel the effects of Kalshi and other prediction markets because it's a bigger market, there's more liquidity. Like, I can think about the just existing futures exchanges. Maybe Kalshi is a better place to hedge your soybean prices or what have you. There's sports bookies, obviously. There's political polling firms where maybe you can get kind of the same information way cheaper. So who do you think will start feeling the effects of prediction markets? Because they have been in some way superseded.
[00:46:43.22] Luana Lopes Lara
Yeah, there's that funny meme of like the guy knocking on the door and it's like, "Who's the next one?"
[00:46:49.11] John Collison
The grim reaper.
[00:46:49.11] Luana Lopes Lara
So you don't want to do that. But I think that a lot of what you mentioned, that just traditional betting that we talked about, all the issues that industry has that we're very different from. There is a traditional futures, now we're going way more into their space. So I think there's going to be a difference there. There is the political polling that I think since the last election, there's just a lot of campaigns are using our data and all of that. There's parametric insurance. Once we have margin, we can start going to more hurricane, natural disaster insurance, all of that side.
[00:47:19.03] Matt Huang
Is there a tragedy of the commons with the polling? Like, part of the reason the prediction markets are accurate is because they interpret the polls, right? So like, if people stop using polls, in some sense, polls are the sensor that you get of what people's opinions are, and then the prediction markets are like the mathematical interpretation of the polls?
[00:47:41.07] Luana Lopes Lara
Yeah, my take is that polls are just going to get a lot better because what people are going to be is like, "Ok, I can make money if my polls are right, so I'm just going to commission this poll and I'm going to do this." And now you can actually compete a lot of polling models into one market.
[00:47:53.23] John Collison
It's like FiveThirtyEight did, kind of a meta-poll interpretation.
[00:47:57.07] Luana Lopes Lara
Exactly. And you can have one number that's aggregating all of that. And even in the last election, there was someone that actually did this. They commission a specific poll to do nearest neighbors type of thing. It's like it was... I don't know how it was, but a different type of poll. And then they were able to make a lot more money in the markets. And that's the whole point of like having money and skin in the game aligns the incentives with truth. And then the polls are not just paid for tell me what I want to hear but the real numbers. So I think it's complementary. Same with the news. A lot of people are like prediction markets will destroy the news. I think it's way more complementary. It's like when you're talking about an election, you're going to give your opinion. The market is not going to give you an opinion. You still need the commentators, but they're going to be able to show a number and be like, this is what the forecast is and this is my opinion on it. I don't think the opinions are going to disappear, but—
[00:48:45.08] John Collison
You guys referenced insider trading earlier, and there's just the policy question as to what the right policy should be around insider trading when it comes to prediction markets. I think it's pretty nuanced, like it's nuanced in the stocks case, right? Where famously there's lots of, you see SEC enforcement actions all the time against the things that aren't allowed. But there are cases like a hedge fund can have proprietary satellite data of the Walmart parking lot and use that to trade earnings. And that is information that only that hedge fund has. But that is permissible. And so similarly, I think there's a complex set of line-drawing exercises here, where presumably we don't think government officials should be trading in advance of military actions. What about leading up to the Super Bowl, you know, predicting the Bad Bunny halftime show length? I mean, you know, people have that information, right? And so where do you think the lines should get drawn on insider trading?
[00:49:43.10] Luana Lopes Lara
Yeah. And as you said it perfectly, it's a very complicated question, and it's a complicated question for stocks. Way more into a bigger scale than it is in prediction markets. The line that we take now is that we follow what the federal law is. So it's basically if you have—
[00:49:58.20] John Collison
And these are CFTC rules.
[00:50:00.17] Luana Lopes Lara
Well, CFTC and SEC, they both have it. So basically, if you have signed an agreement that says you cannot share some part of data. So if I work at the Bureau of Labor Statistics and I have in my confidentiality that I'm not able to say what the inflation number is before, then you cannot be sharing that data. But if you know that they're going to be rehearsing the Thursday before the Super Bowl and you're outside and you're like, "I hear Lady Gaga singing," that's fine. And that's the same thing that a lot of hedge funds do with Starbucks and people noting, ok, there's more people, fewer people in the store. And that's the point. Markets are very good at incentivizing information. We want information to come to the markets, but we don't want it to be unfair. And if you have access to it in an unfair way, you should not be trading on it.
