The following is a conversation with Richard Crabe, founder of Numeri, which is a crowdsourced hedge fund, very much in the spirit of Wall Street Bets, but where the trading is done not directly by humans, but by artificial intelligence systems submitted by those humans. It's a fascinating and extremely difficult machine learning competition, where the incentives of everybody is aligned, the code is kept and owned by the people who develop it, the data, anonymized data, is very well organized and made freely available.

I think this kind of idea has a chance to change the nature of stock trading and even just money management in general by empowering people who are interested in trading stocks with the modern and quickly advancing tools of machine learning. Quick mention of our sponsors. Audible Audiobooks, Trial Labs, Machine Learning Company, Blinkist app that summarizes books, and Athletic Greens, all-in-one nutrition drink.

Click the sponsor links to get a discount and to support this podcast. As a side note, let me say that this whole set of events around GameStop and Wall Street Bets has been really inspiring to me as a demonstration that a distributed system, a large number of regular people are able to coordinate and collaborate in taking on the elite centralized power structures, especially when those elites are misbehaving.

I believe that power in as many cases as possible should be distributed, and in this case, the internet as it is for many cases is the fundamental enabler of that power. And at the core, what the internet in its distributed nature represents is freedom. Of course, the thing about freedom is it enables chaos or progress, or sometimes both.

And that's kind of the point of the thing. Freedom is empowering, but ultimately unpredictable. And I think in the end, freedom wins. If you enjoy this podcast, subscribe on YouTube, review it on Apple Podcasts, follow on Spotify, support on Patreon, or connect with me on Twitter at Lex Friedman.

And now here's my conversation with Richard Crabe. From your perspective, can you summarize the important events around this amazing saga that we've been living through of Wall Street Bets, the subreddit and GameStop, and in general, just what are your thoughts about it from a technical to the philosophical level?

- I think it's amazing. It's like my favorite story ever. Like when I was reading about it, I was like, this is the best. And it's also connected with my company, which we can talk about. But what I liked about it is like, I like decentralized coordination and looking at the mechanisms that these are Wall Street Bets users use to hype each other up, to get excited, to prove that they bought the stock and they're holding.

And then also to see that how big of an impact that that decentralized coordination had. It really was a big deal. - Were you impressed by the distributed coordination, the collaboration amongst like, I don't know what the numbers are. I know in numerized looking at the data, after all of this is over and done, it'd be interesting to see like from a large scale distributed system perspective to see how everything played out.

But just from your current perspective, what we know, is it obvious to you that such incredible level of coordination could happen where a lot of people come together in a distributed sense, there's an emergent behavior that happens after that? - No, it's not at all obvious. And one of the reasons is the lack of kind of like, credibility.

To coordinate with someone, you need to kind of make credible contracts or credible claims. So if you have a username on our Wall Street Bets, like some of them are, like Deep Fucking Value is one of them. The actual username, by the way, we're talking about, there's a website called Reddit and there's subreddits on it.

And a lot of people, mostly anonymous, I think for the most part anonymous, can create user accounts and then can just talk on forum like style boards. You should know what Reddit is. If you don't know what Reddit is, check it out. If you don't know what Reddit is, maybe go to the aww subreddit first, A-W-W with cute pictures of cats and dogs.

That's my recommendation. Anyway. That'd be a good start to Reddit. When you get into it more, go to our Wall Street Bets. - It gets dark quickly. Oh, we'll probably talk about that too. So yeah. So there's these users and there's no contracts, like you were saying. - There's no contracts.

The users are anonymous, but there are little things that do help. So for example, if you've posted a really good investment idea in the past, that exists on Reddit as well. And it might have lots of upvotes. And that's also kind of like giving credibility to your next thing.

And then they are also putting up screenshots. Like this is the, here's the trades I've made and here's a screenshot. Now you could fake the screenshot, but still it seems like if you've got a lot of karma and you've had a good performance on the community, it somehow becomes credible enough for other people to be like, you know what?

He actually probably did put a million dollars into this. And you know what? I can follow that trade easily. - And there's a bunch of people like that. So you're kind of integrating all that information together yourself to see like, huh, there's something happening here. And then you jump onto this little boat of like behavior, like we should buy the stock or sell the stock.

And then another person jumps on, another person jumps on. And all of a sudden you have just a huge number of people behaving in the same direction. It's like flock of whatever birds. - Exactly. What was strange with this one, it wasn't just let's all buy Tesla. We love Elon, we love the Tesla, let's all buy Tesla.

Because that we've heard before, right? Everybody likes Tesla. Well, now they do. So what they did with this, in this case, they're buying a stock that was bad. They're buying it because it was bad. And that's really weird because that's a little bit too galaxy brain for a decentralized community.

How did they come up with it? How did they know that was the right one? And the reason they liked it is because it had really, really high short interest. It had been shorted more than its own float, I believe. And so they figured out that if they all bought this bad stock, they could short squeeze some hedge funds.

And those hedge funds would have to capitulate and buy the stock at really, really high prices. - And we should say that shorted means that these are a bunch of people, when you short a stock, you're betting on the, you're predicting that the stock is going to go down and then you will make money if it does.

And then what's a short squeeze? - It's really that if you are a hedge fund and you take a big short position in a company, there's a certain level at which you can't sustain holding that position. There's no limit to how high a stock can go, but there is a limit to how low it can go, right?

So if you short something, you have infinite loss potential. And if the stock doubles overnight, like GameStop did, you're putting a lot of stress on that hedge fund. And that hedge fund manager might have to say, "You know what? "I have to get out of the trade. "And the only way to get out is to buy the bad stock "that they don't want, like they believe will go down." So it's an interesting situation, particularly because it's not zero sum.

If you say, "Let's all get together "and make a bubble in watermelons." You buy a bunch of watermelons, the price goes up, it comes down again. It's a zero sum game. If someone's already shorted a stock and you can make them short squeeze, it's actually a positive sum game.

So yes, some Redditors will make a lot of money, some will lose a lot, but actually the whole group will make money. And that's really why it was such a clever thing for them to do. - And coupled with the fact that shorting, I mean, maybe you can push back, but to me always from an outsider's perspective, seemed, I hope I'm not using too strong of a word, but it seemed almost unethical.

Maybe not unethical, maybe it's just a asshole thing to do. Okay, I'm speaking not from an economics or financial perspective, I'm speaking from just somebody who loves, I'm a fan of a lot of people, I love celebrating the success of a lot of people. And this is like the stock market equivalent of like haters.

I know that's not what it is. I know that there's efficient, you wanna have an economy efficient mechanism for punishing sort of overhyped, overvalued things. That's what shorting I guess is designed for, but it just always felt like these people are just, because they're not just betting on the loss of the company.

It feels like they're also using their leverage and power to manipulate media or just to write articles or just to hate on you on social media. Then you get to see that with Elon Musk and so on. So this is like the man, so people like hedge funds that were shorting are like the sort of embodiment of the evil or just the bad guy, the overpowerful that's misusing their power.

And here's the crowd, the people that are standing up and rising up. So it's not just that they were able to collaborate on Wall Street bets to sort of effectively make money for themselves. It's also that this is like a symbol of the people getting together and fighting the centralized elites, the powerful.

And that, I don't know what your thoughts are about that in general. At this stage, it feels like that's really exciting that people have power, just like regular people have power. At the same time, it's scary a little bit because just studying history, people can be manipulated by charismatic leaders.

And so just like Elon right now is manipulating, encouraging people to buy Dogecoin or whatever, there can be good charismatic leaders and there can be bad charismatic leaders. And so it's nerve wracking. It's a little bit scary how much power Subreddit can have to destroy somebody. Because right now we're celebrating they might be attacking or destroying somebody that everybody doesn't like.

But what if they attack somebody that is actually good for this world? So that, and that's kind of the awesomeness and the price of freedom. It's like it could destroy the world or it can save the world. But at this stage, it feels like, I don't know, overall, when you sit back, do you think this was just a positive wave of emergent behavior?

Is there something negative about what happened? - Well, yeah, the cool thing is the Reddit people weren't doing anything exotic. It was a creative trade, but it wasn't exotic. It was just buying the stock. Okay, maybe they bought some options too. But it was the hedge fund that was doing the exotic thing.

So I like that. It's hard to say, well, we've got together and we've pulled all our money together and now there's a company out there that's worth more. What's wrong with that? - Yeah. - Right? But it doesn't talk about the motivations, which is, and then we destroyed some hedge funds in the process.

- Is there something to be said about the humor and the, I don't know, the edginess, sometimes viciousness of that subreddit? I haven't looked at it too much, but it feels like people can be quite aggressive on there. So is there, what is that? Is that what freedom looks like?

- I think it does, yeah. You definitely need to let people, one of the things that people have compared it to is the Occupy Wall Street, which is, let's say, some very sincere liberals, like 23 years old, whatever, and they go out with signs and they have some kind of case to make.

But this isn't sincere, really. It's like a little bit more nihilistic, a little bit more YOLO, and therefore a little bit more scary because who's scared of the Occupy Wall Street people with the signs? Nobody. But these hedge funds really are scared. I was scared of the Wall Street bats people.

