Video Author: John Collison
Translation: Peggy, BlockBeats

Editor's Note: Over the past few years, the prediction market has evolved from a relatively niche financial experiment to the center of discussions on technology, finance, and public policy.

It has garnered widespread attention not only because of the allure of "betting on the future" itself, but also because against the backdrop of amplified noise on social media, repeated polling inaccuracies, and the declining credibility of traditional information systems, a more fundamental question has emerged: Can market price become a signaling mechanism closer to reality than opinions, emotions, and narratives?

This conversation revolves around this question. The participants include John Collison, Co-founder of Stripe; Matt Huang, Co-founder of Paradigm; and the two Co-founders of Kalshi, Tarek Mansour and Luana Lopes Lara.

The two Co-founders of Kalshi, Tarek Mansour (right) and Luana Lopes Lara (left)

As one of the most prominent compliant prediction market platforms in the United States, Kalshi quickly gained attention during the 2024 U.S. election. Prior to this breakout, it had gone through years of back and forth with the U.S. Commodity Futures Trading Commission (CFTC) and ultimately, through a key lawsuit, paved the way for the legalization of prediction markets in the U.S.

The first part of the conversation focuses on Kalshi's journey: why the two founders chose not to follow the common "move fast and break things" approach in Silicon Valley, but instead insisted on "compliance first, growth later"; why they were willing to endure lengthy approvals, layoffs, and external scrutiny to secure the "election market"; and how the lawsuit against the CFTC became a turning point for the company's real takeoff.

The second part delves into the operational logic of the prediction market. Tarek and Luana explain the fundamental difference between Kalshi and traditional online betting platforms: it does not rely on a "house model" profiting from user losses but operates as an exchange with fees at its core, encouraging liquidity and information flow into the market. They also highlight a counterintuitive reality: Kalshi's liquidity does not primarily come from traditional large market makers but from a large number of dispersed individual traders, "super forecasters," and small teams. In a sense, the prediction market is not just a financial product but also a mechanism that directly transforms distributed cognition into price signals.

In the latter part of the conversation, the discussion extended to the future boundaries of the prediction market: from elections, sports, to AI, GPU power, macro variables, and policy paths, more and more uncertainties in the real world, can all be broken down into tradable, feedbackable, and decision-assistive market issues? At the same time, a series of unavoidable controversies have also emerged—how to define insider trading, whether sports contracts will amplify online entertainment risks, and how platforms and regulators should establish a new balance between innovation, transparency, and user protection.

It is for this reason that the significance of this conversation is not only about Kalshi itself. What it truly seeks to answer is: will the prediction market become the next-generation financial market, or the next-generation information infrastructure.

The following is the original content (for ease of comprehension, the original content has been slightly reorganized):

TL;DR

· Kalshi chose an unusual path of regulation first, growth second: Spending 3 years obtaining licenses, suing the CFTC to open the election market, the core judgement is whether the prediction market can legally exist, which is more important than growth.

· The essence of the prediction market is to incentivize real information with money: Compared to polls and social media, the market filters information through a profit and loss mechanism, seen as a signaling system closer to the truth.

· Regular people, not institutions, constitute the core liquidity of the market: Over 95% of matching comes from retail users and super predictors, rather than traditional market makers.

· Kalshi emphasizes that it is an exchange, not an online entertainment platform: Revenue comes from fees, not user losses, encouraging skilled participants, rather than restricting winners like the online entertainment industry.

· Elections are a prime scenario, but the future market extends far beyond this: From sports, macro to AI, power variables, the team aims to build a derivatives system where everything can be priced.

· The prediction market is becoming a new information infrastructure: Users are not only trading but also consuming probability; 80% of users mainly use it to assess the world rather than to bet.

· Behind its rise is a lack of trust in the traditional information system: Polarized social media and inaccurate polls are driving people towards price-based judgment mechanisms.

· Core Long-Term Goal: Improve societal decision-making efficiency, not just be a trading platform. Through continuous pricing and feedback, enabling faster formation of true consensus in areas such as politics and economics.

Interview Compilation

John Collison (Stripe Co-Founder & Interviewer):

Tarek Mansour and Luana Lopes Lara are the co-founders of Kalshi. Kalshi is an emerging prediction market company that gained rapid popularity during the November 2024 U.S. election. To establish the first onshore compliant prediction market in the U.S., they spent four years navigating regulatory bodies and seeking approvals before their official launch. Today, Kalshi's monthly transaction volume exceeds $10 billion.

So how do you two typically divide your work? But more than division of work, what I'm curious about is, do you have different perspectives on problem-solving?

Luana (Kalshi Co-Founder & COO):
Actually, our backgrounds are almost identical. We both studied math and computer science at MIT, had similar internship experiences, and were basically indistinguishable. But I am a very optimistic person, I like taking risks, and I always believe things will work out in the end; he, on the other hand, is very cautious, almost a bit pessimistic. So, I think this happens to create a very good balance. Looking back, apart from our daily division of work, what truly complemented between us is this.

Tarek (Kalshi Co-Founder & CEO):
Let me add a bit more background. I originally intended to become a trader, and that was nearly the career path I set for myself. If you've interacted with such people, you might understand that they always have an expected return calculator in their minds.

Matt Huang (Paradigm Co-Founder):
A very typical trader.

John Collison:
Yes, but—

Tarek:
If you are truly that kind of trader, you're constantly thinking about tail risks, the worst-case scenarios. She usually doesn't think like that. I feel it's actually this difference that brings about good outcomes.

Compliance First, Growth Later: Why Kalshi Chose the Hardest Path

John Collison:

I was just about to ask about this. Your starting point is very interesting. After founding Kalshi, for several years you couldn't really operate until you received approval from the CFTC. Most companies don't start this way. On the other hand, there's a very common, albeit often criticized yet widely prevalent pattern in Silicon Valley, where companies like PayPal and Uber initially just did it and figured it out later—start by getting things done, then later fill in the structure and permits, following a model of ask forgiveness later, not permission first.

So can you talk about how it all started? How did the whole approval process go? And I also want to discuss whether this path is also suitable for other companies.

Luana:

I think from the very beginning, we were very clear that if you are in financial services or healthcare, you can't just dive in without a plan. In the financial sector, once user funds are involved, the cost of failure is very high, as seen with FTX; not to mention healthcare, which has numerous catastrophic precedents. We wanted to do things the right way. More importantly, when we looked at this market, the core issue wasn't about whether this thing would grow but rather whether this could be done legally in the U.S. So we decided to address this major issue head-on before proceeding. For a long time, many people thought this was the wrong strategy.

I think before we won the lawsuit regarding the electoral contract, everyone kept saying that those who went to offshore markets were doing better and growing faster. But when we won that lawsuit, proving our understanding of the law was correct, demonstrating that this company could operate legally in the U.S. as we envisioned, that's when things really started to take off.

John Collison:
What was the timeline like? When did you start? And when did you win the lawsuit regarding the electoral contract?

Luana:
We founded the company in 2019 and joined YC that year. It took us three years to finally get regulatory approval and launch, which was around 2022. Later, we won the lawsuit regarding the electoral contract at the end of 2024, and that's when the company truly began to accelerate.

Tarek:
There are actually two aspects to this matter. First, it's a very practical consideration. We felt that if we wanted real mainstream and institutional adoption, the core issue we couldn't avoid was whether this could operate within a regulated, trustworthy, and secure framework. After all, it's a complex market involving the circulation of user funds. We had to tackle this toughest problem first, as that was the path to success.

The second aspect is more principled. What excited us initially was that when we wrote that one-page document on Google Docs, we listed a series of questions: Why are we building this company? Why does this matter excite us so much? Our answer was that we wanted to create the next-generation New York Stock Exchange. We aimed to establish a trustworthy, regulated financial market on U.S. soil. We weren't thrilled about creating a similar thing offshore. The key is, what kind of company do you want to build? Why are you doing this? There are many paths to success, but the other path is not the one we truly want to take. We want this to happen here, on U.S. soil.

John Collison:
You are the first prediction market to receive CFTC approval and reach a certain scale.

Tarek:
Yes, that's right.

John Collison:
And up to today, each of your contracts still needs to be individually approved, right?

Luana:
Yes. Every contract of ours is submitted to the CFTC, and they have 24 hours to halt it.