[00:50:43.08] John Collison
Ok, so you cannot trade on information where you have some duty to hold that information confidential.
[00:50:48.14] Luana Lopes Lara
We actually take it even a step further. For example, if you are a government official, like if you're in Congress, you cannot trade on bills passing, even though I don't know if they have an agreement.
[00:50:57.23] John Collison
Famously, congresspeople can trade on stocks.
[00:51:00.01] Luana Lopes Lara
Right, right. So we're actually like taking a step further there. And we're working a lot with the regulators because obviously it's a very new problem for them and for us. But we have an entire... Part of being regulated, we have an entire surveillance division that is looking at every single flag. They pretty much don't sleep and they just try to figure out everything. And we put out two cases two weeks ago of two insiders that then... Because we're also regulated, we're able to charge them a lot of fines. So we charge them over five times what they made and all those things. They're banned and all that.
[00:51:30.12] John Collison
What's very interesting to me about these stories is that you guys were doing that where with the public equities markets, the SEC is extremely enthusiastic about enforcing their insider trading doctrine. Just what has the CFTC been like on this topic?
[00:51:47.11] Luana Lopes Lara
It's a great question.
[00:51:48.08] Luana Lopes Lara
So obviously the CFTC, you can think about it at three steps, right? The first step is our own surveillance and enforcement. Then the next step is it goes to the CFTC and their own surveillance and enforcement. The last step, if you go to the Department of Justice. And I think the biggest difference is when people look at the SEC cases, most of the time they take a long time because the exchanges did their research and their investigation and they put some fine, they block someone, and it kind of goes through the process. So it's still like that. Every single trade on Kalshi goes to the CFTC. They have every single thing, every single case goes to the CFTC for them to review. So they might take action. We don't know. But now the—
[00:52:27.02] John Collison
Ok, so you refer cases to them?
[00:52:28.13] Luana Lopes Lara
We refer cases to them. But we kind of do our first step of our first level of protection there.
[00:52:35.08] Matt Huang
I'm curious, there are clearly markets where nobody knows the answer yet of some event in the future, so like it's sort of impossible to insider trade. And then there are markets where a single person can change the outcome, like the mention markets in a speech or maybe a sports player doing a specific number of shots. So how do you think about that spectrum?
[00:52:58.12] John Collison
Like, are mention markets a bad idea because they're just so inherently gameable?
[00:53:01.17] Matt Huang
Yeah, or are they like limited in scale fundamentally because—
[00:53:05.13] John Collison
Mention markets like the Brian Armstrong Coinbase earnings thing.
[00:53:07.23] Luana Lopes Lara
Yes, I think that, well, inherently I think major markets are actually great if you think about the Fed. The amount of hedge funds that are just sitting down being like, "Is he going to tilt his head this way or this way? And if he does that, it means he's not very sure."
[00:53:23.13] John Collison
Fed meeting minutes are the original mention markets.
[00:53:25.15] Luana Lopes Lara
Exactly. And it was because we just know that specific words being used mean very specific things, and you can move the market so much. The same thing with Trump. If Trump says we're going to war, that's going to move the markets a lot. Or if he says a lot of things like tariff, everyone knows moves the market a lot. Even in... Yeah, just a lot of everywhere, things that people say move markets and move a lot of different things. So I think the mainstream market is very important. Obviously, the person that is working on the speech or that is saying the speech cannot trade. And that's kind of how we enforce it. If you are Gavin Newsom, there's a market on what you're going to say, you cannot trade it, and your staff cannot trade it.