I'm still scared of them. - Yeah, the anonymity is a bit terrifying and exciting. - Yeah. - I mean, yeah, I don't know what to do with it. I've been following events in Russia, for example. It's like there's a struggle between centralized power and the distributed. I mean, that's the struggle of the history of human civilization, right?

But this, on the internet, just that you can multiply people. Some of them don't have to be real. You can probably create bots. It starts getting me, as a programmer, I start to think, hmm, me as one person, how much chaos can I create by writing some bots? - Yeah.

- And I'm sure I'm not the only one thinking that. I'm sure there's hundreds, thousands of good developers out there listening to this, thinking the same thing. And then as that develops further and further in the next decade or two, what impact does that have on financial markets, on just destruction of reputations, or politics, the bickering of left and right political discourse, the dynamics of that being manipulated by, people talk about Russian bots or whatever.

We're probably in the very early stage of that, right? - Exactly. - And this is a good example. So do you have a sense that most of Wall Street Bets folks are actually individual people? Right, that's the feeling I have, is there's just individual, maybe young investors just doing a little bit of an investment, but just on a large scale.

- Yeah, exactly. The reason I found out, I've known about Wall Street Bets for a while, but the reason I found out about GameStop was this, I met somebody at a party who told me about it, and he was like 21 years old, and he's like, it's gonna go up 100%, in the next one day.

- Are we talking about last year? - This was probably, no, this was, yeah, a few days ago. Yeah, it was like maybe two weeks ago or something. So it was already high, GameStop, but it was just strange to me that there was someone telling me at a party how to trade stocks, who was like 21 years old, and I started to look into it, and yeah, and he did make, he made 140% in one day, he was right, and now he's supercharged.

He's a little bit wealthier, and now he's gonna wait for the next thing, and this decentralized entity is just gonna get bigger and bigger. - And they're gonna together search for the next thing. So there's thousands of folks like him, and they're going to probably search for the next thing to attack.

People that have power in this world, that sit there with power right now, in government, in finance, in any kind of position, are probably a little bit scared right now, and honestly, that's probably a little bit good. It's dangerous, but it's good. - Yeah, it certainly makes you think twice about shorting.

It certainly makes you think twice about putting a lot of money into a short. Like these funds put a lot into one or two names, and so it was very, very badly risk managed. - Do you think shorting is, can you speak at a high level, just for your own as a person, is it good for the world?

Is it good for markets? - I do think that there are two kinds of shorting. Evil shorting. (laughing) And chill shorting. Evil shorting is what Melvin Capital was doing. And it's like you put a huge position down, you get all your buddies to also short it, and you start making press, and trying to bring this company down.

And I don't think, in some cases, you go out after fraudulent companies, say this company's a fraud, maybe that's okay. But they weren't even saying it, they're just saying it's a bad company, and we're gonna bring it to the ground, bring it to its knees. A quant fund, like Numerai, we always have lots of positions, and we never have a position that's more than 1% of our fund.

So we actually have, right now, 250 shorts. I don't know any of them, except for one, because it was one of the meme stocks. (laughing) But we shorting them, not to make them go, we don't even want them to go down necessarily. That doesn't sound a bit strange that I say that, but we just want them to not go up as much as our longs.

- Right. - So by shorting a little bit, we can actually go long more in the things we do believe in. So when we were going long in Tesla, we could do it with more money than we had, because we'd borrow from banks, who would lend us money, because we had longs and shorts, because we didn't have market exposure, we didn't have market risk.

And so I think that's a good thing, because that means we can short the oil companies and go long Tesla and make the future come forward faster. And I do think that's not a bad thing. - So we talked about this incredible distributed system created by Wall Street Bets.

And then there's a platform, which is Robinhood, which allows investors to efficiently, as far as you can correct me if I'm wrong, but there's those and there's others and there's Numeri that allow you to make it accessible for people to invest. But that said, Robinhood was, in a centralized way, applied its power to restrict trading on the stock that we're referring to.

Do you have a thought on actually, like the things that happened? I don't know how much you were paying attention to sort of the shadiness around the whole thing. Do you think it was forced to do it? Or was there something shady going on? What are your thoughts in general?

- Well, I think I wanna see the alternate history. Like I wanna see the counterfactual history of them not doing that. - Not doing it. - How bad would it have gotten for hedge funds? How much more damage could have been done if the momentum of these short squeezes could continue?

What happens when there are short squeezes, even if they're in a few stocks, they affect kind of all the other shorts too. And suddenly brokers are saying things like, you need to put up more collateral. So we had a short. It wasn't GameStop, luckily, it was BlackBerry. And it went up like 100% in a day.

It was one of these meme stocks, super bad company. The AIs don't like it, okay? The AIs think it's going down. - What's a meme stock? - A meme stock is kind of a new term for these stocks that catch mimetic momentum on Reddit. And so the meme stocks were GameStop, the biggest one, GameStonk, as Elon calls it, AMC.

And BlackBerry was one, Nokia was one. So these are high short interest stocks as well. So these are targeted stocks. Some people say, isn't it adorable that these people are investing money in these companies that are nostalgic? It's like you go into the AMC movie theater, it's like nostalgic.

It's like, no, it's not why they're doing it. It's that they had a lot of short interest. That was the main thing. And so they were high chance of short squeeze. - In saying, I would love to see an alternate history, do you have a sense that that, what is your prediction of what that history would have looked like?

- Well, you wouldn't have needed very many more days of that kind of chaos to hurt hedge funds. I think it's underrated how damaging it could have been. Because when your shorts go up, your collateral requirements for them go up. Similar to Robinhood. Like we have a prime broker that said to us, you need to put up like $40 per $100 of short exposure.

And then the next day they said, actually you have to put up all of it, 100%. And we were like, what? But if that happens to all the short, all the commonly held hedge fund shorts, because they're all kind of holding the same things. If that happens, not only do you have to cover the short, which means you're buying the bad companies, you need to sell your good companies in order to cover the short.

So suddenly like all the good companies, all the ones that the hedge funds like are coming down and all the ones that the hedge funds hate are going up in a cascading way. So I believe that if you could have had a few more days of GameStop doubling, AMC doubling, you would have had more and more hedge fund deleveraging.

- But so hedge funds, I mean, they get a lot of shit, but do you have a sense that they do some good for the world? I mean, ultimately, so, okay, first of all, Wall Street Bets itself is a kind of distributed hedge fund, Numeri is a kind of hedge fund.

So hedge fund is a very broad category. I mean, like if some of those were destroyed, would that be good for the world? Or would there be coupled with the destroying the evil shorting, would there be just a lot of pain in terms of investment in good companies? - Yeah, a thing I like to tell people if they hate hedge funds is, I don't think you wanna rerun American economic history without hedge funds.

- So on mass, they're good. - Yeah, you really wouldn't want to, because hedge funds are kind of like picking up, they're making liquidity, right, in stocks. And so if you love venture capitalists, they're investing in new technology, it's so good, you have to also kind of like hedge funds, because they're the reason venture capitalists exist, because their companies can have a liquidity event when they go to the public markets.

So it's kind of essential that we have them. There are many different kinds of them. I believe we could maybe get away with only having an AI hedge fund, but we don't necessarily need these evil billions type hedge funds that make the media and try to kill companies, but we definitely need hedge funds.

- Maybe from your perspective, because you run such an organization, and Vlad, the CEO of Robinhood, sort of had to make decisions really quickly, probably had to wake up in the middle of the night kind of thing, and he also had a conversation with Elon Musk on Clubhouse, which I just signed up for.

It was a fascinating, one of the great journalistic performances of our time with Elon Musk. - Pull a surprise for Elon. - How hilarious would it be if he gets a pull surprise? (laughing) And then his Wikipedia would be like, journalist and part-time entrepreneur. - Business magnate. - And business magnate.

I don't know if you can comment on any aspects of that, but if you were Vlad, how would you do things differently? What are your thoughts about his interaction with Elon, how he should have played it differently? - I guess there's a lot of aspects to this interaction. One is about transparency, like how much do you want to tell people about really what went down, there's NDAs potentially involved, how much in private do you want to push back and say no, fuck you, to centralized power?

Whatever the phone calls you're getting, which I'm sure he was getting some kind of phone calls that might not be contractual, like it's not contracts that are forcing him, but he was being, what do you call it, like pressured to behave in certain kinds of ways from all kinds of directions.

Like what do you take from this whole situation? - I was very excited to see Vlad's response. I mean, it was pretty cool to have him talk to Elon. And one of the things that struck me in the first few seconds of Vlad speaking was, I was like, is Vlad like a boomer?

(laughing) Like, but here we are. He seemed like a 55 year old man talking to a 20 year old. Elon was like the 20 year old, and he's like the 55 year old man. You can see why Citadel and him are buddies, right? Like you can, you can see why.

It's like, this is a nice, it's not a bad thing. It's like he's got a respectable, professional attitude. - Well, he also tried to do like a jokey thing, like, no, we're not being ageist here, boomer, but like a 60 year old CEO of Bank of America would try to make a joke for the kids.

That's what Vlad sounded like. - Yeah, I was like, what is this? This guy's like, what is he, 30? - Yeah. - And I'm like, this is weird. - Yeah. - But I think, and maybe that's also what I like about Elon's kind of influence on American business. It's like, he's super like anti the professional.