John Collison:
So, they are almost receiving your contract information flow in real-time?

Luana:
Yes, you could understand it that way.

Tarek:
Yes. To reach the current state of this contract processing network has actually been a very long process. You have to imagine, the first time we walked into the CFTC building, this concept was all in our minds, and the regulators had to keep up as well. Because you're discussing a product not backed by traditional financial underlying assets and facing the possibility of dozens or even hundreds of contracts every week. Now, of course, we have done much more, but initially, this regulatory model was not prepared for this scenario at all.

So, this process is actually very similar to iterating on a product, except you're not making a product for customers, but you're exploring with regulatory agencies how to regulate this thing. What are their concerns? What can we do to address these concerns?

Luana:
In a sense, this is about finding a regulatory-market fit.

Matt Huang:
So now you are more accustomed to this rhythm. You send out the contracts first unless they explicitly stop you. Have they recently vetoed anything?

Luana:
Not recently. The biggest rejection was actually the election contract, so in the end, we had to sue them. They rejected us around that for two years. But now, we have worked with them for too long, and we both understand where the boundaries are, they trust us, knowing us as a self-regulatory entity, understanding what can and cannot be done. For example, markets like war, assassination, we don't do. As long as everything is within the established boundaries, the whole process will be much faster.

John Collison:
So let me confirm, the core of that election lawsuit was that they are generally willing to approve various contracts, but they are not willing to approve contracts like who will win the election, which happens to be the most popular type, especially during the U.S. presidential election. So you sued the CFTC.

Tarek:
Yes. It's actually their own rule—

John Collison:
And generally, suing your regulator is not considered best practice.

Tarek:
Exactly. The thing is, we started pushing for the prediction market since the end of 2021, initiating conversations with policymakers, engaging with Congress, regulatory agencies. Everyone would say, "That sounds great." But then they never really pushed it forward, and we started feeling off about it. By the end of 2022, they effectively delayed the approval until after the election, essentially pocket vetoing it. It was a very tough time for the company; we had to let go of many people. Even harder was that the team, investors, and most investors, even, started losing faith in this path.

John Collison:
Not in the idea itself but in the strategy.

Tarek:
Right, they no longer believed in the strategy, even starting to question the idea itself. People felt like things were a bit unhealthy, should you be doing something else? Clearly, this path seemed undoable. But we just couldn't force ourselves to do something else; we really couldn't. So we said, okay, let's try again.

You can imagine, at that time, the team's morale was at an all-time low, everyone was waiting for a new strategy. Many people left, many were let go because we had to downsize. And then at the next stand-up, we told everyone, the strategy for 2023 is—let's try again.

John Collison:
So, we're going to keep doing the same thing, but this time, it will work.

Tarek:
Yes, that's exactly it, this time it will work. Even though almost all evidence points in the opposite direction. I have to say, a large part of this was really driven by her. Of course, I also really wanted this thing to succeed, but my rational brain kept saying, listen to these people, this path is not viable. But she was more steadfast. So we tried again. By the end of 2023, they blocked it again. By then, I almost felt—

John Collison:
Well, this prediction market thing is just not achievable.

Tarek:
Yes, that's exactly how I felt at the time. Then she said, now, among all possible options, the only thing left to do is to sue the government. My initial reaction was, this is too crazy. We took this to the board to discuss; I remember Alfred, Michael, and Seibel on my side were there.

John Collison:
That would be Alfred Lin and Michael Seibel.

Tarek:
Yes. I remember those board meetings where it always started with, we have to be very clear with you, this idea is terrible. There are many reasons, your opponents are regulatory bodies; you are just over twenty people; the government really wants to go after you, there are countless ways, they can shut you down, revoke your license, they can do it all. And this is not just theoretical risk. Even if you win, you might have been bled dry in the process.

I also remember, before the official board discussions, we had an internal meeting. It was the night before we had lined up lawyers, prepared to file lawsuits, and I suddenly backed off. I said, maybe we should just go back to being a clearinghouse, or focus more on financial products, not put everything on this, not go all in. And in that call, I can't recall the exact words, but the essence was, are you kidding me?

Luana:
That does sound like something I would say.

Tarek:
At that moment, I realized, okay, I can't win this argument. But another part of me knew we just had to do this. Later, when we went to the board to discuss, their response was essentially, this is clearly an anti-pattern, a bad idea. But many great companies are built on some form of anti-pattern, there is always something abnormal happening, maybe this thing is your abnormality.

John Collison:
That's a great way to put it. Every company eventually emerges in some new, unconventional way, so maybe this is your way. So when you later won the election lawsuit, what was the legal basis? Were there any particularly interesting policy aspects?

Luana:
The core is actually quite simple, the government cannot arbitrarily ban a type of contract unless it deems that contract against public interest, and that ban must fall within specific categories, like war, terrorism, assassination, etc. The CFTC at that time was trying to shoehorn elections into these categories. They would say, elections might be illegal under certain state laws, they even brought up some state laws regarding bucket shops, trying to find any reason to block it.

But we were very clear about the law, elections have economic impacts, and as long as there is an economic impact, it should be tradable on a futures exchange or derivatives exchange. That lawsuit was essentially telling the CFTC, you can't just do as you please.

John Collison:
So, the so-called banned category must truly belong to one of those explicitly prohibited categories, and elections clearly do not belong to that.

Luana:
Exactly, that's right.

Tarek:
This point is very important. We often talk about how the law constrains businesses, but the law also constrains the government.

John Collison:
Right. Matt, were you just about to mention the point about suing the government over the past couple of years?

Matt Huang:
Yes. I think in the realm of crypto and prediction markets, suing the government may seem unusual, but I later found that it's actually much more common than in the traditional Silicon Valley mindset. Coinbase has sued its main regulatory agency; in the GovTech field, SpaceX, Anduril, Palantir have all sued the government for various reasons. So I'm curious, since you've dealt with the government so much, what advice would you give to those looking to do similar business? In what circumstances do you think initiating such a challenge is the right move?

Tarek:
I think it should only be done when there is no other choice. It's still very painful.

John Collison:
But do you really have no other choice? Can't you continue without an election market? Of course, elections are a very prominent, very attractive category, but I guess it's not the primary source of contracts for most of you today, right?

Luana:
I think it's just too important. Perhaps this sounds a bit obsessive, but it really is the Holy Grail market. It best illustrates the use of data in this market and best reflects the value of this market. Take the 2024 election, for example; the polls were way off, while the market significantly outperformed in information integration. I think it is the most shining example that demonstrates why prediction markets are beneficial, why the US needs to have them within a regulated framework. Other markets do not have such a strong demonstrative effect.

The Core Logic of Prediction Markets: Using Real Money to Produce Information

Matt Huang:
John just mentioned the logic of PayPal and Uber, where you act first and explain later. In fact, there were already other prediction markets operating offshore at the time, showing real demand. So I'm curious, did this matter help you in your lawsuit? For example, did it help to illustrate that election markets do not go against the public interest—after all, people were already engaging in them?

Tarek:
I'm not sure. But in terms of the courtroom, the focus actually fell more squarely on the legal text itself. We discussed the Commodity Exchange Act, which is one of the core laws of financial regulation; the other corresponding part is the Securities Exchange Act. The focus was on reading through each section, interpreting these laws, and then determining whether the regulatory agency had overstepped.

However, from our own perspective, the existence of offshore markets has indeed been helpful because it allows us to still reference some external data on the path of regulation first, product later. At that time, we couldn't learn directly from our own product because we insisted on getting a license first before doing anything. So, in a sense, some external data, external evidence, can indeed help us make decisions. It also allows more people to start to understand what prediction markets are and how they can be used. But in terms of policy, have offshore players been very helpful to us? I think not necessarily.

John Collison:
If Kalshi had appeared ten or fifteen years earlier, would it not have taken off at all? Is it because the current CFTC is more open, or is it necessary for certain technological conditions to mature, such as stablecoins?

Luana:
I do think that part of the factors come from crypto. Back then, early prediction markets like Augur had already appeared. I think the existence of these things did make the CFTC feel that we needed a legal, regulated alternative. In the past, they might have just said no. I think this did play a role, but perhaps only 5% to 10%, not more.