[00:54:08.04] Luana Lopes Lara
That's part of the political kind of cuts that we do there that they cannot trade. But I think the point is like, if there is a way to restrict some players in the market so that the market's fair and the market's positive and there's an economic utility for it, the market should exist. We shouldn't say "Ok, there are five people that could manipulate the market and the market shouldn't exist." Then you say the stock market shouldn't exist, right? So I think it's more about how do we build a system that is strong and resilient enough and with the right prohibitions that you can have the market.
[00:54:38.07] John Collison
The other big debate you guys are in the middle of is just that about sports contracts generally. And I was trying to reason about my own thoughts here on this debate, where on the one hand, the criticism is that with more sports gambling comes proven bad effects. You can measure some of the bad effects that it has on people. And, you know, we have this especially in the US, where there was a lot of legalization of sports betting over the past 10 years, and there's some data on that. On the other hand, I have no real issue with alcohol, despite the fact that it has kind of a similar distribution where many people enjoy it and then there's a very bad set of outcomes for a small fraction of the population. And so I feel like the societal discussion of, you know, the morals of alcohol and the morals of betting are different, despite the fact that, again, it's a similar shape to the distribution. And, you know, also just thinking from my experience, online sports betting has been legal in—well, legal is a complex term—has been available in Europe for a very long time, basically since the start of the internet.
[00:55:52.06] John Collison
And there was all these cross-border hacks in Malta. And now it's a bit more regularized. But it's been available to people who want it for a long time. And life goes on. And it hasn't led to any kind of major societal collapse over there. But clearly, this is one of the big debates that raged around Kalshi. And so I'm curious how you guys think about increasing access to sports contracts and the effects there.
[00:56:16.10] Luana Lopes Lara
Well, there's a lot of parts to what you're saying. I think that the way that we think about sports, how we decided to first list sports, is obviously something that a lot of people are interested in. That's unquestionable. But also there's something that a lot of people do a lot of research and know a lot about. And there aren't traditional ways for you to make money on that. That are actually good, or we talk about the winners, they get cut and all those things, and it doesn't really work. And regardless of whether people like that some people bet or dislike that some people bet, people bet. And it's just about what is the best way for them to have access to something that they can get exposure to sports. And I think that the whole point of markets versus a bookie is that markets are just objectively better. I think that it's almost... I'd never heard someone make a case that a state-by-state regulated casino is actually a good thing. I'm actually hearing nowadays that a lot of the paid propaganda by the gambling guys trying to say that.
[00:57:10.17] John Collison
And just to put numbers on that, the order of magnitude rake for sports betting companies is around 10% and the order of magnitude for prediction markets is, you know, 1% or a few points.
[00:57:22.23] Luana Lopes Lara
Right. But the predatory part doesn't even come from that. Like for sports betting, if you start losing—because they want the losers, they don't want the winners—if you start losing, the first thing that they're going to do is give you a bonus. They're going to be like, "Oh, here it is." Oh, look at that.
[00:57:36.03] John Collison
Welcome back.
[00:57:38.09] Luana Lopes Lara
We were talking about sports, and what they do is you start losing and then they're going to give you $1,000 for you to come back or a deposit boost and all those things. So that they can hook you to keep you coming back because they want to incentivize the losers. We don't do that.
[00:57:52.06] John Collison
The people losing the most money are the most profitable for sports bookies, which creates a bad incentive.
[00:58:01.13] Luana Lopes Lara
Yeah, and we don't have that at all. And I think that the whole point's like, there's a moral question. Like, some people are going to go into Robinhood or Coinbase or whatever and speculate on stocks and speculate in crypto and whatever they want to do. And some people want to speculate on sports, and they should have the best, the access to the best possible thing for that. And right now it's just the sportsbooks are just not it. And we really firmly believe that what we do in our markets are significantly safer for all of that. And if you just take a stance of prohibiting, it's like same what you said with alcohol. It didn't change. People just went to a speakeasy and drank. And people are just going to go offshore when there's way less protections. There's none of the self-exclusion, deposit limits, all those things that we do. Don't have any information about them. And they're going to actually... It's actually very bad. Bad for them. So I think it's just this prohibition concept just never really works.