Like why say, you know, a hundred words about nothing? And so I liked how he was cutting in and saying, Vlad, what do you mean? Spill the beans, bro. - Yeah, so you don't have to be courteous. It's like the first principles thinking, it's like, what the hell happened?

- Yes. - And let's just talk like normal people. The problem of course is, you know, for Elon, it's cost them, what is it? Tens of millions of dollars, his tweeting like that. But perhaps it's a worthy price to pay because ultimately there's something magical about just being real and honest and just going off the cuff and making mistakes and paying for them, but just being real.

And then moments like this, that was an opportunity for Vlad to be that. And it felt like he wasn't. Do you think we'll ever find out what really went down if there was something shady underneath it all? - Yeah, I mean, it would be sad if nothing shady happened.

- Right. - And then his presence made it shady. - Sometimes I feel like that would Mark Zuckerberg, the CEO of Facebook. Sometimes I feel like, yeah, there's a lot of shitty things that Facebook is doing, but sometimes I think he makes it look worse by the way he presents himself about those things.

Like, I honestly think that a large amount of people at Facebook just have a huge, unstable, chaotic system and they're all, not all, but a mass are trying to do good with this chaotic system. But the presentation is like, it sounds like there's a lot of back room conversations that are trying to manipulate people.

And there's something about the realness that Elon has that it feels like CEO should have and Vlad had that opportunity. - I think Mark Zuckerberg had that too when he was younger. - Younger. - And somebody said, you gotta be more professional, man. You can't say, you know, lol to an interview.

And then suddenly he became like this distant person that was hot. Like, you'd rather have him make mistakes, but be honest, than be like professional and never make mistakes. - Yeah, one of the difficult hires, I think, is like marketing people or like PR people is you have to hire people that get the fact that you can say lol on an interview.

Or like, you know, take risks as opposed to what the PR, I've talked to quite a few big CEOs and the people around them are trying to constantly minimize risk of like, what if he says the wrong thing? What if she says the wrong thing? It's like, what? Like, be careful.

It's constantly like, ooh, like, I don't know. And there's this nervous energy that builds up over time with larger, larger teams where the whole thing, like I visited YouTube, for example, everybody I talked to at YouTube, incredible engineering and incredible system, but everybody's scared. Like, let's be honest about this like madness that we have going on of huge amounts of video that we can't possibly ever handle.

There's a bunch of hate on YouTube. There's this chaos of comments, bunch of conspiracy theories, some of which might be true. And then just like this mess that we're dealing with and it's exciting, it's beautiful. It's a place where like democratizes education, all that kind of stuff. And instead they're all like sitting in like, trying to be very polite and saying like, well, we're just want to improve the health of our platform.

Like, it's like this discussion like, all right, man, let's just be real. Let's both advertise how amazing this fricking thing is, but also to say like, we don't know what we're doing. We have all these Nazis posting videos on YouTube. We don't know how to like handle it. And just being real like that, 'cause I suppose that's just a skill.

Maybe it can't be taught, but over time, the whatever the dynamics of the company is, it does seem like Zuckerberg and others get worn down. They just get tired. - Yeah. - They get tired of-- - Not being real. - Of not being real, which is sad. So let's talk about Numeroid, which is an incredible company system idea, I think.

But good place to start. What is Numeroid and how does it work? - So Numeroid is the first hedge fund that gives away all of its data. So this is like probably the last thing a hedge fund would do, right? Why would we give away a data? It's like giving away your edge.

But the reason we do it is because we're looking for people to model our data. And the way we do it is by obfuscating the data. So when you look at Numeroid's data that you can download for free, it just looks like a million rows and of numbers between zero and one.

And you have no idea what the columns mean. But you do know that if you're good at machine learning or have done regressions before, you know that I can still find patterns in this data, even though I don't know what the features mean. - And the data itself is a time series data.

And even though it's obfuscated, anonymized, what is the source data? Like approximately, what are we talking about? - So we are buying data from lots of different data vendors. And they would also never want us to share that data. So we have strict contracts with them. But that's the kind of data you could never buy yourself unless you had maybe a million dollars a year of budget to buy data.

So what's happened with the hedge fund industry is you have a lot of talented people who used to be able to trade and still can trade, but now they have such a data disadvantage, it would never make sense for them to trade themselves. But Numeroid, by giving away this obfuscated data, we can give them a really, really high quality data set that would otherwise be very expensive.

And they can use whatever new machine learning technique they want to find patents in that data that we can use in our hedge fund. - And so how much variety is there in underlying data? We're talking about, I apologize if I'm using the wrong terms but one is just like the stock price.

The other, there's like options and all that kind of stuff, like the, what are they called, order books or whatever. Is there maybe other totally unrelated to directly to the stock market data? Like natural language as well, all that kind of stuff. - Yeah, we were really focused on stock data that's specific to stocks.

So things like you can have like a P, every stock has like a PE ratio. For some stocks, it's not as meaningful, but every stock has that. Every stock has one year momentum, how much they went up in the last year. But those are very common factors. But we try to get lots and lots of those factors that we have for many, many years, like 15, 20 years history.

And then the setup of the problem is commonly in quant called like cross-sectional global equity. You're not really trying to say, I want, I believe the stock will go up. You're trying to say the like relative position of this stock in feature space makes it not a bad buy in a portfolio.

- So it captures some period of time and you're trying to find the patterns, the dynamics captured by the data of that period of time in order to make short-term predictions about what's going to happen. - Yeah, so our predictions are also not that short. We're not really caring about things like order books and tick data, not high frequency at all.

We're actually holding things for quite a bit longer. So our prediction time horizon is about one month. We end up holding stocks for maybe like three or four months. So I kind of believe that's a little bit more like investing than kind of plumbing. Like to go long a stock that's mispriced on one exchange and short on another exchange, that's just arbitrage.

But what we're trying to do is really know something more about the longer term future of the stock. - Yeah, so from the patterns, from these like periods of time series data, you're trying to understand something fundamental about the stock, not like about deep value, about like it's big in the context of the market, is it underpriced, overpriced, all that kind of stuff.

So like, this is about investing. It's not about just like you said, high frequency trading, which I think is a fascinating open question from a machine learning perspective. But just to like sort of build on that. So you've anonymized the data and now you're giving away the data. And then now anyone can try to build algorithms that make investing decisions on top of that data or predictions on the top of that data.

- Exactly. - And so that's, what does that look like? What's the goal of that? What are the underlying principles of that? - So the first thing is, we could obviously model that data in house, right? We can make an XGBoost model on the data and that would be quite good too.

But what we're trying to do is by opening it up and letting anybody participate, we can do quite a lot better than if we modeled it ourselves and a lot better on the stock market doesn't need to be very much. Like it really matters the difference between if you can make 10 and 12% in an equity market neutral hedge fund, because the whole, usually you're charging 2% fees.

So if you can do 2% better, that's like all your fees, it's worth it. So we're trying to make sure that we always have the best possible model as new machine learning libraries come out, new techniques come out, they get automatically synthesized. Like if there's a great paper on supervised learning, someone on Numeri will figure out how to use it on Numeri's data.

- And is there an ensemble of models going on or is it more towards kind of like one or two or three, like best performing models? - So the way we decide on how to weight all of the predictions together is by how much the users are staking on them.

How much of the cryptocurrency that they're putting behind their models. So they're saying, I believe in my model. You can trust me because I'm going to put skin in the game. And so we can take the stake weighted predictions from all our users, add those together, average those together, and that's a much better model than any one model in the sum because ensembling a lot of models together is kind of the key thing you need to do in investing too.

- Yeah, so you're putting, so there's a kind of duality from the user, from the perspective of a machine learning engineer, where you're, it's both a competition, just a really interesting, difficult machine learning problem, and it's a way to invest algorithmically. So like, and, but the way to invest algorithmically also is a way to put skin in the game that communicates to you that you're, the quality of the algorithm, and also forces you to really be serious about the models that you build.

So it's like, everything just works nicely together. Like, I guess one way to say that is the interests are aligned. - Exactly. - Okay, so it's just like poker is not fun when it's like for very low stakes. The higher the stakes, the more the dynamics of the system starts playing out correctly.

Like as a small side note, is there something you can say about which kind, looking at the big broad view of machine learning today, or AI, what kind of algorithms seem to do good in these kinds of competitions at this time? Is there some universal thing you can say, like neural networks suck, recurring neural networks suck, transformers suck, or they're awesome, like old school, sort of more basic kind of classifiers are better, all that.

Is there some kind of conclusions so far that you can say? - There is, there's definitely something pretty nice about tree models, like XGBoost. And they just seem to work pretty nicely on this type of data. So out of the box, if you're trying to come 100th in the competition, in the tournament, maybe you would try to use that.

But what's particularly interesting about the problem that not many people understand, if you're familiar with machine learning, this typically will surprise you when you model our data. So one of the things that you look at in finance is you don't want to be too exposed to any one risk.

Like even if the best sector in the world to invest in over the last 10 years was tech, does not mean you should put all of your money into tech. So if you train a model, it would say, put all your money into tech, it's super good. But what you want to do is actually be very careful of how much of this exposure you have to certain features.