Tarek:
More broadly, I think people's intellectual interest in prediction markets has always been there, starting from the 1950s. Everyone has long known that this is a better source of signals than many other information mechanisms. However, ten or fifteen years ago, there was not such a strong real-world pain point; in recent years, this pain point has been real. I think the country is more divided, and the world is more divided. Social media has fragmented the information flow into different camps, clickbait headlines are prevalent, and a large amount of the content we read today—whether it's traditional news, social media, or something else—has increasingly skewed toward sensationalism. It is precisely because of this that more pressing issues have emerged, and these issues have brought about this wave of adoption of prediction markets. I don't think the situation we see today would have occurred fifteen years ago because the problems then were not as severe as they are now.

Luana:
Also, most of our users are not actually here to trade. About 80% of users are more information consumers. They come in to see, for example, who will win the Texas primary election yesterday, take a look and know that the polls say the two sides are evenly matched, but in reality, they are not. The function of serving as an information carrier is much more important today than before.

John Collison:
So, are you saying that in the age of algorithmic information flow, markets like Kalshi are very suitable; if placed ten or fifteen years ago, people might not have been so interested?

Tarek:
Yes. I think more accurately, people's distrust of traditional sources of information is significantly and continuously increasing. So you need a new source, and this mechanism does work. The incentive mechanism of prediction markets points to the truth, more trading volume, more liquidity, ultimately leading to more accurate predictions. This process takes some time; people need a few rounds of validation to start believing in it. But once it has established a record, people will no longer want to use a product that is clearly inferior.

John Collison:
When it comes to transaction volume, could you outline for us Kalshi's growth trajectory? It seems to be growing exceptionally fast.

Tarek:
The transaction volume in February this year was $10.4 billion.

John Collison:
So, that's a transaction volume of $10.4 billion in contracts.

Tarek:
Yes. Compared to six months ago, it has grown roughly 11 times, maybe a bit more.

John Collison:
It's growing so fast that you can't even be bothered to look back a year because that's already ancient history.

Luana:
Indeed, a year ago feels like ancient history. For example, a year ago, we only had one sports market.

Tarek:
Yes, that was in February. In any case, the growth is indeed very rapid.

Matt Huang:
Aside from AI, it's probably one of the fastest-growing companies.

Tarek:
I think so. It may even be on par with some of the top AI companies. I'm not sure what the latest data is for Cursor or Anthropic, but—

John Collison:
And even within AI, an 11x growth in six months is already quite extreme.

Tarek:
Indeed, very fast. I think the reason is that we are a true market and possess the inherent attributes of a market, such as network effects. As the market categories grow, liquidity deepens, user retention improves, and user engagement and transaction volume continue to increase over time. This, of course, will drive their own usage growth, but will also drive the growth of others because there is more liquidity in the system, the product is easier to use. Then, users will also be more willing to share with others. So, these forces combined are driving the current growth.

Matt Huang:
A significant portion of your early growth actually came from other brokerage platforms. Today, this structure has changed. How do you view the broker line? What is the current proportion?

John Collison:
What does broker refer to here? Like Robinhood?

Tarek:
Yes. This question is very interesting.

Luana:
I can first explain the broker part, and let him talk about the specific numbers. In simple terms, because we are essentially an exchange plus clearinghouse, our role is more like the New York Stock Exchange, or more accurately, more like the Chicago Mercantile Exchange. This means that brokers can connect to us. You can trade stocks on Robinhood, and similarly, you can trade Kalshi's contracts on Robinhood; in the future, it will be the same on Coinbase or other platforms.

From the very beginning, we have been very clear that we are first and foremost an exchange and clearinghouse, not anything else. Being able to connect with institutions like Goldman Sachs and Robinhood is crucial to our understanding of the entire ecosystem.

At the beginning of last year, our first broker partners to go live were actually Robinhood and Webull. During that initial rapid growth phase, the broker channel accounted for a very high proportion, which was actually good because brokers would bring in a lot of demand, and once the demand is there, market makers are willing to participate because they want to offset retail flow. In this way, we also bought ourselves time to gradually refine our user-facing products to where they are today.

Our current understanding is that the core will always be the exchange + clearinghouse. Users can access directly through our app, website, API, or through any broker. We are now increasing our focus on institutions and international brokers. In the future, even if you are in Brazil, you can trade Kalshi directly, all of this is coming soon. As for the numbers, you can share.

Tarek:
She is not willing to disclose specific numbers, but our so-called direct, which is the part that directly faces consumers like kalshi.com and the Kalshi App, has seen a growth rate that has clearly surpassed that of the intermediary channel, namely the broker channel. I think this is mainly because the brand has become well-known. Now, many people, as soon as they have a differing opinion on something, their first reaction is, "Let me open Kalshi to check the odds," or "Let me place a bet on Kalshi." The brand has become synonymous with this behavior itself. There has already been a lot of organic growth, and I believe this trend will continue in the coming months.

An Counterintuitive Market: Retail More Important Than Institutions

John Collison:
What you just talked about is the retail side, how individual users are growing, some coming in through brokers like Robinhood, and some coming directly to the Kalshi website. But as an exchange, you also have another crucial thing to address, which is market-making. The New York Stock Exchange doesn't have to worry too much about market-making because the economic incentives themselves are strong enough, and once the scale is sufficient, this is no longer a big issue. But I am curious about how you did it in the early days. Did you do market-making yourselves? Did you collaborate with external market makers? How did you incentivize them to participate? How did you build your market-making system from scratch?

Luana:
The market on Kalshi can actually be divided into two categories, each behaving very differently, so the market-making incentives are also completely different.

One type is the long-tail market, such as a market on whether One Direction will reunite. These markets are actually quite hard to price, and demand is usually low, so we really need to attract market makers through various incentives, including recruitment incentives, and so on. One of our long-term focuses is: How can we establish stable, sustainable liquidity for these long-tail markets? We may have around ten thousand markets now, so how can we still ensure liquidity if we have fifty thousand or a hundred thousand markets in the future?

The other type is more classic markets, such as crypto, sports, and so on. For this type, market-making is actually much easier because the demand is clear, and the pricing logic is more mature. The market-making incentives on this side are not direct monetary rewards but rather a partial fee rebate. However, at the same time, we set very strict performance requirements, such as maintaining a certain spread or a certain order book depth within a certain time frame. Because in these markets, we are more like incentivizing the stability of the order book rather than just encouraging market makers to participate.

John Collison:
When you talk about incentivizing the stability of the order book, what exactly do you mean?

Luana:
For example, in a live match or in a crypto market where you settle trades hourly—

John Collison:
If there is no new information coming in, you don't want the price to wildly fluctuate for no reason, right?

Luana:
Exactly. Even if there is actually new information, such as someone about to score, you don't want the entire order book to lose all liquidity instantly. Of course, you allow the spread to widen appropriately, but you still want users to be able to trade. Especially as we move towards a broker model, brokers will come to us with their expectations from traditional markets. They will say, we want the spread and depth to be maintained at a certain level no matter what time it is. So we have to negotiate with market makers on how to design incentives in this scenario. Because if you let the market decide purely on its own, it may naturally widen a lot during high volatility; but to serve all users, including broker channels, we must do more design at the incentive level.

Matt Huang:
So in those moments when the spread widens significantly under normal circumstances, do market makers lose money? Are they using profits from other more stable periods to subsidize these moments?

Luana:
Now, because overall demand is already very strong, they can actually make money even with a slightly narrower spread. But this is also the meaning of our incentive plan; you have to look at all the benefits that the entire plan brings. Perhaps you might lose a little at certain times, but as long as the overall revenue is high enough, then it's worth it.

Matt Huang:
So your goal is to always maintain a tight spread in the main markets.

John Collison:
That is, the main markets need to achieve a tight spread around the clock. This actually requires careful design.

Tarek:
Exactly, this is very challenging. But what's even more interesting is that the truly unique aspect of the prediction market is that a large part of the liquidity doesn't actually come from the market makers you usually think of but from regular people.

This brings us back to the initial logic. We've solved the regulatory issues, but next comes the liquidity issue. Traditional exchanges like the New York Stock Exchange, CME, will say, "We want to launch a grain futures contract," so they spend two years designing the product, bring in the familiar fifty market makers, prepare together months in advance, and spend a few years promoting the product. That's the traditional way. However, the prediction market is entirely different because you have to constantly generate liquidity on a weekly, daily, or even hourly basis for new events. How do you do this? Its rhythm is highly dynamic, with new things always emerging.