[00:58:50.02] John Collison
It's also the whole policy discussion around this stuff is also very interesting when it kind of reminds me of, in Canada all the liquor stores are run by the government, or at least in British Columbia. And you have the government saying this must be very carefully controlled, but also we will sell it to you and have the revenue source. And obviously that's much more of a factor in lotteries and things like that.
[00:59:12.11] Luana Lopes Lara
It's all about money at the end of the day. The states want their money, the casinos want their money. It's just, yeah, it is what it is.
[00:59:18.17] Matt Huang
Speaking of sports, one thing we were talking about while you were gone was just what interesting new verticals are there going to be? And so I'm just curious, which ones are you most excited about?
[00:59:28.07] Tarek Mansour
Well, I think that anything around dissecting like a stock into its more atomic components.
[00:59:38.20] Matt Huang
So this would be like betting directly on Nvidia GPU shipments—
[00:59:44.06] Luana Lopes Lara
Versus Tesla deliveries or whatever.
[00:59:46.03] Tarek Mansour
Their earnings...Because then you can expand that to things like dissecting the macro economy and what are the main sort of like factors are influencing the economy broadly, like AI. And I have a series of questions to price what's going on with AI. Things like health scares like COVID and so on. But the sort of where things get really interesting is this idea where... So there was a paper written by Kevin Hassett around this idea that like as society gets increasingly more complex, our asset prices, our understanding of asset prices will naturally decay. Entropy will go up because the things that influence or the vector that, the number of factors—
[01:00:38.05] Matt Huang
Much higher dimensional vector.
[01:00:39.17] Tarek Mansour
Becomes very high dimensional. And if those dimensions you don't have a good understanding of x1 to xN, you cannot get a good estimate of Y. And so the paper basically says you need infinite markets. And prediction markets started this notion of infinite markets, which is like you have to have a market for each one of these Xs so that you can then take that back into pricing traditional asset, getting good traditional asset prices.
[01:01:07.09] Tarek Mansour
And an example of that this last week, so you know that there was the Citrini put out a research report. And it got a surprising amount of—
[01:01:17.09] Luana Lopes Lara
People got obsessed with it.
[01:01:18.15] Tarek Mansour
Yeah, I would say it got a surprising amount of love and interest.
[01:01:21.06] John Collison
This is the AI 2028 we're all doing.
[01:01:23.16] Luana Lopes Lara
Yeah, 38% on a bull run.
[01:01:26.10] Tarek Mansour
I think there's a little bit of like our society wants to believe that AI is going to end us all. I think right now there's a bit of that. But this is where... And that impacted markets. Like the stocks got, there was a sell-off. And so we launched a prediction market on that. And well, before that, Citadel came out with a rebuttal and we launched a prediction market on that. And, you know, the odds are 10%, right?
[01:01:46.13] John Collison
Of the economic scenario that they predicted being true as of 2028.
[01:01:51.03] Luana Lopes Lara
It's three out of five things. So they have like five conditions. If three hit, we—
[01:01:53.11] Tarek Mansour
Yeah, if three out of these five conditions hit, you could reasonably say that, ok, this outcome has somewhat materialized. And it's just 10%. Five out of five is much lower. So that is important. If you can put that back into pricing models, maybe the markets wouldn't have reacted, maybe people wouldn't have sold off DoorDash. Because at least I believe, but you don't have to trust me, maybe you should trust markets, that this sort of analysis around DoorDash was actually quite poor.
[01:02:19.17] Matt Huang
So one thing you're sort of envisioning is this world where everything has a price all the time. And I'm curious, is that a good world to live in? And I would note that Stripe, I think, benefits from being private and not having the real-time price and smoothness for employees and comp and all that.
[01:02:38.23] John Collison
I think there's some noise in the sub-second pricing.