So on Numerai, what a lot of people figure out is actually if you train a model on this kind of data, you want to somehow neutralize or minimize your exposure to these certain features, which is unusual because if you did train a stoplight or stop street detection on computer vision, your favorite feature, let's say you could, and you have an auto encoder and it's figuring out, okay, it's going to be red and it's going to be white.

That's the last thing you want to be, you want to reduce your exposure to. Why would you reduce your exposure to the thing that's helping you, your model the most? And that's actually this counterintuitive thing you have to do with machine learning on financial data. - So reducing, it's reducing your exposure would help you generalize the things that are, so basically financial data has a large amount of patterns that appeared in the past and also a large amount of patterns that have not appeared in the past.

And so like in that sense, you have to reduce the exposure to red lights, to the color red. That's interesting, but how much of this is art and how much of it is science from your perspective so far in terms of as you start to climb from the 100th position to the 95th in the competition?

- Yeah, well, if you do make yourself super exposed to one or two features, you can have a lot of volatility when you're playing numeri. You could maybe very rapidly rise to be high if you were getting lucky. - Yes. - And that's a bit like the stock market.

Sure, take on massive risk exposure, put all your money into one stock and you might make 100%, but it doesn't in the long run work out very well. And so the best users are trying to stay high for as long as possible, not necessarily try to be first for a little bit.

- So to me, a developer, machine learning researcher, how do I, Lex Friedman, participate in this competition and how do others, which I'm sure there'll be a lot of others interested in participating in this competition, what are, let's see, there's like a million questions, but like first one is how do I get started?

- Well, you can go to numeri.ai, sign up, download the data, and on, the data is pretty small. In the data pack you download, there's like an example script, Python script, that just builds a XGBoost model very quickly from the data. And so in a very short time, you can have an example model.

- Is it a particular structure? Like what, is this model then submitted somewhere? So there needs to be some kind of structure that communicates with some kind of API. Like how does the whole, how does your model, once you build it, once you create a little baby Frankenstein, how does it then live in its-- - Okay, well, we want you to keep your baby Frankenstein at home and take care of it.

We don't want it. So you never upload your model to us. You always only giving us predictions. So we never see the code that wrote your model, which is pretty cool. That our whole hedge fund is built from models where we've never ever seen the code. But it's important for the users because it's their IP, why would they wanna give it to us?

- That's brilliant. - So they've got it themselves, but they can basically almost like license the predictions from that model to us. - License the prediction, yeah, yeah. - So-- - Think about it. - What some users do is they set up a compute server and we call it numeric compute.

It's like a little AWS kind of image and you can automate this process. So we can ping you, we can be like, we need more predictions now, and then you send it to us. - Okay, cool. So that's, is that described somewhere, like what the preferred is, the AWS, or whether another cloud platform, is there, I mean, is there sort of specific technical things you wanna say that comes to mind that is a good path for getting started?

So download the data, maybe play around, see if you can modify the basic algorithm provided in the example, and then you would set up a little server on AWS that then runs this model and takes pings and then makes predictions. And so how does your own money actually come into play doing the stake of cryptocurrency?

- Yeah, so you don't have to stake. You can start without staking, and many users might try for months without staking anything at all to see if their model works on the real life data, right? And is not overfit. But then you can get Numeraire many different ways. You can buy it on, you can buy some on Coinbase, you can buy some on Uniswap, you can buy some on Binance.

- So what did you say this is, how do you pronounce it? So this is the Numeraire cryptocurrency. - Yeah, NMR. - NMR, what's, did you just say NMR? - It is technically called Numeraire. - Numeraire, I like it. - Yeah, but NMR is simple. - NMR, Numeraire, okay.

So, and you could buy it, you know, basically anywhere. - Yeah, so it's a bit strange 'cause sometimes people be like, is this like pay to play? - Right. - And it's like, yeah, you need to put some money down to show us you believe in your model. But weirdly, we're not selling you the, like you can't buy the cryptocurrency from us.

- Right. - It's like, it's also, we never, if you do badly, we destroy your cryptocurrency. Okay, that's not good, right? You don't want it to be destroyed. But what's good about it is it's also not coming to us. - Right. - It's not like we win when you lose or something, like we're the house.

Like we're definitely on the same team. - Yes. - You're helping us make a hedge fund that's never been done before. - Yeah, so again, interests are aligned. There's no tension there at all, which is really fascinating. You're giving away everything, and then the IP is owned by, so does the code, you never share the code.

That's fascinating. So, since I have you here, and you said a hundredth, I didn't ask out of how many, so we'll just. (laughing) But if I then, once you get started and you find this interesting, how do you then win or do well, but also how do you potentially try to win if this is something you want to take on seriously from the machine learning perspective, not from a financial perspective?

- Yeah, I think that, first of all, you want to talk to the community. People are pretty open. We give out really interesting scripts and ideas for things you might want to try. But you're also going to need a lot of compute probably. And so some of the best users are, actually the very first time someone won on Numero, I wrote them a personal email.

I was like, "You've won some money. "We're so excited to give you $300." And then they said, "I spent way more on the compute." (laughing) But-- - So this is fundamentally a machine learning problem first, I think, is this is one of the exciting things, I don't know if we'll, in how many ways we can approach this, but really this is less about kind of, no offense, but like finance people, finance-minded people.

They're also, I'm sure, great people. But it feels like from the community that I've experienced, these are people who see finance as a fascinating problem space, source of data, but ultimately they're machine learning people or AI people, which is a very different kind of flavor of community. And I mean, I should say to that, I'd love to participate in this, and I will participate in this.

And I'd love to hear from other people, if you're listening to this, if you're a machine learning person, you should participate in it and tell me, give me some hints how I can do well at this thing. 'Cause this Boomer, I'm not sure I still got it, but 'cause some of it is, it's like Kaggle competitions, like some of it is certainly set of ideas, like research ideas, like fundamental innovation, but I'm sure some of it is like deeply understanding, getting like an intuition about the data.

And then like a lot of it will be like figuring out like what works, like tricks. I mean, you could argue most of deep learning research is just tricks on top of tricks, but there's some of it is just the art of getting to know how to work in a really difficult machine learning problem.

- And I think what's important, the important difference with something like a Kaggle competition, where they'll set up this kind of toy problem, and then there will be an out of sample test, like, hey, you did well out of sample. And this is like, okay, cool. But what's cool with Numera is, the out of sample is the real life stock market.

We don't even know, like we don't know the answer to the problem. We don't, like, you'll have to find out live. And so we've had users who've like submitted every week for like four years, because it's kind of, we say it's the hardest data science problem on the planet, right?

And it sounds maybe sounds like maybe a bit too much for like a marketing thing, but it's the hardest because it's the stock market. It's like, literally there are like billions of dollars at stake and like no one's like letting it be inefficient on purpose. So if you can find something that works on Numera, you really have something that is like working on the real stock market.

- Yeah, because there's like humans involved in the stock market. I mean, it's, you know, you could argue there might be harder data sets, like maybe predicting the weather, all those kinds of things. But the fundamental statement here is, which I like, I was thinking like, is this really the hardest data science problem?

And you start thinking about that, but ultimately it also boils down to a problem where the data is accessible. It's made accessible, made really easy and efficient at like submitting algorithms. So it's not just, you know, it's not about the data being out there, like the weather. It's about making the data super accessible, making, building a community around it.

Like this is what ImageNet did. - Exactly. - Like, it's not just, there's always images. The point is you aggregate them together, you give it a little title, this is community, and that was one of the hardest, right, for a time, and most important data science problems in the world because it was accessible, because it was made sort of like, there was mechanisms by which like standards and mechanisms by which to judge your performance, all those kinds of things.

And NumerEyes, I actually have to step up from that. Is there something more you can say about why, from your perspective, it's the hardest problem in the world? I mean, you said it's connected to the market. So if you can find a pattern in the market, that's a really difficult thing to do because a lot of people are trying to do it.

- Exactly. But there's also the biggest one is it's non-stationary time series. We've tried to regularize the data so you can find patterns by doing certain things to the features and the target. But ultimately, you're in a space where you don't, there's no guarantees that the out-of-sample distributions will conform to any of the training data.

And every single era, which we call on the website, like every single era in the data, which is like sort of showing you the order of the time, even the training data has the same dislocations. And so, yeah, and then there's, yeah, there's so many things that you might wanna try.

There's unlimited possible number of models, right? And so by having it be open, we can at least search that space. - Zooming back out to the philosophical, you said that Numeri is very much like Wall Street Bets. Is there, I think it'd be interesting to dig in why you think so.

I think you're speaking to the distributed nature of the two and the power of the people nature of the two. So maybe can you speak to the similarities and the differences and in which way is Numeri more powerful? In which way is Wall Street Bets more powerful? - Yeah, this is why the Wall Street Bets story is so interesting to me because it's like, feels like we're connected.

And looking at how, just looking at the forum of Wall Street Bets, it's, I was talking earlier about how, how can you make credible claims? You're anonymous. Okay, well, maybe you can take a screenshot. Or maybe you can upvote someone. Maybe you can have karma on Reddit. And those kinds of things make this emerging thing possible.

Numeri, it didn't work at all when we started. It didn't work at all. Why? People made multiple accounts. They made really random models and hoped they would get lucky. And some of them did. - Yes. - Staking was our like solution to, could we make it so that we could trust, we could know which model people believed in the most.