John Collison:
Many people find this counterintuitive, that you actually need to incentivize market makers to provide liquidity. Because in the stock market, you don't need to incentivize high-frequency trading firms; they will eagerly invest and build low-latency links between New York and Chicago to do this. So is this because the prediction market is still in its early stages, or is there some more fundamental difference here?

Tarek:
This brings us back to the key point I mentioned earlier. What you're facing now is a model that requires instant liquidity provision, much faster and more dynamic than the traditional market. Traditional Wall Street market makers do not operate in this way. You can't expect them to set up a desk on short notice within an hour and price topics like politics or culture.

What's really interesting is that the prediction market has a very counterintuitive feature. In many markets, the people best at pricing it may not necessarily be the so-called experts or authorities, but rather ordinary individuals.

Matt Huang:
Internet anonymous individuals.

Tarek:
Yes, exactly, those super forecasters. This ability is highly decentralized. It's hard to say that a specific demographic group is the best at pricing. And the reason we are where we are today is that it took a long time for us to eventually cultivate an entire community, a group of super forecasters on Kalshi who can efficiently price these things. Initially, it was hard to turn their interest from a hobby into a part-time job and then from a part-time job into a full-time job. But as the market pie grew larger, this eventually happened.

Luana:
Here's some data we can share: On the platform, the traditional large institutional market makers, in the conventional sense, represent less than 5% of all executed market orders. Even the largest among them.

Tarek:
So, they only account for a very small portion of the matched liquidity.

John Collison:
Really?

Luana:
In other words, in all the filled orders, the proportion of those well-known large institutional market makers is less than 5%. In other words, over 95% is peer-to-peer, or those small funds or teams with only two people involved.

Tarek:
This is very rare on exchanges.

Matt Huang:
So, how many of these small but full-time market-making teams are there?

Luana:
There are approximately over 2,000 people market-making on Kalshi.

John Collison:
Matt was actually asking, who are these so-called market makers on Kalshi? Are they institutions like Jane Street, Akuna, or are they those people coding in their garage at 3 AM with a can of Red Bull?

Tarek:
Ironically, it's the garage people who are the most critical.

Matt Huang:
And you just said they account for 95%.

Tarek:
Yes. They are crucial to the entire system because they price quickly, are always watching the order book, and constantly monitoring the situation. They are truly the most primitive observers of the situation.

John Collison:
So, Kalshi is built on top of a group of people who are constantly watching the situation.

Tarek:
Exactly. Let me give you an example. In recent years, the best inflation predictor on Kalshi wasn't those institutions or well-known hedge funds. It was a person in Kansas. He had never traded in financial markets before, just liked to read the news, had a sense about inflation, and thought he could feel where it was going. And you'll find many such people on the platform. While there may be a few thousand formally and fully engaged, in a broader sense, there are tens of thousands of people like this. They know a lot about various topics and actively price in on these events and are rewarded for this ability.

Luana:
Let's talk about my favorite user.

Tarek:
By the way, I also recently have a new favorite user.

John Collison:

Alright, each of you talk about your favorite user.

Tarek:
I was thinking about this earlier today. It's actually the person from that Wall Street Journal article about taxes—

Luana:
Oh, yes, he's definitely a strong contender as well. But my favorite user is a huge Ariana Grande fan. He found Kalshi during the election season, but he actually doesn't like elections at all, no interest whatsoever. Then he found our Billboard market, which is like a ranking market.

John Collison:
He sees this as a very important market.

Luana:
It's very important to him. He has made over $150,000, paid off his student loans, got a master's degree, and bought a car with that money. He had never traded before or done anything like it, but he has a very strong, almost obsessive interest in music rankings. For the first time, he was able to monetize that passion. And he's been really friendly to us on Twitter too.

Tarek:
I have many favorites, but this one has been up there recently. Last week, The Wall Street Journal wrote an article about a very active tax accountant on Kalshi, named Alan. When DOGE first came out, everyone was talking about how much it would reduce expenses. He went and dug through a ton of tax law and regulations, researched deeply, and came to a conclusion that it wouldn't meet the external expectations no matter what. He almost definitively made that call. Then he went back to tell his wife, I have very high conviction in this trade. In a way, it's a bit like Michael Burry from The Big Short, except this time he was shorting DOGE. And he ended up making a big bet and winning.

This is where the prediction markets are powerful—if you have that kind of knowledge, and that knowledge is often esoteric—like, I guess none of us here has read that stack of tax law—then you can really go do research, understand the world better, and get rewarded for it. It's amazing.

John Collison:
An early important application of AI was poker bots. Have you guys seen any really good AI market makers now? Because no one has quite done the tax law thing, and outside of Claude, it probably doesn't hold up anymore.

Luana:
That's true. Maybe we should indeed ask it.

Tarek:
We have indeed seen more and more people using agents to trade, especially on the API side, that's been quite apparent.

John Collison:
So have you had users already running successful highly agentic market-making operations? Where most of the process is driven by the intelligent agent?

Tarek:
Users usually don't tell us the detailed workings of their strategy.

John Collison:
But you do chat with them, right?

Tarek:
We do, John, the answer is yes. My understanding is, early Renaissance Technologies, weren't they already using some agent-based model for trading? Of course, it was a very early version. I think today it's just an ongoing evolution, and it will only get stronger. In the systems of many traders on our platform, there are already some AI-based summarization and decision-making modules.

John Collison:
What I'm really curious about is those fully autonomous systems, no humans in the loop, absorbing information and making quotes on their own. Feels like that's coming very soon, if it's not already here. Like your Claude doing market-making on its own.

Luana:
I'm not sure if there are fully unmanned systems yet. For example, in international election-type scenarios, I know many systems already automatically translate documents, conduct polls, and perform various processing tasks. But I'm not sure if it's fully automated already.

Tarek:
We actually don't know if the models have reached that stage yet. We recently launched Kalshi Research, one of the directions is to collaborate with some research labs to establish a brand new benchmark to see which models are better at predicting the future. This benchmark is very unique because it tests not memory patterns, but the model's understanding of the world itself. I'm personally looking forward to the results.

John Collison:
So how are you planning to evaluate?

Tarek:
It's not fully decided yet. But the general idea is to have the models run on the same set of markets continuously for a month or two, and then see who performs better, like prediction accuracy, long-term PnL, and so on.

The Fundamental Difference Between Online Entertainment: Trading vs. Gambling

John Collison:
Another question about market making. In sports online entertainment, there is a well-known phenomenon where online entertainment companies target so-called sharps, meaning people who are too clever or too good at betting. Many people do not realize that, for online entertainment companies, the most ideal bettor is actually someone who is not very professional, just a fan supporting their home team; and the worst is someone who is especially good at identifying mispriced niche markets. Because online entertainment companies may offer odds on one thousand markets, just one or two mistakes, and professional players will specifically target those errors. So they will identify you through behavioral signals, if you just registered and only bet on your home team, that's good; if you appear too professional, they will ban you.

This is actually very interesting, you think you are just betting according to the odds they provide, but if you are too good at it, they don't want you. It's a bit like a Las Vegas casino asking you to leave. Would Kalshi also have this issue of dealing with sharps? I originally thought not because you should welcome them. But would market makers worry about having too clever opponents on the other side?

Tarek:
The issue is that sharps themselves are part of the market. Let me clarify one thing first—

Luana:
We don't limit winners. We don't do that kind of thing. We want the smartest people to come.

Tarek:
We need these sharps. Without them, how would the market be more accurate? This is our biggest difference from online entertainment companies.

John Collison:
That's easier said than done, it's not entirely the same. Because what you want is liquidity, to maintain a narrow spread throughout the entire match, throughout the entire election process. Whereas sharps may just suddenly rush in when the odds are wrong, make a big profit, and then disappear. So, providing liquidity and making correct judgments are two different things after all.