[01:02:40.12] Matt Huang
Right. Because clearly sentiments swings publicly. Like there's on average in the long run, it's a truth-telling machine, but in the short run can be a panic. So just curious for you guys to think about, yeah, is that the world we want to live in?
[01:02:54.21] Luana Lopes Lara
I mean, we are obviously biased because we love markets. We think markets are good, so we're definitely biased here. But I think our view on this is that it's always better to have more data than less data. If you don't think the data is good, if you don't think the second-by-second stock price is good, you can just choose to ignore it. It might... The world might not just ignore it, but you could use it.
[01:03:17.04] John Collison
I think CEOs of public companies would say it is not always possible to choose to ignore it.
[01:03:20.22] Luana Lopes Lara
That's fair. But in a way, it's like it's better to have the data and then use it as an input to something than not. But when we say we want to have prices on a lot of things, it doesn't mean everything. There are a lot of things that we wouldn't do, like wildfires we don't do. War, terrorism, assassination. Those things are bad. There's a moral side of these markets and we're not going to ever go there. But in general, it's in a world of social media is like, you don't know what's true anymore. My feed is like, is it real? Is it not real? Did this happen? It's just better to have a source, an unbiased source of information that you can kind of use it for other things. And I think that that value is there.
[01:03:58.10] Tarek Mansour
I think that maybe the simple frame for this is you are increasing market efficiency for all these questions. That's what's happening. It's including potentially some events or things that relate to maybe private companies. And I was just thinking about the question, it's an interesting question, why do companies go public? Why is it important to get a real-time market price? Because there are downsides. Sometimes markets are erratic. Sometimes they overshoot in either direction. But the market on the long enough time horizon is a good sort of allocator. It's a good weighing mechanism. It's a good allocator of capital. And I don't really see that... I just think that pricing a lot of these questions will just increase efficiency. Make our function, our allocating function better over time. And there will be some net losers. Like some people that maybe capital shouldn't be allocated to.
[01:05:03.17] Luana Lopes Lara
It's also a good feedback loop. If you're a CEO of a public company and you announce something and it just keeps going down, you're like, "Maybe I'm wrong." And I think the same thing you see with politicians where you can see in the live debate, if they say some answers and they see their prices going lower, they're like, "Well, maybe the answers aren't great." And I think a lot of these things, when we see even the, for example, the use case of prediction markets in government, a lot of it is conditional markets. They can say, if we pass this bill, will unemployment go up or down? And you can price these things for better decision-making and just like a tighter feedback loop tied with good incentives.
[01:05:37.16] John Collison
Do you guys feel like we have started to see... Like, clearly we have seen the effects of social media on politics, where politics is a different game now. And different politicians are popular and the political discourse is changed by the existence of first Twitter and now to some extent short-form video. Do you guys feel like we have seen the effects on politics of prediction markets yet?
[01:06:04.06] Tarek Mansour
Definitely. I mean, the—
[01:06:05.17] John Collison
What are they?
[01:06:06.16] Tarek Mansour
Well, the candidates are using the prediction market prices.
[01:06:09.21] John Collison
Sure, but that could just be like a handy thing, whereas again, I think with social media, the candidates are different, the debates are different.
[01:06:15.18] Tarek Mansour
Reflexive.
[01:06:16.23] Luana Lopes Lara
I do think that what the markets are good at is that they are more unbiased by party dynamics. So if there is an underdog that the public really likes, there's a lot of... Maybe there's like a party that's like, "We don't like this guy, we like this guy from the establishment." But the markets are very good at actually showing the real odds for that person.
[01:06:36.05] John Collison
Ok, so you think the party machines have lost a little bit of power?
[01:06:39.12] Luana Lopes Lara
I think you can shine more light into what people really want, which might not be necessarily what the parties want to put forward. I don't think we've seen that necessarily yet, but I feel like if I were to say, there was a, for example, the Texas primary, and I think that the polls were saying one person was really going to win, another person that was way higher in prediction markets won. I think a lot of it was more of like, they give a more, a fairer view of the state of the race than a lot of the party polls.