And we could weight models that had high stake more and effectively coordinate this group of people to be like, well, actually there's no incentive to creating bot accounts anymore. Either I stake my accounts, in which case I should believe in them 'cause I could lose my stake, or I don't.

And that's a very powerful thing that having a negative incentive and a positive incentive can make things a lot better. And staking is like this, is this really nice like key thing about blockchain. It's like something special you can do where they're not even trusting us with their stake in some ways.

They're trusting the blockchain, right? So the incentives, like you say, it's about making these perfect incentives so that you can have coordination to solve one problem. And nowadays I sleep easy because I have less money in my own hedge fund than our users are staking on their models. - That's powerful.

In some sense, from a human psychology perspective, it's fascinating that the Wall Street bets worked at all. Amidst that chaos emerging behavior, like behavior that made sense emerged. It would be fascinating to think if numerized style staking could then be transferred to places like Reddit, and not necessarily for financial investments, but I wish sometimes people would have to stake something in the comments they make on the internet.

That's the problem with anonymity, is like anonymity is freedom and power that you don't have to, you can speak your mind, but it's too easy to just be shitty. - Exactly. - So this, I mean, you're making me realize from like a profoundly philosophical aspect, numerized staking is a really clean way to solve this problem.

It's a really beautiful way. Of course, it only with numerized currently works for a very particular problem, right? Not for human interaction on the internet, but that's fascinating. - Yeah, there's nothing to stop people. In fact, we've open sourced like the code we use for staking in a protocol we call erasure.

And if Reddit wanted to, they could even use that code to have enabled staking on our Wall Street bets. And they're actually researching now, they've had some Ethereum grants on how could they have more crypto stuff in there, in Ethereum, because wouldn't that be interesting? Like imagine you could, instead of seeing a screenshot, like guys, I promise I will not sell my GameStop.

We're just gonna go huge. We're not gonna sell at all. And here is a smart contract, which no one in the world, including me, can undo. That says, I have staked millions against this claim. - That's powerful. - And then what could you do? - And of course, it doesn't have to be millions.

It could be just very small amount, but then just a huge number of users doing that kind of stake. - Exactly. - That could change the internet. - It would change, and then Wall Street. - It would change Wall Street. They would never have been able to, they would still be short squeezing one day after the next, every single hedge fund collapsing.

- If we look into the future, do you think it's possible that numeri-style infrastructure where AI systems backed by humans are doing the trading is what the entirety of the stock market is, or the entirety of the economy, is run by basically this army of AI systems with high level human supervision?

- Yeah, the thing is that some of them could be bad actors. - Some of the humans? - No, well, these systems could be tricky. So actually I once met a hedge fund manager, this is kind of interesting. He said, very famous one, and he said, sometimes we can see things in the market where we know we can make money, but it will mesh it up.

We know we can make money, but it will mess things up. And we choose not to do those things. And on the one hand, maybe this is like, oh, you're being super arrogant. Of course you can't do this, but maybe he can. And maybe he really isn't doing things he knows he could do, but would change, be pretty bad.

Would the Reddit army have that kind of morality or concern for what they're doing? Probably not based on what we've seen. - The madness of crowds. There'll be like one person that says, hey, maybe, and then they get trampled over. That's the terrifying thing, actually. A lot of people have written about this, is somehow that like little voice, that's human morality, gets silenced when we get into groups and start chanting.

And that's terrifying. But I think maybe I misunderstood. I thought that you're saying AI systems can be dangerous, but you just described how humans can be dangerous. So which is safer? - So one thing is, so Wall Street bets these kinds of attacks, it's not possible to model numerized data and then come up with the idea from the model, let's short squeeze, just game stop.

It's not even framed in that way. It's not like possible to have that idea. But it is possible for like kind of a bunch of humans. So I think this, numeri could get very powerful without it being dangerous, but Wall Street bets needs to get a little bit more powerful and it'll be pretty dangerous.

- Yeah, well, I mean, so this is a good place to kind of think about numeri data today. Numeri signals and what that looks like in 10, 20, 30, 50, 100 years. Like right now, I guess maybe you can correct me, but the data that we're working with is like a window.

It's a anonymized obfuscated window into a particular aspect, time period of the market. And you can expand that more and more and more and more potentially. You can imagine in different dimensions to where it encapsulates all the things that where you could include kind of human to human communication that was available for like to buy GameStop, for example, on Wall Street bets.

So maybe the step back, can you speak to what is numeri signals and what are the different data sets that are involved? - So with numeri signals, you're still providing predictions to us but you can do that from your own data sets. So numeri, it's all, you have to model our data to come up with predictions.

Numeri signals is whatever data you can find out there, you can turn it into a signal and give it to us. So it's a way for us to import signals on data we don't yet have. And that's why it's particularly valuable because it's gonna be signals, you're only rewarded for signals that are orthogonal to our core signal.

So you have to be doing something uncorrelated. And so strange alternative data tends to have that property. There isn't too many other signals that are correlated with what's happening on Wall Street bets. That's not gonna be like correlated with the price to earnings ratio, right? And we have some users as of recently, as of like a week ago, there was a user that created, I think he's in India.

He created a signal that is scraped from Wall Street bets. And now we have that signal as one of our signals in thousands that we use at Numeri. - And the structure of the signal is similar. So is this just numbers and time series data? - It's exactly, and it's just like, it's kind of a, you're providing a ranking of stocks.

So you just say, give a one means you like the stock, zero means you don't like the stock and you provide that for 5,000 stocks in the world. - And they somehow converted the natural language that's in the Wall Street bet. - Exactly, so there's, and they made, they open sourced this Colab notebook.

You can go and see it and look at it. And so, yeah, it's taking that, making a sentiment score and then turning it into a rank of stocks. - A sentiment score. - Yeah. - Like this stock sucks or this stock is awesome. And then converting that's, that's fast.

Just even looking at that data would be fascinating. So on the signal side, what's the vision? This long-term, what do you see that becoming? - So we wanna manage all the money in the world. That's NumerEyes mission. And to get that, we need to have all the data and have all of the talent.

Like there's no way, so first principles, if you had really good modeling and really good data that you would lose, right? It's just a question of how much do you need to get really good? So NumerEye already has some really nice data that we give out. This year, we are 10X-ing that.

And I actually think we'll 10X the amount of data we have on NumerEye every year for at least the next 10 years. - Wow. - So it's gonna get very big, the data we give out. And signals is more data. People with any other random dataset can turn that into a signal and give it to us.

- And in some sense, that kind of data is the edge cases, the weirdness is the, so you're focused on like the bulk, like the main data, and then there's just like weirdness from all over the place that just can enter through this back door of the process. - Exactly, and it's also a little bit shorter term.

So the signals are about a seven day time horizon, and on NumerEye, it's like a 30 day. So it's often for faster situations. - You've written about a master plan, and you've mentioned, which I love, in a similar sort of style of big style thinking, you would like NumerEye to manage all of the world's money.

So how do we get there from yesterday to several years from now? Like what is the plan? You've already started to allure to it, to get all the data and get it-- - Yeah. - And get the talent, humans, models. - Exactly, I mean, the important thing to note there is, what would that mean, right?

And I think the biggest thing it means is like, if there was one hedge fund, you would have not so much talent wasted on all the other hedge funds. Like it's super weird how the industry works. It's like one hedge fund gets a data source and hires a PhD, and another hedge fund has to buy the same data source and hire a PhD, and suddenly a third of American PhDs are working at hedge funds, and we're not even on Mars.

And like, so in some ways, NumerEye, it's all about freeing up people who work at hedge funds to go work for Elon. - Yeah, and also the people who are working on NumerEye problem, it feels like a lot of the knowledge there is also transferable to other domains. - Exactly.

One of our top users is, he works at NASA Jet Propulsion Lab. And he's like amazing. I went to go visit him there. And it's like, he's got like NumerEye posters, and it looks like the movies, like it looks like Apollo 11 or whatever. Yeah, the point is he didn't quit his job to join full time.

He's working on getting us to Jupiter's moon. That's his mission, the Europa Clipper mission. - Actually, literally what you're saying. - Literally. He's smart enough that we really want his intelligence to reach the stock market, 'cause the stock market's a good thing, hedge funds are a good thing, all kinds of hedge funds, especially.

But we don't want him to quit his job. So he can just do NumerEye on the weekends. And that's what he does. He just made a model, and it just automatically submits to us. And he's like one of our best users. - You mentioned briefly that stock markets are good.

From my sort of outsider perspective, is there a sense, do you think trading stocks is closer to gambling, or is it closer to investing? Sometimes it feels like it's gambling, as opposed to betting on companies to succeed. And this is maybe connected to our discussion of shorting in general.

But from your sense, the way you think about it, is it fundamentally still investing? - I do think, I mean, it's a good question. I've also seen lately people say, this is like speculation. Is there too much speculation in the market? And it's like, but all the trades are speculative.

Like all the trades have a horizon. Like people want them to work. So I would say that, there's certainly a lot of aspects of gambling math that applies to investing. Like one thing you don't do in gambling is put all your money in one bet. You have bankroll management, and it's a key part of it.