Tarek:
But many sharps would actually earn more if they were to provide liquidity. They could easily become part of the liquidity. This is especially crucial. Many people would say I'm not gambling, I'm trading, although this statement sometimes sounds like self-consolation, but I think it does point out a fundamental difference. The business model of gambling is the house eating the losses of the customers, your income comes from the users' losses. So the behaviors you just described are completely reasonable for online entertainment companies, if someone makes money, I have to stop him because that is a direct loss on my profit and loss statement; if someone loses money, I have to figure out how to bring him back.

This is completely different from the traditional financial markets. In the financial market, the core of institutional design is fairness and transparency. You need to create a set of rules that are fair to all participants. Maybe Matt is better than Luana, maybe Luana is better than Matt, they can compete inside and determine the outcome for themselves.

John Collison:
So your incentive system is completely different. You don't make money by one party losing money as in a casino; you earn through transaction fees.

Luana:
Exactly. For us, the best outcome is for users to perceive this market as fair, with good prices and stable liquidity, so they are willing to trade here. Of course, to achieve this result, we also need to design different incentives for different roles. That's why we have various liquidity programs. If you are providing liquidity and taking on a higher impermanent loss risk, then your fee should be lower; if you are actively taking liquidity or sniping, then your fee should be a bit higher because you need to pay for that behavior.

John Collison:
So you use fees to incentivize pro-social behavior.

Luana:
Yes, I believe that's how the financial market naturally operates.

Tarek:
The traditional financial market essentially follows the same logic.

Luana:
It's just not stated as explicitly.

Tarek:
Essentially, it's about tilting the playing field slightly in favor of those who truly create value for the market, giving less advantage to those who merely extract value.

John Collison:
So what behavior is considered pro-social, and what is considered anti-social?

Tarek:
Insider trading is certainly considered anti-social behavior.

Luana:
That's the most typical example.

Tarek:
And it's illegal. As for sniping, it is actually part of the market. Suddenly someone gets new information and then trades on it; this happens every day in the traditional financial market. It's just that if you want to maintain liquidity, make these market makers willing to allocate resources, you must give them a certain incentive.

And I think that's also one of the reasons why prediction markets are becoming more widely accepted. Everyone likes this kind of mechanism—your advantage is directly proportional to your depth of research, level of information, and the time and energy you invest. The traditional financial market is actually the same; it's just that for many, those markets are not as interesting. Compared to pondering IBM's quarterly reports every quarter, I think researching DOGE, elections, how people perceive an election, why they vote as they do, is much more interesting.

John Collison:
Established by Kalshi, it is a new type of market where real-world outcomes themselves can be traded. For example, whether the US will confirm the existence of aliens by 2027. Thousands of participants can open positions, transfer funds, settle in real-time, with the underlying being a highly intricate multiparty fund flow system. The fund orchestration behind Kalshi is powered by Stripe Connect, facilitating participant onboarding, payment processing, fund routing, and payout management. When the fund flow itself becomes programmable, new products and even new market structures can emerge. So, if you are building a brand-new, fund-flow-intensive product, Stripe Connect is tailored for you.

Tarek:
Sorry, everyone.

Everything Can Be Priced? The Future Frontier of Prediction Markets

Matt Huang:
Let's talk about different market verticals. Now everyone can understand markets like elections, sports, economic indicators. But I think prediction markets are actually like a search function applied to all the interesting markets humans want to trade. In the past, traditional exchanges like CME decided what would be listed, often commodities like wheat, oil, corn. But now you can launch a thousand markets in a day. So as this happens, what do you think we will ultimately discover?

Luana:

There is one direction we are particularly excited about, and we have already started moving towards it, such as the more collectible side like watches, handbags. In fact, it is entirely feasible to create derivatives around these things.

Another thing worth mentioning is GPU computing power. This is also a very interesting direction. We are increasingly thinking that many assets are actually more suitable for traditional futures structures rather than binary yes/no markets. In other words, it is not a binary question of whether the price will reach a certain point, but more like real futures where you can have margin, more institutionalized liquidity. This direction is crucial because once we move away from purely binary markets to more traditional derivative structures, we are essentially expanding the trading subject from grains to computing power.

Matt Huang:
Historically, the most successful futures markets have always been these large commodity categories because they are naturally large expenditure categories. And we are now in an era where humans are spending most on a new commodity.

John Collison:
A new commodity category.

Matt Huang:
Yes, a new commodity. And the traditional markets have not really been attacking this category of computing power.

Luana:
Our approach to this is that we want to be the world's largest derivatives exchange. To achieve this, there are four key things in our product roadmap.

First is the breadth of market themes. This means we want to cover topics like computing power, sports, elections, securities, and more.

Second is market structure. Currently, we only have binary yes/no products, but in the future, we want to have futures, swaps, options, and more.

Third is the margin system. Currently, this system is actually very poor, as you have to put in all your funds upfront.

John Collison:
Meaning the capital requirement is too heavy.

Luana:
Yes. This makes it very difficult to operate in many markets, such as markets like whether there will be a hurricane this year. If you want to make a market or sell these contracts, the capital efficiency is almost unacceptable. Fourth, is liquidity.

Internally, our view is that if we can succeed in these four areas—having the broadest market themes, the richest market structure, an excellent and efficient margin system, and good liquidity—then we will succeed no matter what we do. So, everything in the company ultimately falls into one of these four buckets.

And the so-called thematic expansion is essentially about how to match the right theme with the right market structure and the right margin model, and then secure the necessary liquidity. You are absolutely right, if we can rebuild these margin systems from scratch, we will be able to more quickly introduce new margin models in the future and list more new markets.

Matt Huang:
Because you have a direct-to-consumer mobile product, I'm curious, do you naturally lean more towards markets that retail customers find interesting, rather than more professional or institutional markets? For example, computing power, which in my opinion is more of a professional-to-professional market. How do you think about the liquidity and interest cultivation in such markets?

Luana:
To be honest, internally, we almost see the company as two parts as well. One part is about improving existing markets, such as sports, crypto, and other things that are already up and running; the other part is about constantly driving the next new thing. Because I believe what has truly brought Kalshi to where it is today is not regulation, not anything else, but our continuous push for the next new thing. First, it was elections, then sports. For us, consistently defining the next big thing and executing it well is essential. If we stop doing that, the company won't succeed.

The current organization of the company is also roughly like this, with the market operations and engineering teams more inclined to maintaining and optimizing existing things, while new teams, such as the institutional team, margin team, international team, are more inclined to driving things forward. Then there is a layer of platform focused on core exchange, compliance, and other infrastructure. We've always been balancing between different themes. But in the end, we are only 120 people, there is a lot to do, and it's indeed very challenging.

Matt Huang:
So now on the institutional side, they have started reaching out to you on certain topics?

Luana:
Yes, of course. For example, we just launched a feature called Block Trades a week ago. You know what a block trade is, right? This is a very institutionalized way of trading, where I can call you directly, agree on a trade, and then submit it to the exchange for confirmation, instead of everything having to go through the public order book. We are now working on many features like this, hoping to attract more institutions.

Matt Huang:
Are they trading the same things as Kalshi's retail side? Or will you offer new products to them?

Luana:
Yes, and no. For example, if they are interested in a specific topic—like many people were interested in the tariff issue before, whether tariffs would be imposed; and recently, some are focusing on the future of oil reserves—as long as we hear that they want to trade this market, we usually just list it directly, and it is open to everyone. I do think that in the future, what institutions will ultimately trade will be very different from what ordinary retail users are interested in. But at least in terms of listing on the market, the cost for us now is very low.

John Collison:
When Uber first appeared, it didn't actually harm the taxi industry much because initially, it mainly targeted idle supply and met incremental demand. But later, as it grew larger, it truly began to challenge the taxi industry. So, what about Kalshi and other prediction markets? Are there any existing industries that will increasingly feel your impact? Because once this market is larger and more liquid, it may replace some of their functions. For example, traditional futures exchanges, maybe in the future to hedge soybean prices, going to Kalshi might be more convenient. And then sports online entertainment companies, political polling institutions—the latter may find that similar information can be provided by you more cheaply. So, who do you think will be the first to feel the impact of being overtaken by the prediction market?

Luana:
You know that meme, the one where the bouncer at the door asks who's next? Actually, we don't really want to call out who will be taken down directly. But the directions you mentioned are indeed all in there. Traditional online entertainment industry is certainly one, we just talked about many structural problems in this industry, and we are completely different from it. The traditional futures market is also, as we are starting to enter more and more into their territory. Political polling institutions will also be affected, I think since the last election, many campaign teams have been using our data. And parametric insurance. When we have a more mature margin system, we can enter scenarios of true risk pricing like hurricanes and disaster insurance.