[01:07:08.02] Tarek Mansour
I think that we do see this a lot with also... There's sort of the piece where people use it and that's sort of reacting in real time to certain things. But I think there's some degree of depolarization, and that's what Luana's alluding to. And in some ways, it's an antidote to social media. Social media has really polarized. When you have two candidates running for a Senate race, we're sort of set. Your feed is set. Either your feed is saying the Republican candidate is awesome, or the feed is saying the Democrat candidate is awesome. Prediction markets don't really have that. Because the people that are engaging in this are not in the sort of like who's great and who's, you know—
[01:07:48.11] John Collison
I think what you're saying is social media feeds ultimately try to resolve upfront. They're like, "I need to figure you out. Are you a Democrat? Are you a Republican? Like what post should I show to you?" And pretty quickly, and you know, people have complained about this, of this phenomenon where you end up down a particular rabbit hole because they have pigeonholed you as kind of this type. And you're saying just prediction markets do not have that phenomenon.
[01:08:08.13] Tarek Mansour
There's a little bit of a reverse phenomenon. It's like, are we feeling too certain about this person? And I think that depolarizes things because the dimension is not, we're not one-dimensional anymore, which is like Republican, Democrat. And that's what you're seeing and what you're hearing about. Now it's like, "Well, this guy's kind of cool." And that's Talarico, and maybe he might do good in Texas even though he's a Democrat.
[01:08:29.10] Luana Lopes Lara
The same thing happened with the New York mayor situation. Everyone was like Cuomo was going to win 100%, 100% Cuomo was going to win, there's not even a chance. And we were just seeing Mamdani's odds going up. And I think it's just the progressive message was really working with New Yorkers and the markets were really seeing that uptick. And that decrease in polarization really comes from people taking a step back and being like, "What do I actually think is going to happen?"
[01:08:51.16] John Collison
So there's a bit of like an Iowa-New Hampshire Iowa-New Hampshire effect here, where in presidential elections in the US, there's like some big name leading into the election. You know, maybe it was Hillary Clinton in '08. And then Iowa and New Hampshire are measurements of the sentiment of those two states, but they also create narrative. And I think what you guys are saying is prediction markets create this Iowa-New Hampshire effect, where they can create narrative in a way that changes the ultimate outcome.
[01:09:21.10] Luana Lopes Lara
I don't know if it changes, though. I wouldn't say it changes the ultimate outcome. I think it sheds light into what the ultimate outcome is going to be.
[01:09:28.04] John Collison
But it potentially changes it.
[01:09:29.18] Matt Huang
There clearly is some reflexivity, right?
[01:09:31.21] Tarek Mansour
I mean, there's always, but it's like with the polls too. Like, yes, I mean, there's polls—
[01:09:36.18] John Collison
Aren't you guys hiding your lamp under a bushel here? You're like, "Well, we're not changing anything here."
[01:09:41.12] Tarek Mansour
I just think that the one thing I would say, the reason why we're like—
[01:09:45.13] John Collison
Prediction markets are a big deal. It's ok to say they'll change things a little bit.
[01:09:48.16] Tarek Mansour
High odds don't always correlate to like a good outcome. So you saw Mamdani when his odds were 94% on Kalshi, the thing that he was mentioning pretty consistently, and I think there was a little bit of worry there, it's like you've got to show up. Because if you're very high odds, that could also lead people to be like, "Ok, this isn't it." So it's not as clear. And I don't think this changes things more than polls changes. Does that make sense? So the response is more like in a vacuum. If you had nothing else If you didn't have social media, if you didn't have news, if you didn't have any of that, yes. But because we have all these other things and you add prediction market to it, like the impact model... But I want to make one point that I think is actually very interesting, and we're seeing this often.
[01:10:26.18] John Collison
We'll be the judge of that.
[01:10:28.12] Tarek Mansour
Ok, you can judge that. Let me know if it's interesting. But a lot of times when you see people start participating in prediction markets, they get more engaged in the underlying. They legitimately just get more informed.