And small alterations to your bankroll management might be better than improvements to your skill. So there, and then there are things we care about in our fund, like we wanna make a lot of independent bets. We talk about it, like we wanna make a lot of independent bets, because that's gonna be a higher sharp than if you have a lot of bets that depend on each other, like all in one sector.

But yeah, I mean, the point is that you want the prices of the stocks to be reflective of how, of their value. - Of the underlying value. - Yeah, like you shouldn't have there be like a hedge fund that's able to say, well, I've looked at some data and all of this stuff's super mispriced.

Like that's super bad for society if it looks like that to someone. - I guess the underlying question then is, do you see that the market often like drifts away from the underlying value of companies and it becomes a game in itself? Like would these, whatever they're called, like derivatives, like the option, options and shorting and all that kind of stuff, it's like layers of game on top of the actual, what you said, which is like the basic thing that the Wall Street Bets was doing, which is like just buying stocks.

- Yeah, there are a lot of games that people play that are in the derivatives market. And I think a lot of the stuff people dislike when they look at the history of what's happened, they hate like credit default swaps or collateralized debt obligations. Like these are the kind of like enemies of 2008.

And then the long-term capital management thing, it was like they had 30 times leverage or something. Just that no one, like you could just go to a gas station and ask anybody at the gas station, is it a good idea to have 30 times leverage? And they just say no.

It's like common sense just like went out the window. So yeah, I don't respect long-term capital management. (laughing) - Okay, but Nimrod doesn't actually use any derivatives unless you call shorting derivative. We do put money into companies. We, and that does help the companies we're investing in. It's just in little ways.

We really did buy Tesla and it did. And we were not, we played some role in its success. Super small, make no mistake. But still, I think that's important. - Can I ask you a pothead question, which is what is money, man? So if we just kind of zoom out and look at, 'cause let's talk to you about cryptocurrency, which perhaps could be the future of money.

In general, how do you think about money? You said Nimrod, the vision, the goal is to run, to manage the world's money. What is money in your view? - I don't have a good answer to that, but it's definitely in my personal life, it's become more and more warped.

And you start to care about the real thing, like what's really going on here. Elon talks about things like this, like what is a company really? And it's like, it's a bunch of people who like kind of show up to work together and like they solve a problem. And there might not be a stock out there that's trading that represents what they're doing, but it's not the real thing.

And being involved in crypto, like I put in a crowd sale of Ethereum and all these other things and different crypto hedge funds and things that I've invested in. And it's just kind of like, it feels like how I used to think about money stuff is just like totally warped.

Because you stop caring about the price and you just care about the product. - So by the product, you mean like the different mechanisms that money is exchanged. I mean, money is ultimately a kind of, on one is a store of wealth, but it's also a mechanism of exchanging wealth.

But what wealth means becomes a totally different thing, especially with cryptocurrency to where it's almost like these little contracts, these little agreements, these transactions between human beings that represent something that's bigger than just like cash being exchanged at 7-Eleven, it feels like. - Yeah, maybe I'll answer what is finance?

It's what are you doing when you have the ability to take out a loan? You can bring a whole new future into being with finance. If you couldn't get a student loan to get a college degree, you couldn't get a college degree if you didn't have the money. But now like weirdly you can get it with, and like, yeah, all you have is this like loan, which is like, so now you can bring a different future into the world.

And that's how when I was saying earlier about if you rerun American history, economic history without these things, like you're not allowed to take out loans, you're not allowed to have derivatives, you're not allowed to have money, it just doesn't really work. And it's a really magic thing, how much you can do with finance by kind of bringing the future forward.

- Finance is empowering. It's, we sometimes forget this, but it enables innovation, it enables big risk takers and bold builders that ultimately make this world better. You said you were early in on cryptocurrency. Can you give your high level overview of just your thoughts about the past, present and future of cryptocurrency?

- Yeah, so my friends told me about Bitcoin and I was interested in equities a lot. And I was like, well, it has no net present value. It has no future cash flows. Bitcoin pays no dividends. So I really couldn't get my head around it. And like that this could be valuable.

And then I, but I did, so I didn't feel like I was early in cryptocurrency. In fact, 'cause I was like, it was like 2014, it felt like a long time after Bitcoin. And then, but then I really liked some of the things that Ethereum was doing. It seemed like a super visionary thing.

Like I was reading something that was, that was just gonna change the world when I was reading the white paper. And I liked the different constructs you could have inside of Ethereum that you couldn't have on Bitcoin. - Like smart contracts and all that kind of stuff. - Exactly, yeah.

And even the, they were, yeah, even spoke about different, yeah, different constructions you could have. - Yeah, that's a cool dance between Bitcoin and Ethereum of it's in the space of ideas. Feels so young. Like I wonder what cryptocurrencies will look like in the future. Like if Bitcoin or Ethereum 2.0 or some version will stick around or any of those, like who's gonna win out or if there's even a concept of winning out at all.

Is there a cryptocurrency that you're especially find interesting that technically, financially, philosophically you think is something you're keeping your eye on? - Well, I don't really, I'm not looking to like invest in cryptocurrencies anymore, but I, they are, I mean, and many are almost identical. I mean, there's not, there wasn't too much difference between even Ethereum and Bitcoin in some ways, right?

But there are some that I like the privacy ones. I mean, I was like, I like Zcash for it's like coolness. It's actually, it's like a different kind of invention compared to some of the other things. - Okay, can you speak to just briefly to privacy? What is there some mechanisms of preserving some privacy of the, so I guess everything is public.

- Yeah. - Is that the problem? - Yeah, none of the transactions are private. And so, even like I have some of my, I have some numeraire and you can just see it. In fact, you can go to a website and says like, you can go to like ether scan and it'll say like numeraire founder.

And I'm like, how the hell do you guys know this? - So they can reverse engineer, whatever that's called. - Yeah, and so they can see me move it too. They can see me, oh, why is he moving it? So, but yeah, Zcash, then they also, when you can make private transactions, you can also play different games.

- Yes. - And it's unclear, it's like, what's quite cool about Zcash is I wonder what games are being played there. No one will know. - So from a deeply analytical perspective, can you describe why Dogecoin is going to win? Which it surely will. Like it very likely will take over the world.

And once we expand out into the universe, we'll take over the universe. Or on a more serious note, like what are your thoughts on the recent success of Dogecoin where you've spoken to sort of the meme stocks, the memetics of the whole thing. It feels like the joke can become the reality like the meme, the joke has power in this world.

- Yeah. - It's fascinating. - Exactly. It's like, why is it correlated with Elon tweeting about it? - It's not just Elon alone tweeting, right? It's like Elon tweeting and that becomes a catalyst for everybody on the internet kind of like spreading the joke, right? - Exactly. - The joke of it.

So it's the initial spark of the fire for Wall Street bets type of situation. - Yeah. - And that's fascinating because jokes seem to spread faster than other mechanisms. - Yeah. - Like funny shit is very effective at captivating the discourse on the internet. - Yeah, and I think you can have, like I like the one meme, like Doge, I haven't heard that name in a long time.

(laughing) - Yeah. - Like I think back to that meme often. That's like funny. And every time I think back to it, there's a little probability that I might buy it. - Buy some Dogecoin. - Right, and so I imagine you just have millions of people who have had all these great jokes told to them.

And every now and then they reminisce, oh, that was really funny. And then they're like, let me buy some. - Wouldn't that be interesting if like the entirety, if we travel in time, like multiple centuries, where the entirety of the communication of the human species is like humor. Like it's all just jokes.

Like we're high on probably some really advanced drugs. And we're all just laughing nonstop. It's a weird, like dystopian future of just humor. Elon has made me realize how like good it feels to just not take shit seriously every once in a while and just relieve like the pressure of the world.

At the same time, the reason I don't always like when people finish their sentences with lol is like that you don't, when you don't take anything seriously. When everything becomes a joke, then it feels like that way of thinking feels like it will destroy the world. It's like, I often think it like, will memes save the world or destroy?

Because I think both are possible directions. - Yeah, I think this is a big problem. I mean, America, I always felt that about America, a lot of people are telling jokes kind of all the time and they're kind of good at it. And you take someone aside, an American, you're like, I really want to have a sincere conversation.

It's like hard to even keep a straight face because everything is so, there's so much levity. So it's complicated. I like how sincere actually like your Twitter can be. You're like, I am in love with the world today. - I get so much shit for it. It's hilarious. I'm never gonna stop because I realized like, you have to be able to sometimes just be real and be positive and just be, say the cliche things, which ultimately those things actually capture some fundamental truths about life.

But it's a dance. And I think Elon does a good job of that. Now from an engineering perspective of being able to joke, but everyone's mostly to pull back and be like, here's real problems, let's solve them and so on. And then be able to jump back to a joke.

So it's ultimately, I think, I guess a skill that we have to learn. But I guess your advice is to invest everything anyone listening owns into Dogecoin. That's what I heard from this interaction. - Yeah, no, exactly. Yeah. My hedge fund is unavailable. Just go straight to Dogecoin. - You're running a successful company.

It's just interesting 'cause my mind has been in that space of potentially being one of the millions of other entrepreneurs. What's your advice on how to build a successful startup? How to build a successful company? - I think that one thing I do like, and it might be a particular thing about America, but there is something about playing, tell people what you really want to happen in the world.