Matt Huang:
Will there be a tragedy of the commons in polling? After all, part of the reason prediction markets are accurate is that they rely on interpreting polls. In other words, polls are the sensors, and prediction markets are the mathematical interpretation of these sensor outputs. So, if everyone stops polling, could it weaken the prediction markets instead?

Luana:
My view is the opposite; polling will only get better. In the future, people will realize that if their polls are more accurate, they can make money. So, they will be willing to commission better polls and design better methods. You can now actually have many different polling models competing in the same market.

John Collison:
Just like the metapoll aggregator that 538 does.

Luana:
Yes, even further. You can ultimately arrive at a single number that aggregates all these different things. In the last election, someone actually did this; they specifically commissioned a new polling method, similar to the nearest neighbors approach. Although I don't quite remember the specific details, the result was that they made more money in the market. This is the power that real money on the line brings—it aligns incentives with the truth. Polls are no longer about telling me what I want to hear but about providing actual numbers.

So, I think they complement each other, just like the news. Many people say prediction markets will destroy the news industry, but I see it more as a complement. For example, when discussing an election, news commentators will still give their opinions; the market does not provide opinions but a number. You still need commentators, except now they can say during commentary, this is the probability the market is giving, and this is my personal judgment. I don't think opinions will disappear.

John Collison:
You mentioned insider trading earlier, and there is actually a very complex policy issue here — where should the line be drawn regarding insider trading in prediction markets? Even in the stock market, this is already quite complex. For example, we all know that the SEC often enforces against illegal insider trading, but there are legal scenarios too, such as a hedge fund using its unique Walmart parking lot satellite data to predict earnings, which is obviously information others don't have but is legal. Similarly, in prediction markets, the boundary will also be very complex. For example, we probably all agree that government officials should not trade before a military action. But what about before the Super Bowl when someone knows in advance how long Bad Bunny's halftime show will be? Some people will indeed know this information in advance. So, how do you think the boundary of insider trading should be defined?

Luana:
Actually, your statement is already quite accurate. This is a very complex issue. It's even more complicated in the stock market and on a larger scale. The principle we are now following is to comply with federal law.

John Collison:
Does this include both CFTC rules and other financial regulatory rules?

Luana:
Yes, CFTC and SEC both have relevant frameworks. In simple terms, if you have signed a confidentiality agreement, promising not to disclose certain data, then you cannot trade using that information. For example, if I work at the Bureau of Labor Statistics, my confidentiality obligation prevents me from disclosing inflation numbers before they are released, so I certainly cannot trade on that information. However, if you happen to know that the Super Bowl halftime show rehearsal is on Thursday and you are just standing outside and happen to overhear Lady Gaga singing, that is okay. The same logic applies to traditional hedge funds using alternative data such as Starbucks store foot traffic. The role of the market is to encourage information to enter the market. We welcome information into the market, but the condition is that it cannot be unfair information. If you obtain it unfairly, then you should not trade on it.

John Collison:
So, if you have a confidentiality obligation regarding such information, you cannot use it to trade.

Luana:
Right. In fact, we go even further. For example, if you are a government official, you cannot trade on whether a certain bill will pass, even though I am not sure if they have signed some formal confidentiality agreement.

John Collison:
Very interesting, because it is widely known that members of Congress can actually trade stocks now.

Luana:
Yes. So, in this matter, we are actually more stringent than the current system. We have also been in constant communication with regulatory agencies because this is a new issue for both them and us. We operate in a regulated market, so we have a robust internal monitoring department that almost never sleeps, monitoring every anomaly signal, trying to resolve all issues. Just about two weeks ago, we announced two insider trading cases. And precisely because we are within a regulated framework, we can fine these individuals very heavily, fines exceeding five times their ill-gotten gains, and ban them permanently, and so on.

John Collison:
One interesting point I find is that you are already doing this whole set of actions yourselves. In the public stock market, the SEC has always been very proactive in enforcing insider trading. So, what is the CFTC's stance on this issue?

Luana:
That's a good question. You can understand the whole mechanism as a three-tier system. The first tier is our own monitoring and enforcement. The second tier is the CFTC's own monitoring and enforcement. The third tier, if it is more severe, would involve the Department of Justice. The reason why SEC cases are more prominent is that in many cases, the exchanges themselves will investigate, take initial action, levy fines, and freeze accounts before the subsequent processes. We have a similar process here. Every transaction on Kalshi is synchronized with the CFTC, and they can see everything, every case. We also submit these cases for CFTC review. As for whether they will take further action, we do not know, but at least the ball is now in their court.

John Collison:
So you would hand off the case to them, but with the first layer of protection in place.

Luana:
Yes, exactly.

Matt Huang:
I'm also thinking of another scenario. In some markets, the future outcome is unknown to anyone right now, so by definition, insider trading is almost impossible to exist. But there are also markets where a single person can influence the outcome, such as whether a certain political figure will mention a specific word in a speech, or if a particular player will attempt a certain number of shots. How do you view this spectrum?

John Collison:
For example, in a so-called mention market, is it inherently a bad idea because it's too easily manipulable?

Matt Huang:
Or could it be that such markets are inherently unscalable?

John Collison:
Like when Brian Armstrong might mention a word during the Coinbase earnings call, these markets are also similar.

Luana:
I actually think mention markets are great. Just think about the Federal Reserve. Many hedge funds are there, watching which way the chair leans during a speech, whether a specific word is used, or if there's any change in tone. Because we all know that the presence or absence of certain words conveys very specific meanings and can strongly move the market. The same goes for Trump. If he says we are going to war, the market will certainly fluctuate significantly; if he mentions tariffs, it will also see a big swing. In fact, many public figures' words have always been influencing and driving the market. So I think mention markets are very important.

Of course, those involved in writing the speeches or the individuals giving the speeches themselves cannot trade; neither can their teams. That's our constraint. For example, if you're Gavin Newsom, and the market is betting on whether you will say a particular phrase, then you and your staff cannot participate in trading. My principle is, as long as we can safeguard market fairness by limiting certain participants and the market itself has positive value and economic utility, then it should exist. We shouldn't say this market shouldn't exist just because five people might be able to manipulate it. Otherwise, the stock market shouldn't exist either. The key is not to shut down the market but to establish a set of robust rules and prohibitions that allow the market to exist and remain fair.

John Collison:
Another big controversy you are deeply involved in is sports contracts. Critics would say that sports betting expansion has brought about many well-known negative consequences, which are measurable. Especially in the U.S., the legalization of sports betting has made significant progress over the past decade, and there is now some data. On the other hand, I personally don't have a strong moral objection to alcohol, even though the distribution of alcohol's consequences is quite similar; the vast majority of people just enjoy it normally, but a few have very bad outcomes. However, the moral discussions in society about alcohol and online entertainment are clearly different, even though their distribution shapes are very similar.

In addition to my own experience—online sports betting has actually been around in Europe for a long time, since the early days of the internet. Initially, it was through various loopholes, Malta licenses, and the like, before gradually becoming more regulated. People have always had access to it, and life has not collapsed as a result. Obviously, this is still a major point of contention around Kalshi. What is your view on expanding sports contract availability?

Luana:
There are actually many layers to this. Why did we decide to launch in the sports market? First, undoubtedly, sports represent a huge demand; second, sports are an area where people are willing to spend a lot of time researching and are genuinely knowledgeable. However, traditionally, people have not had a really good way to turn their knowledge in this area into income. Traditional online entertainment restricts you as soon as you win too much, the whole system actually does not work well. So, whether you like it or not, there will always be people betting on sports; the question is just how to make them engage with this in the best way.

I think the market has a key difference compared to the house, the market is objectively a better mechanism. I have hardly ever heard anyone seriously argue that a state-regulated casino system is a good thing. Recently, you can hear some propaganda from the online entertainment industry side, saying how reasonable this system is, but if you really ask them for data—

John Collison:
Just from the scale, you can tell. Sports online entertainment companies probably have around a 10% house edge, while prediction markets may only have a 1% or a few percentage points.