[01:10:40.23] John Collison
It pulls people in.
[01:10:42.02] Tarek Mansour
It pulls people in, but to do research. Now you have... Because you have some skin in the game, or you're about to put some skin in the game, you read, everything changes. You're not just like saying something crazy on Twitter anymore. You're putting money. And now it's like, "Let me read, let me actually figure out, oh, who's this person, what's happening?"
[01:10:56.04] Luana Lopes Lara
It's like, are they pro this, are they against this, what's their view on everything?
[01:10:59.02] Tarek Mansour
And you go... It's amazing because this happened a bit in the New York race. Or it happened a bit with Brexit. People voted for Brexit. And after they voted for it, they're like, "Oh God, wait, I don't think we wanted this. Wait, we didn't even understand what we were voting for." And I think this heightened engagement, it engages people further to learn and understand. You know, in sports, basically, if you ask a lot of the leagues, they would tell you the same thing, but people got more engaged with the statistic, which player is good, what's happening. And I think that will happen, is already happening, it's sentiment in politics, which is a good thing.
[01:11:32.10] Luana Lopes Lara
I think that the I'm going to say something and then I'm going to qualify what I'm saying. I think that politics are going to get better because you're going to have a way faster feedback loop on the messaging and the policies. Right now, when a candidate says 10 things and they win or they lose, you're trying to make one assessment of so many things that the candidate did. And did that go well or not? And which point was it? And even if you do a poll, it's like always delayed. It's a very specific example. But now you can have real-time of they said this, what was the response? And you can get that and that faster feedback loop, which I think makes startups great. You're able to iterate very fast. And I think if candidates are able—everyone wants to win at the end of the day, and if they're able to optimize their message to what really people want and what policies people want, I think they're going to end up being better because they're just going to know what people want better.
[01:12:20.21] Tarek Mansour
It's like you get a score on all of them, all the different things you've done, not one score that encompasses the good and bad.
[01:12:26.04] Luana Lopes Lara
Which is basically like when you ship a new feature, you're able to have like 10 metrics. And you're like, "This went up, this went down." And you're able to, markets can kind of contribute to that. But also I think to a lot of other things as well. Music charts. When someone listens to the song and they're like, "Definitely not going to hit number 1," you're like, "Ok, that was that song. So we should do something else."
[01:12:45.19] Matt Huang
Do you guys try to use prediction markets in any way internally to make decisions?
[01:12:50.19] Luana Lopes Lara
Every single decision we make is always probability. Even the election lawsuit, right, when we were doing it, it's like—
[01:12:56.13] Matt Huang
But that's you guys evaluating the probability. Do you ever think about creating markets for your employees to participate in?
[01:13:02.13] Tarek Mansour
So we have one nit. Which is, as a regulated exchange, we can't trade.
[01:13:09.19] Luana Lopes Lara
But we do it like internal markets.
[01:13:10.21] John Collison
There's like a separation of church and state thing.
[01:13:13.01] Tarek Mansour
Exactly, exactly. And so we've been asking, we've been working with regulators, could we do something small? We do small dollars where, so that, you know, because obviously that's one that we really want to do.
[01:13:23.20] Matt Huang
And you can't even dogfood the product, I guess.
[01:13:25.09] Tarek Mansour
And then dogfooding the product. Right. Which has been hard. It's hard.
[01:13:29.18] John Collison
Sorry, employees cannot trade in their personal capacity?
[01:13:32.19] Tarek Mansour
They can't.
[01:13:33.06] Luana Lopes Lara
Not at all.
[01:13:33.18] John Collison
Oh, interesting. Yeah. But yeah, that makes it very hard when you, as you say, you just can't dogfood your—People at Facebook use Facebook and that's how you make the product good.
[01:13:42.15] Luana Lopes Lara
So that's why it's so important for us to you should just be asking the users all the time.
[01:13:47.02] John Collison
Yeah. Oh, that's so interesting.
[01:13:48.13] Luana Lopes Lara
Yeah. It sucks.