Don't stop. It's not gonna make it, like if you're asking someone to invest in your company, don't say, I think maybe one day we might make a million dollars. When you actually believe something else, you actually believe, you're actually more optimistic, but you're toning down your optimism because you want to appear low risk.

But actually it's super high risk if your company becomes mediocre because no one wants to work in a mediocre company, no one wants to invest in a mediocre company. So you should play the real game. And obviously this doesn't apply to all businesses, but if you play a venture-backed startup kind of game, like play for keeps, play to win, go big.

And it's very hard to do that. I've always feel like, you can start narrowing your focus because 10 people are telling you, you got to care about this boring thing that won't matter five years from now. And you should push back and play the real game. - So be bold.

So both, I mean, there's an interesting duality there. So there's the way you speak to other people about like your plans and what you are like privately, just in your own mind. - And maybe it's connected with what you were saying about, yeah, sincerity as well. Like if you appear to be sincerely optimistic about something that's big or crazy, it's putting yourself up to be kind of like ridiculed or something.

And so if you say, my mission is to, yeah, go to Mars. It's just so bonkers that it's hard to say. - It is. But one powerful thing, just like you said, is if you say it and you believe it, then actually amazing people come and work with you.

- Exactly. - It's not just skill, but the dreams. There's something about optimism, like that fire that you have when you're optimistic of actually having the hope of building something totally cool, something totally new, that when those people get in a room together, like they can actually do it.

- Yeah. - And also it makes life really fun when you're in that room. So all of that together, ultimately, I don't know. That's what makes this crazy ride of a startup really look fun. And Elon is an example of a person who succeeded at that. There's not many other inspiring figures, which is sad.

I used to be at Google and there's something that happens that sometimes when the company grows bigger and bigger and bigger, where that kind of ambition kind of quiets down a little bit. - Yeah. - Google had this ambition, still does, of making the world's information accessible to everyone.

And I remember, I don't know, that's beautiful. I still love that dream of, they used to scan books, but just in every way possible, make the world's information accessible. Same with Wikipedia. Every time I open up Wikipedia, I'm just awe-inspired by how awesome humans are, man. At creating this together, I don't know what the meanings are over there, but it's just beautiful.

Like what they've created is incredible. And I'd love to be able to be part of something like that. And you're right, for that you have to be bold. - And it's strange to me also, I think you're right that there's how many boring companies there are. Something I always talk about, especially in FinTech, it's like, why am I excited about this?

This is so lame. Like what is, this isn't even important. Even if you succeed, this is gonna be terrible. This is not good. And it's just strange how people can kind of get fake enthusiastic about boring ideas when there's so many bigger ideas that, yeah, I mean, you read these things, like this company raises money, and it's just like, that's a lot of money for the worst idea I've ever heard.

- Some ideas are really big. So like I worked on autonomous vehicles quite a bit, and there's so many ways in which you can present that idea to yourself, to the team you work with, to just, yeah, like to yourself when you're quietly looking in the mirror in the morning, that's really boring or really exciting.

Like if you're really ambitious with autonomous vehicles, it changes the nature of like human robot interaction, it changes the nature of how we move. Forget money, forget all that stuff. It changes like everything about robotics and AI, machine learning, it changes everything about manufacturing. I mean, the cars, the transportation is so fundamentally connected to cars, and if that changes, it's changing the fabric of society, of movies, of everything.

And if you go bold and take risks and be willing to go bankrupt with your company, as opposed to cautiously, you could really change the world. And it's so sad for me to see all these autonomous companies, autonomous vehicle companies, they're like really more focused about fundraising and kind of like smoke and mirrors.

They're really afraid, the entirety of their marketing is grounded in fear and presenting enough smoke to where they keep raising funds so they can cautiously use technology of a previous decade or previous two decades to kind of test vehicles here and there, as opposed to do crazy things and bold and go huge at scale to huge data collection.

So that's just an example. Like the idea can be big, but if you don't allow yourself to take that idea and think really big with it, then you're not gonna make anything happen. Yeah, you're absolutely right in that. So you've been connected in your work with a bunch of amazing people.

How much interaction do you have with investors? I get that whole process is an entire mystery to me. Is there some people that just have influence on the trajectory of your thinking completely? Or is it just this collective energy that they behind the company? Yeah, I mean, I came here and I was amazed how, yeah, people would, I was only here for a few months and I met some incredible investors and I almost run out of money.

And once they invested, I was like, I am not gonna let you down. And I was like, okay, I'm gonna send them like an email update every like three minutes. And then they don't care at all. So they kind of wanna, I don't know. Like, so for some, I like it when it's just like, they're always available to talk, but a lot of building a business, especially a high tech business, there's little for them to add, right?

There's little for them to add on product. There's a lot for them to add on like business development. And if we are doing product research, which is for us research into the market, research into how to make a great hedge fund. And we do that for years, there's not much to tell the investors.

So they're basically is like, I believe in you. There's something, I like the cut of your jib. There's something in your idea, in your ambition, in your plans that I like. And it's almost like a pat on the back. It's like, go get them kid. - Yeah, it is a bit like that.

And that's cool. That's a good way to do it. I'm glad they do it that way. Like the one in meeting I had, which was like really good with this was meeting Howard Morgan, who's actually a co-founder of Renaissance Technologies in the like 1980s and worked with Jim Simons.

And he was in the room and I was meeting some other guy and he was in the room. And I was explaining how quantitative finance works. I was like, so, you know, they use mathematical models. And then he was like, yeah, I started Renaissance. I know a bit about this.

And then I was like, oh my God. So yeah, and then I think he kind of said, well, yeah, he said, well, 'cause I was talking, he was working at First Round Capital as a partner. And they kind of said they didn't want to invest. And then I wrote a blog post describing the idea.

And I was like, I really think you guys should invest. And then they end up. - Oh, interesting. You convinced them. - Yeah, 'cause they were like, we don't really invest in hedge funds. And I was like, you don't see like what I'm doing here. This is a tech company, not a hedge fund, right?

- Yeah, and Numerai is brilliant. It's, when it caught my eye, there's something special there. So I really do hope you succeed in the, obviously it's a risky thing you're taking on, the ambition of it, the size of it. But I do hope you succeed. You mentioned Jim Simons.

He comes up in another world of mine really often. He's just a brilliant guy on the mathematics side as a mathematician, but he's also a brilliant finance, hedge fund manager guy. Have you gotten a chance to interact with him at all? Have you learned anything from him on the math, on the finance, on the philosophy life side of things?

- I've played poker with him. It was pretty cool. It was like, actually in the show "Billions", they kind of do a little thing about this poker tournament thing with all the hedge fund managers. And that's real life thing. And they have a lot of like World Series of Bracelet, World Series of Poker Bracelet holders.

But it's kind of Jim's thing. And I met him there and yeah, it was kind of brief, but I was just like, he's like, "Oh, how do you, why are you here?" And I was like, "Oh, Howard sent me." He's like, "Go play this tournament, meet some of the other players." And then- - Was it Texas Hold'em?

- Yeah, Texas Hold'em tournament. Yeah. - Do you play poker yourself or was it- - Yeah, I do. I mean, it was crazy. On my right was the CEO, who's the current CEO of Renaissance, Peter Brown. And Peter Muller, who's a hedge fund manager at PDT. And yeah, I mean, it was just like, and then just everyone.

And then all these bracelet World Series, like people that I know from like TV. And Robert Mercer, who's fucking crazy. - Who's that? - He's the guy who donated the most money to Trump. And he's just like- - It's a lot of personality. - Character, yeah. Jeez, it's crazy.

So it's quite cool how, yeah, like the, it was really fun. And then I managed to knock out Peter Muller. I have a, I got a little trophy for knocking him out because he was a previous champion. In fact, I think he's won the most. I think he's won three times.

Super smart guy. But I will say Jim outlasted me in the tournament. And they're all extremely good at poker. But they're also, so it was a $10,000 buy-in. And I was like, this is kind of expensive, but it all goes to charity, Jim's math charity. But then the way they play, they have like rebuys.

And like, they all do a shit ton of rebuys. This is for charity. So immediately they're like going all in and I'm like, man, like, so I ended up, you know, adding more as well. So the stakes- It's like you couldn't play at all without doing that. Yeah, the stakes are high.

But you're connected to a lot of these folks that are kind of titans of just, of economics and tech in general. Do you feel a burden from this? You're a young guy. I did feel a bit out of place there. Like the company was quite new and they also don't speak about things, right?

So it's not like going to meet a famous rocket engineer who will tell you how to make a rocket. They do not want to tell you anything about how to make a hedge fund. It's like all secretive. And that part I didn't like. And they were also kind of making fun of me a little bit.

Like they would say, like they'd call me like, I don't know, the Bitcoin kid. Yeah. And then they would say even things like, I remember Peter said to me something like, I don't think AI is going to have a big role in finance. And I was like, hearing this from the CEO of Renaissance was like weird to hear because I was like, of course it will.

And he's like, but he can see, he's like, I can see it having a really big impact on things like self-driving cars. But finance, it's too noisy and whatever. And so I don't think it's like the perfect application. And I was like, that was interesting to hear. 'Cause it's like, and I think there was that same day that Libra, I think it is, the poker playing AI started to beat like the human.

So it was kind of funny hearing them like say, oh, I'm not sure AI could ever attack that problem. And then that very day it's attacking the problem of the game we're playing. - Well, there's a kind of a magic to somebody who's exceptionally successful, looking at you, giving you respect, but also saying that what you're doing is not going to succeed in a sense.

Like they're not really saying it, but I tend to believe from my interactions with people that it's a kind of prod to say, like prove me wrong. - Yeah. - That's ultimately, that's how those guys talk. They see good talent and they're like. - Yeah. And I think they're also saying it's not gonna succeed quickly in some way.

They're like, this is gonna take a long time and maybe that's good to know. - And certainly AI in trading, that's one of the most philosophically interesting questions about artificial intelligence and the nature of money, because it's like, how much can you extract in terms of patterns from all of these millions of humans interacting using this methodology of money?

It's like one of the open questions in the artificial intelligence. In that sense, you converting it to a dataset is one of the biggest gifts to the research community, to the whole, anyone who loves data science and AI, this is kind of fascinating. And I'd love to see where this goes actually.

- Thing I say sometimes, long before AGI destroys the world, a narrow intelligence will win all the money in the stock market. Way, like just a narrow AI. - Yeah. - And I don't know if I'm gonna be the one who invents that. So I'm building Numerai to make sure that that narrow AI uses our data.

(laughing) - So you're giving a platform to where millions of people can participate and do build that narrow AI themselves. People love it when I ask this kind of question about books, about ideas and philosophers and so on. I was wondering if you had books or ideas, philosophers, thinkers that had an influence on your life when you were growing up or just today that you would recommend that people check out, blog posts, podcasts, videos, all that kind of stuff.

Is there something that just kind of had an impact on you that you can recommend? - A super kind of obvious one. I really was reading "Zero to One" while coming up with Numerai. It's like, I was like halfway through the book. And I really do like a lot of the ideas there.

And it's also about kind of thinking big and also it's like peculiar little book. It's like, why, like there's a little picture of the hipster versus Unabomber. And it's a weird little book. So I like, there's kind of like some depth there. - In terms of a book on a, if you're thinking of doing a startup, it's a good book.

- A book I like a lot is maybe my favorite book is David Deutsch's "Beginning of Infinity." I just found that so optimistic. It puts you, everything you read in science, it like makes the world feel like kind of colder. 'Cause like, it's like, you know, we're just coming from evolution and coming from nothing should be this way or whatever.

And humans are not very powerful. We're just like scum on the earth. And the way David Deutsch sees things and argues, he argues them with the same rigor that the cynics often use and then has a much better conclusion. That's, you know, some of the statements of things like, you know, anything that doesn't violate the laws of physics can be solved.

- So ultimately arriving at a hopeful, like a hopeful path forward. - Yeah, without being like a hippie. - You've mentioned kind of advice for startups. Is there in general, whether you do a startup or not, do you have advice for young people today? You're like an example of somebody who's paved their own path and were, I would say, exceptionally successful.

Is there advice, somebody who's like 20 today, 18, undergrad or thinking about going to college or in college and so on that you would give them? - I think I often tell young people don't start companies. Is it not, don't start a company unless you're prepared to make it your life's work.

Like that's a really good way of putting it. And a lot of people think, well, you know, this semester I'm gonna take a semester off and in that one semester, I'm gonna start a company and sell it or whatever. And it's just like, what are you talking about? It doesn't really work that way.

You should be like super into the idea, so into it that you wanna spend a really long time on it. - Is that more about psychology or actually time allocation? Like, is it literally the fact that you need to give 100% for potentially years for it to succeed? Or is it more about just the mindset that's required?

- Yeah, I mean, I think, well, any, I think, yeah, you don't wanna have, certainly don't wanna have a plan to sell the company like quickly or something. What's like, or it's like a company that has a very, it's like a big fashion component. Like it'll only work now.

It's like an app or something. So yeah, that's a big one. And then I also think something I've thought about recently is I had a job as a quant at a fund for about two and a half years. And part of me thinks if I had spent another two years there I would have learned a lot more and had even more knowledge to be where, to basically accelerate how long Numerai took.

So the idea that you can sit in an air conditioned room and get free food, or even sit at home now in your underwear and make a huge amount of money and learn whatever you want and get, it's just crazy. It's such a good deal. - Yeah. Oh, that's interesting.

That's the case for, I was terrified of that. Like at Google, I thought I would become really comfortable in that air conditioned room. And that I was afraid the quant situation is, I mean, what you present is really brilliant that it's exceptionally valuable, the lessons you learn 'cause you get to get paid while you learn from others.

If you see that, if you see jobs in the space of your passion that way, that it's just an education. It's like the best kind of education. But of course you have, from my perspective you have to be really careful on that to get comfortable. Again, a relationship, then you buy a house or whatever the hell it is.

And then you get, and then you convince yourself like, well, I have to pay these fees for the car, for the house, blah, blah, blah. And then there's momentum and all of a sudden you're on your deathbed and there's grandchildren and you're drinking whiskey and complaining about kids these days.

So I'm afraid of that momentum, but you're right. Like there's something special about the education you get working at these companies. - Yeah, and I remember on my desk, I had like a bunch of papers on quant finance, a bunch of papers on optimization and then the paper on Ethereum, just on my desk as well.

And the white paper, and it's like, it's amazing how kind of, and you can learn about, so that, I also thought, I think this idea of like learning about intersections of things. I don't think there are too many people that know like as much about crypto and quant finance and machine learning as I do.

And that's a really nice set of three things to know stuff about. And that was because I had like free time in my job. - Okay, let me ask the perfectly impractical, but the most important question. What's the meaning of all the things you're trying to do so many amazing things, why?

What's the meaning of this life of yours or ours? - I don't know. - Humans. - Yeah, so have you had people say, asking what meaning of life is, is like asking the wrong question or something. - The question is wrong. - Yeah. - No, usually people get too nervous to be able to say that 'cause it's like, your question sucks.

I don't think there's an answer. It's like the searching for it. It's like sometimes asking it. It's like sometimes sitting back and looking up at the stars and being like, huh, I wonder if there's aliens up there. There's a useful like a palate cleanser aspect to it 'cause it kind of wakes you up to like all the little busy, hurried day-to-day activities, all the meetings, all the things you like a part of.

We're just like ants, a part of a system, a part of another system. And then asking this bigger question allows you to kind of zoom out and think about it. But there's ultimately, I think it's an impossible thing for a limited cognitive capacity to capture. But it's fun to listen to somebody who's exceptionally successful, exceptionally busy now, who's also young like you, to ask these kinds of questions about like death.

Do you consider your own mortality kind of thing? And life, whether that enters your mind because it often doesn't. It kind of almost gets in the way. - Yeah, it's amazing how many things you can like that are trivial that could like occupy a lot of your mind until something bad happens or something flips you.

- And then you start thinking about the people you love that are in your life. Then you start thinking about like, holy shit, this ride ends. - Exactly, yeah. I just had COVID and I had it quite bad. It wasn't really bad. It was just like, I also got a simultaneous like lung infection.

So I had like almost like bronchitis or whatever. I don't even, I don't understand that stuff, but I started, and then you're forced to be isolated. - Right. - And so it's actually kind of nice because it's very depressing. And then I've heard stories of, I think it's Sean Parker.

He had like all these diseases as a child and he had to like just stay in bed for years. And then he like made Napster. It's like pretty cool. So yeah, I had about 15 days of this recently, just last month. And it feels like it did shock me into a new kind of energy and ambition.

- Were there moments when you were just like terrified at the combination of loneliness? And like, you know, the thing about COVID is like there's some degree of uncertainty. Like it feels like it's a new thing, a new monster that's arrived on this earth. And so, you know, dealing with it alone, a lot of people are dying.

It's like wondering like-- - Yeah, you do wonder, I mean, for sure. And then there are even new strains in South Africa, which is where I was. And maybe the new strain had some interaction with my genes and I'm just gonna die. - But ultimately it was liberating somehow.

- I loved it. Oh, I love that I got out of it. - Okay. - 'Cause it also affects your mind. You get confusion and kind of a lot of fatigue and you can't do your usual tricks of psyching yourself out of it. So, you know, sometimes it's like, oh man, I feel tired.

Okay, I'm just gonna go have coffee and then I'll be fine. It's like, now it's like, I feel tired, I don't even wanna get out of bed to get coffee 'cause I feel so tired. And then you have to confront, there's no like quick fix cure and you're trapped at home.

- So now you have this little thing that happened to you that was a reminder that you're mortal and you get to carry that flag in trying to create something special in this world, right? With Neumeri. Listen, this was like one of my favorite conversation 'cause the way you think about this world of money and just this world in general is so clear and you're able to explain it so eloquently.

Richard, it was really fun. Really appreciate you talking to me. - Thank you, thank you. - Thanks for listening to this conversation with Richard Kray and thank you to our sponsors. Audible Audiobooks, Trial Labs, Machine Learning Company, Blinkist app that summarizes books and Athletic Greens, all-in-one nutrition drink. Click the sponsor links to get a discount and to support this podcast.

And now let me leave you with some words from Warren Buffett. "Games are won by players who focus on the playing field, not by those whose eyes are glued to the scoreboard." Thank you for listening and hope to see you next time. (upbeat music) (upbeat music)