Luana:
But the predatory nature is not only reflected in the house edge. The worst thing about traditional sports online entertainment is that when you start losing money, the first thing they do is give you a bonus, give you $1000 cashback, deposit bonuses, to lure you back because what they really want is losers, not winners. Their core is to make you come back and continue to lose. We don't do any of that.

John Collison:
At sports online entertainment companies, the people who lose the most money are actually the most profitable customers.

Luana:
Yes, this creates an extremely distorted incentive. We don't have any of that. I think the fundamental issue is that there will always be people going to Robinhood, Coinbase, to speculate on stocks or crypto; and there will always be people wanting to speculate on sports. They should have access to the best infrastructure. And today's sports online entertainment companies are not it. We firmly believe that our market is far superior to them in terms of security.

As for prohibitionism, it's actually like what you just said about alcohol. Prohibiting alcohol doesn't stop people from drinking; it just drives them to underground bars. Prohibiting sports speculation is the same; people will only run to offshore platforms where there is less protection, almost no self-exclusion mechanisms, no limits, none of the protective measures we have in place now, regulators don't even know what these people are doing. That's worse for users. So I think, simply relying on prohibition never works.

John Collison:
The policy discussions around this matter are also quite interesting. It reminds me of Canada, for example in British Columbia, where most liquor stores are government-operated. On one hand, the government says this thing must be strictly controlled, but on the other hand, it sells it itself and treats it as a source of revenue. The same is even more true for lotteries.

Luana:
At the end of the day, everyone is making money from this thing. The state government wants to make money, and the casinos want to make money. That's just how it is.

Matt Huang:
Speaking of niche markets. Earlier, when you weren't here, we were discussing what other new directions are worth getting excited about. What are your top picks?

Tarek:
I think a very interesting direction is to break down a stock into more atomized components.

Matt Huang:
For example, directly trading Nvidia's GPU shipment volume, rather than—

Luana:
Instead of delivery volume or other metrics.

Tarek:
Yes, rather than just looking at its overall financial report. This idea, when expanded further, will enter a more macroscopic level, examining which key factors are driving the entire economy. For example, AI, can we break out a series of questions around AI to price it? Or for instance, healthcare events, like something similar to COVID-19.

But the truly interesting part is that there is a paper by Kevin Hassett that once suggested, as society becomes more complex, the information carried by asset prices naturally decays because there are increasingly more factors influencing an outcome, and the vector dimensionality becomes higher. If you don't have a good understanding of dimensions X1 to XN, you cannot estimate the final Y very well.

The core conclusion of this paper is that you need infinite markets, and predictive markets are essentially a way to get to infinite markets. In other words, you need a price for each X, then feed these prices back to help with traditional asset pricing. Actually, a very good example happened last week. You know Citini released a research report, right?

Luana:
Yes, and surprisingly many people really bought into it.

Tarek:
Yes, I would say it received attention and appreciation beyond expectations.

John Collison:
That's the report about AI in 2028.

Luana:
Yes, everyone has been talking about that lately. In fact, 38% of people took it seriously.

Tarek:
I think to some extent, there is a societal tendency right now to really want to believe that AI will ultimately destroy us. This sentiment exists, and it will affect the market; stocks will fall because of it. So, we launched a prediction market about that report. Earlier on, Citadel also released a rebuttal report, and we launched a market around that as well. The market's probability was 10%.

John Collison:
So, you're saying that the probability of that economic scenario predicted in the report actually occurring is only 10%.

Luana:
More precisely, it proposed five conditions, and if three of them are met, it can be considered somewhat realized.

Tarek:
Right. If three out of the five conditions are met, it can be said that its outcome is somewhat true. But the market only gave a 10% probability to this result; if all five conditions are met, the probability is even lower.

So, this is significant because if you can feed this market price feedback into a pricing model, perhaps the market wouldn't react so strongly. For example, maybe people wouldn't just sell off DoorDash solely because of that report. You don't have to trust me necessarily, but you might trust the market; the market will tell you that the analysis surrounding DoorDash is actually quite mediocre.

Matt Huang:
The world you're imagining is actually a world where everything has a price at any time. I'm curious, is this really a world we want to live in? For example, Stripe currently benefits from remaining private largely because it doesn't have real-time pricing, which is smoother for employees, compensation, and emotional stability.

Luana:
By the way—

John Collison:
Because public pricing itself brings noise.

Matt Huang:
Right. In the long term, prices indeed get closer to the truth, but in the short term, it also amplifies panic. So I wonder, do you really think this is the world we want to live in?

Luana:
We certainly have biases because we love the market; we think the market is a good thing, so we definitely have a stance. But our view is: more data is always better than less data. Even if you think some data is not great, like second-to-second stock prices being too noisy, then you can choose to ignore it.

John Collison:
The CEO of a public company would probably say that choosing to ignore is not that easy.

Luana:
That's true. But overall, I still feel that having this data and then using it as one of the inputs is always better than not having it. Just because we say many things should have a price doesn't mean everything should have a price. There are many markets we will never engage in, such as wildfires, terrorism, assassination – these are bad markets. Of course, there is a moral boundary in prediction markets that we won't cross.

But overall, in a world surrounded by social media, you increasingly don't know what's real. Every day in my feed, it's "Is this true? Is that true? Did this really happen?" In this context, having a source of signals that is independent of personal stances and can be used to make judgments about the world is valuable. I think this value is very real.

Tarek:
I would frame it in a simpler way. You're essentially increasing the market efficiency of all these issues. Even though some issues may involve private companies, the logic is the same. Think about why companies go public? Why is real-time market pricing so important? Indeed, public markets have disadvantages, sometimes being too volatile, sometimes overreacting in any direction. But over a long enough time frame, the market is still a good weighing mechanism and a good capital allocation mechanism. So I don't think pricing more issues would fundamentally be bad. On the contrary, it would enhance efficiency and make the whole resource allocation system work better. Of course, there will be net losers, some people or things that shouldn't have received capital allocation in the first place may not get it in the future.

Luana:
And this is also a good feedback mechanism. For example, if you're a CEO of a public company and you announce something but the stock price keeps dropping, you would think, maybe I'm wrong. Politicians are the same way. In a debate, if they say certain things and their market probability goes down, they would realize that maybe that answer wasn't good enough.

I think in many areas, this faster feedback loop makes decision-making better. For example, in a government scenario, prediction markets have a very important application called conditional markets, where the government can ask, if we pass this bill, will employment go up or down? You can price these questions to assist decision-making. This means a tighter feedback loop and better incentives.

John Collison:
Do you feel like we're already starting to see the impact of prediction markets on politics? Social media has clearly changed politics, the political game has changed, popular candidates have changed, and the way political discussions happen has changed. What about prediction markets? Has their impact on politics started to show?

Tarek:
Definitely. The candidate themselves is using the prediction market price.

John Collison:
But that may just be a convenient reference. Social media is not only used by the candidate, it changes the candidate themselves, the campaign style itself.

Tarek:
Yes, it will reflect these changes.

Luana:
I think the most important point of the prediction market is that it is less biased than the party machine. If there is an underdog candidate that the public actually likes, but the party elites prefer another more establishment figure, the market often more accurately shows the odds of this underdog candidate.

John Collison:
So you mean, the power of the party machine has been slightly weakened?

Luana:
I think so. It clarifies what people actually want, which may not necessarily align with what the party wants to push. Maybe we haven't seen very obvious cases nationwide yet, but in a primary like in Texas, the polls said one person was a safe bet, but the person with higher odds in the prediction market ended up winning. I think many times, the market presents the situation more fairly than the party narrative.

Tarek:
Another crucial point is that it somewhat depolarizes, as Luana just mentioned. In a way, it is even an antidote to social media. Social media forcefully reduces a senatorial race into one dimension; your feed quickly decides if you are Republican or Democrat, and then continuously feeds you corresponding content. But the prediction market is not like that because the people participating in this market are not asking who is the good guy or the bad guy.

John Collison:
I think what you mean is, social media's feed quickly labels you, are you a Democrat or a Republican? And then it shows you things within that framework. Many people complain that they are pushed into a rabbit hole because the system quickly classifies you. And you are saying, the prediction market does not exhibit this phenomenon.