[01:13:49.05] Tarek Mansour
But the power users, the superforecasters, are a lot of what influences sort of where it goes because they're very engaged.
[01:13:54.08] John Collison
You guys must presumably spend a lot of time with those power user, superforecaster types. Yeah. Just have them on speed dial.
[01:14:00.17] Luana Lopes Lara
Yeah. They were there day one. So, we want to make sure they're happy.
[01:14:04.21] John Collison
Last question. Where do you guys want to see prediction markets policy go? Like when you're talking to someone in government or if you had a magic wand, what are you arguing for?
[01:14:17.07] Tarek Mansour
So our stance as a company, and I think this may differ a little bit from I would say like your average tech company or big tech company. So we are pro-innovation and innovation needs to happen in America. You know, we have to lead and we have to do it right, and we have to win. We have to be, you know, all the things that Americans want to do or we need to win as a country, we should have here. But we're also pro-regulation. And so at a high-level principle, there's usually this tension that generally it's like the policymakers want to regulate—
[01:14:52.16] John Collison
And I think you're maybe more like a traditional financial firm in that way, right? Where maybe Silicon Valley, a lot of firms grew up in an unregulated way, but financial firms have always had a regulator. And that's just a fact of life.
[01:15:05.11] Tarek Mansour
Yes. So it's part of the culture. And I mean, again, we spent four years getting regulated up front. So, so it's part of... But we believe in regulation. I think it's important because, regulation is a bit like insurance. It's like it's protecting you from things going wrong at bad times. And so when I think about where this lands, I think anything that is oriented around preserving these in America and making sure we win, but then elevating the fairness and transparency of the market. Anything that's oriented like how do we make it more fair? Ban insider trading, add more restrictions on, for example, government officials, members of Congress trading on like, you know, information they shouldn't trade. You know, I'm obviously a big fan of like banning insider trading for members of Congress.
[01:15:44.09] Luana Lopes Lara
That's what we talked about.
[01:15:45.14] John Collison
Yeah. But presumably you mean just banning trading period for members?
[01:15:49.13] Tarek Mansour
I think it's not a bad idea.
[01:15:52.13] Luana Lopes Lara
That's how we kind of—
[01:15:53.13] Tarek Mansour
But I think... And then things around like creating also social fairness and transparency because of all the questions that we asked, if people are basically trading on politics, let's have all the trade data be as public as possible so anyone can audit it, anyone can see it, which is a good thing. You don't, now you don't know, imagine a poll where you can check every single person that basically got, the general public can check who was polled and what the sample was like. And then anything around customer protection. And because that is important long term in the sense that when you build a consumer product and it goes mainstream, there is a massive burden on the other companies to educate. And you've seen it over and over. And in our case, you want to make sure that people know what they're getting into. They're not overextending themselves in terms of how much they're sort of trading. They're not getting into an area of discomfort.
[01:16:50.11] Tarek Mansour
And how do you, we can do as much as we can do on the marketing and on the product side, but like we need policymakers and regulators' help to basically make it an industry standard, but also help us elevate ours.
[01:17:00.13] Luana Lopes Lara
And by the way, we're pro that. I think that like even the classic retail brokerages should also be adopting a lot of these customer protections that we're talking about that they don't. I think that every retail trading platform should be taking a lot of these steps.
[01:17:14.03] Tarek Mansour
Yeah, I mean, that's our general view, and we hope that this is sort of the direction that things take. Because there's kind of... You can have a variety of views. And some people believe that any type of speculation should be banned, whether it's in the stock market or crypto or prediction market. We don't believe that. We think that would be a bad outcome for all the reasons, because there's a lot of upside to having liquid markets and a variety of different things. But also because if you ban it, you're actually heightening the risks that you're trying to prevent because now that activity is going offshore where you cannot monitor it or police it or do anything to protect it.
[01:17:50.08] John Collison
Awesome. Cool. Tarek, Luana, thank you guys.
[01:17:51.21] Luana Lopes Lara
Thank you.