Tarek:
Quite the opposite, in fact. It will instead make you question, are we too sure about this person? This mechanism forces discussions to depolarize because the issue is no longer a one-dimensional Republican/Democrat, but more dimensions come into play. So you will start to hear judgments like, this person is actually quite cool, maybe although he is a Democrat, he would do well in Texas.

Luana:
The New York City mayor's race was the same. Almost everyone thought Cuomo was a sure win, 100% certain. But we saw Mamdani's odds consistently rising. I think this reflects that progressive messaging does resonate among New Yorkers. The market captured this change, and this depolarization comes from people willing to take a step back, no longer just judging by camp but seriously considering what is more likely to happen. Because the incentives are aligned.

John Collison:
It's a bit like the Iowa / New Hampshire effect in the U.S. presidential primary. Before the election, everyone thinks a certain big figure will win, like Hillary Clinton in 2008. But Iowa and New Hampshire not only gauge the mood of these two states, they also create a narrative. What you're saying now is, to some extent, the prediction market can also create this Iowa / New Hampshire effect, thus influencing the final result.

Luana:
I'm not sure if it will change the result. I'm more inclined to say that it will reveal the outcome that is more likely in the end. If it really does change the result—

Tarek:
I think there will always be some impact, but—

Matt Huang:
Clearly, there is some influence.

Tarek:
There will definitely be some, just like with opinion polls.

John Collison:
You guys seem a bit modest. You can totally say that the prediction market has value, and there's nothing wrong with changing things.

Tarek:
I just want to emphasize that high odds don't necessarily mean a good outcome. For example, Mamdani reached 94% on Kalshi, and he kept reminding his supporters that they must go out and vote. Because the odds were high, it might make people feel too secure, thinking, "I don't need to go." So this matter is not that simple. I don't think the change that the prediction market brings to reality will be greater than the change brought by opinion polls to reality. It will certainly have an impact, but it doesn't act alone in a vacuum. In reality, there's also social media, chat rooms, various opinion platforms. You are adding the prediction market on top of that, so its marginal impact is not that linear.

But I want to mention something that I find very interesting, whether you find it meaningful or not is up to you. We found that many people, after participating in the prediction market, will take a more serious approach to understanding the underlying events. They really become more knowledgeable. Because once you are ready to bet, or you already have real money at stake, your way of reading information changes. You won't just casually tweet anymore. You will research, read, and understand who this person is, what happened, what they support, what they oppose, what their stance is. This really immerses people in a research state.

Luana:
Right.

Tarek:
This happened in the New York City mayoral election and in the Brexit referendum. Many people voted for Brexit at the time because they thought we should leave the EU; only after voting did they realize, wait, did we really want this? Did we even understand what we were voting for? And the prediction market will give this kind of participation more informational content. Sports leagues actually tell you the same thing. Once the audience places a bet, they will care more about statistics, which player performs well, what really happened in the game. I think, this kind of thing has already begun to happen on a large scale in politics, and this is a good thing.

Luana:
I would even go a step further and say that I think politics would be better for it because candidates' message delivery and policy positions would receive quicker feedback. Now, a candidate talks about ten things, and in the end, whether they win or lose, we can only make one overall judgment, what really worked, what didn't? Even with polls, there is always a lag, and the sample is very limited. But now you can see in real-time, he said this, how did the market react? Such a faster feedback loop, I think, would make politics more like entrepreneurship—you can iterate quickly. All candidates ultimately want to win, and if they can continuously optimize their expression to make it closer to the policies and information the public truly wants, then I think they will ultimately become better because they will have a clearer understanding of what people really want.

Tarek:
It's like you no longer just get one total score, but everything you do has a score.

Luana:
Yes. It's like launching a new feature, and you can see ten metrics at the same time, this one went up, that one went down, and then you can iterate. The market can provide this ability in many things. Music is the same, such as when doing a chart market, a user listens to a song and says, this song will never be a champion. Then you will know, maybe it's time to change direction.

Matt Huang:
Do you also use prediction markets internally to make decisions?

Luana:
Every decision we make is essentially about judging probability. Even election litigation is the same.

Matt Huang:
But that's still your own probability estimate. Have you ever thought about really creating an internal market for employees to participate?

Tarek:
The answer is yes. As a regulated exchange, we actually cannot trade ourselves.

John Collison:
It's a bit like the separation of church and state issue here.

Luana:
Exactly.

Tarek:
So we have been asking the regulators if we can do some small-scale trading ourselves. Like small amounts? Because this is really something we would like to do.

Matt Huang:
After all, you do "eat your own dog food."

Tarek:
Yes, we also want to really use our own product.

Luana:
But this is really difficult.

Tarek:
Because all of our employees are under restrictions.

John Collison:
So, employees can't even transact in a personal capacity. That's just too onerous. As you said, it basically prevents you from dogfooding.

Tarek:
Yeah.

John Collison:
Facebook employees all use Facebook, so they can polish the product better. You can't do that on your end.

Luana:
So, for us, continually asking users becomes particularly important.

John Collison:
That's really interesting.

Tarek:
Even though it's painful, those power users, super forecasters, actually have a huge impact on the product direction.

John Collison:
I guess you spend a lot of time with these power users, super forecasters.

Luana:
Yes, they will win, so of course, we have to try to keep them happy.

John Collison:
One last question, how do you want the policy environment around prediction markets to evolve? If you were talking to folks in government or if you had a magic wand, what would you advocate for?

Tarek:
I think our company's stance may be a bit different from many typical large tech companies. Our stance is that innovation must happen in the U.S., the U.S. must lead on this, and it must win in the right way. Anything Americans want to do and that America as a country needs to win should happen within the U.S. But at the same time, we are also pro-regulation.

At a higher level, there's always a tension between innovation and regulation. Many people's default assumption is that policymakers always want to regulate, and companies always want to avoid regulation.

John Collison:
At this point, you might be more like a traditional financial company rather than a typical Silicon Valley company. Because many Silicon Valley companies grew up in an unregulated environment, but financial companies always had regulators from the start; that's just the reality.

Tarek:
Yes, this is part of the financial culture. After all, it took us four years to navigate the regulation. But we do believe in regulation. I think of regulation as a bit like insurance; it may make you feel constrained at times, but when things really go wrong, it protects you from major harm.

So if you ask me what kind of policy is good, I would say any policy that helps keep these innovations in the United States, ensures that the United States wins this competition, and enhances market fairness and transparency is a good policy. For example, how to make insider trading harder to occur; for example, how to impose more restrictions on government officials, members of Congress in their trading activities to prevent them from using privileged information.

Luana:
We actually just talked about this.

Tarek:
Right. To be honest, I personally strongly support a comprehensive ban on congressional insider trading.

John Collison:
It may even be not just prohibiting insider trading but directly banning congressional trading.

Tarek:
I don't think that would necessarily be a bad idea.

Luana:
That's actually how we see it.

Tarek:
Another point is social equity and transparency. Because we are addressing many public issues, if people can trade around politics, we should make such transaction data as open as possible, allowing anyone to audit and see it. I think that would be a good thing. Imagine if opinion polls could be open to the extreme – you could check every person surveyed, check what the sample looks like – that would be very different.

Another area is user protection. This is very important in the long run. Because as a consumer-facing product becomes mainstream, the platform itself bears a significant educational responsibility. This kind of thing has happened repeatedly in history. In our scenario, we want to ensure that users know what they are doing, do not overtrade, and do not put themselves in uncomfortable situations. Of course, we will do our best at the product and marketing level, but we also need policymakers and regulators to help us turn these requirements into industry standards and assist us in raising the bar on this protection.

Luana:
By the way, I actually think that not only us, even those traditional retail brokers should start adopting many of the user protection measures we are talking about today, which they are not doing right now. Any trading platform targeting retail investors should move in this direction.

Tarek:
Yes, that's probably our overall view. We also hope that the future direction will move in this direction. Because, of course, you can have many different viewpoints, some people may think that all speculation should be banned, whether in the stock market, crypto market, or prediction market. We do not agree with this. We think that would be a very bad outcome, partly because the liquidity market itself has many positive values; and partly because if you really ban it, you will end up amplifying the risks you originally wanted to mitigate because these activities will only move offshore, where you cannot monitor, enforce, or protect users.

John Collison:
Awesome. Thank you all.

[Video Link]

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia