Transcript
00:00:00 The following is a conversation with Richard Crabe,
00:00:02 founder of Numeri, which is a crowdsourced hedge fund,
00:00:07 very much in the spirit of Wall Street Bets,
00:00:09 but where the trading is done not directly by humans,
00:00:13 but by artificial intelligence systems
00:00:15 submitted by those humans.
00:00:18 It’s a fascinating and extremely difficult
00:00:21 machine learning competition
00:00:22 where the incentives of everybody is aligned,
00:00:26 the code is kept and owned by the people who develop it,
00:00:29 the data, anonymized data is very well organized
00:00:33 and made freely available.
00:00:35 I think this kind of idea has a chance to change
00:00:38 the nature of stock trading
00:00:40 and even just money management in general
00:00:42 by empowering people who are interested in trading stocks
00:00:47 with the modern and quickly advancing tools
00:00:50 of machine learning.
00:00:51 Quick mention of our sponsors,
00:00:54 Audible Audio Books,
00:00:55 Trial Labs Machine Learning Company,
00:00:57 Blinkist app that summarizes books,
00:01:00 and Athletic Greens, all in one nutrition drink.
00:01:04 Click the sponsor links to get a discount
00:01:06 and to support this podcast.
00:01:08 As a side note, let me say that this whole set of events
00:01:11 around GameStop and Wall Street Bets
00:01:14 has been really inspiring to me as a demonstration
00:01:18 that a distributed system,
00:01:20 a large number of regular people
00:01:24 are able to coordinate and collaborate
00:01:26 in taking on the elite centralized power structures,
00:01:32 especially when those elites are misbehaving.
00:01:36 I believe that power in as many cases as possible
00:01:39 should be distributed.
00:01:40 And in this case, the internet, as it is for many cases,
00:01:44 is the fundamental enabler of that power.
00:01:48 And at the core, what the internet
00:01:50 in its distributed nature represents is freedom.
00:01:53 Of course, the thing about freedom
00:01:55 is it enables chaos or progress, or sometimes both.
00:02:02 And that’s kind of the point of the thing.
00:02:04 Freedom is empowering, but ultimately unpredictable.
00:02:09 And I think in the end, freedom wins.
00:02:12 If you enjoy this podcast, subscribe on YouTube,
00:02:15 review it on Apple Podcasts, follow on Spotify,
00:02:18 support on Patreon, or connect with me on Twitter
00:02:21 at Lex Friedman.
00:02:23 And now, here’s my conversation with Richard Crabe.
00:02:28 From your perspective, can you summarize
00:02:29 the important events around this amazing saga
00:02:32 that we’ve been living through of Wall Street Bets,
00:02:34 the subreddit and GameStop, and in general,
00:02:38 just what are your thoughts about it
00:02:39 from a technical to the philosophical level?
00:02:42 I think it’s amazing.
00:02:42 It’s like my favorite story ever.
00:02:46 Like when I was reading about it,
00:02:47 I was like, this is the best.
00:02:49 And it’s also connected with my company,
00:02:53 which we can talk about.
00:02:54 But what I liked about it is like,
00:02:57 I like decentralized coordination
00:02:59 and looking at the mechanisms
00:03:01 that these are Wall Street Bets users use
00:03:04 to hype each other up, to get excited,
00:03:08 to prove that they bought the stock and they’re holding.
00:03:12 And then also to see that how big of an impact
00:03:15 that that decentralized coordination had.
00:03:18 So it really was a big deal.
00:03:21 Were you impressed by the distributed coordination,
00:03:24 the collaboration amongst like,
00:03:26 I don’t know what the numbers are.
00:03:27 I know I’m numerized looking at the data.
00:03:30 After all of this is over and done,
00:03:32 it’d be interesting to see like
00:03:34 from a large scale distributed system perspective
00:03:38 to see how everything played out.
00:03:40 But just from your current perspective, what we know,
00:03:44 is it obvious to you that such incredible level
00:03:47 of coordination could happen
00:03:49 where a lot of people come together in a distributed sense,
00:03:52 there’s an emergent behavior that happens after that.
00:03:54 No, it’s not at all obvious.
00:03:57 And one of the reasons is the lack of credibility.
00:04:01 To coordinate with someone,
00:04:02 you need to make credible contracts or credible claims.
00:04:06 So if you have a username on our Wall Street Bets,
00:04:11 like some of them are, like deep fucking value
00:04:14 is one of them.
00:04:15 That’s an actual username.
00:04:16 By the way, we’re talking about,
00:04:18 there’s a website called Reddit
00:04:19 and there’s subreddits on it.
00:04:21 And a lot of people, mostly anonymous,
00:04:24 I think for the most part anonymous,
00:04:27 can create user accounts
00:04:28 and then can then just talk on forum like style boards.
00:04:31 You should know what Reddit is.
00:04:32 If you don’t know what Reddit is, check it out.
00:04:34 If you don’t know what Reddit is,
00:04:36 maybe go to the awesome subreddit first,
00:04:40 aww with cute pictures of cats and dogs.
00:04:43 That’s my recommendation.
00:04:44 Anyway.
00:04:44 Okay, yeah, that would be a good start to Reddit.
00:04:46 When you get into it more, go to our Wall Street Bets.
00:04:48 It gets dark quickly.
00:04:50 We’ll probably talk about that too.
00:04:53 So yeah, so there’s these users
00:04:56 and there’s no contracts, like you were saying.
00:04:58 There’s no contracts, the users are anonymous,
00:05:01 but there are little things that do help.
00:05:03 So for example,
00:05:04 if you’ve posted a really good investment idea in the past,
00:05:07 that exists on Reddit as well.
00:05:10 And it might have lots of upvotes.
00:05:12 And that’s also kind of like giving credibility
00:05:14 to your next thing.
00:05:17 And then they are also putting up screenshots,
00:05:20 like here’s the trades I’ve made and here’s a screenshot.
00:05:26 Now you could fake the screenshot,
00:05:28 but still it seems like if you’ve got a lot of karma
00:05:32 and you’ve had a good performance on the community,
00:05:35 it somehow becomes credible enough
00:05:37 for other people to be like, you know what?
00:05:39 He actually probably did put a million dollars into this.
00:05:43 And you know what, I can follow that trade easily.
00:05:46 And there’s a bunch of people like that.
00:05:47 So you’re kind of integrating all that information
00:05:50 together yourself to see like, huh,
00:05:52 there’s something happening here.
00:05:53 And then you jump onto this little boat of like behavior,
00:05:57 like we should buy the stock or sell the stock.
00:06:00 And then another person jumps on,
00:06:02 another person jumps on.
00:06:04 And all of a sudden you have just a huge number of people
00:06:07 behaving in the same direction.
00:06:08 It’s like flock of whatever birds.
00:06:10 Exactly.
00:06:11 What was strange with this one,
00:06:12 it wasn’t just let’s all buy Tesla.
00:06:15 We love Elon, we love Tesla, let’s all buy Tesla.
00:06:18 Because that we’ve heard before, right?
00:06:21 Everybody likes Tesla.
00:06:23 Well, now they do.
00:06:26 So what they did with this in this case,
00:06:29 they’re buying a stock that was bad.
00:06:31 They’re buying it because it was bad.
00:06:33 And that’s really weird because that’s a little bit
00:06:37 too galaxy brain for a decentralized community.
00:06:41 How did they come up with it?
00:06:43 How did they know that was the right one?
00:06:44 And the reason they liked it
00:06:46 is because it had really, really high short interest.
00:06:49 It had been shorted more than its own float, I believe.
00:06:54 And so they figured out that if they all bought
00:06:57 this bad stock, they could short squeeze some hedge funds.
00:07:03 And those hedge funds would have to capitulate
00:07:05 and buy the stock at really, really high prices.
00:07:08 And we should say that shorted means that
00:07:10 these are a bunch of people, when you short a stock,
00:07:13 you’re betting on the, you’re predicting
00:07:16 that the stock’s going to go down
00:07:17 and then you will make money if it does.
00:07:19 And then what’s a short squeeze?
00:07:22 It’s really that if you are a hedge fund
00:07:24 and you take a big short position in a company,
00:07:28 there’s a certain level at which
00:07:31 you can’t sustain holding that position.
00:07:34 There’s no limit to how high a stock can go,
00:07:37 but there is a limit to how low it can go, right?
00:07:39 So if you short something, you have infinite loss potential.
00:07:43 And if the stock doubles overnight, like GameStop did,
00:07:48 you’re putting a lot of stress on that hedge fund.
00:07:51 And that hedge fund manager might have to say,
00:07:53 you know what, I have to get out of the trade.
00:07:56 And the only way to get out is to buy the bad stock
00:07:59 that they don’t want, like they believe will go down.
00:08:02 So it’s an interesting situation,
00:08:05 particularly because it’s not zero sum.
00:08:09 If you say, let’s all get together
00:08:11 and make a bubble in watermelons,
00:08:13 you buy a bunch of watermelons,
00:08:14 the price goes up, it comes down again,
00:08:18 it’s a zero sum game.
00:08:20 If someone’s already shorted a stock
00:08:22 and you can make them short squeeze,
00:08:24 it’s actually a positive sum game.
00:08:25 So yes, some Redditors will make a lot of money,
00:08:28 some will lose a lot,
00:08:29 but actually the whole group will make money.
00:08:32 And that’s really why it was such a clever thing
00:08:36 for them to do.
00:08:38 And coupled to the fact that shorting,
00:08:40 I mean, maybe you can push back,
00:08:42 but to me always from an outsider’s perspective,
00:08:45 seemed, I hope I’m not using too strong of a word,
00:08:48 but it seemed almost unethical.
00:08:51 Maybe not unethical, maybe it’s just a asshole thing to do.
00:08:55 Okay, I’m speaking not from an economics
00:08:57 or financial perspective,
00:08:58 I’m speaking from just somebody who loves,
00:09:02 I’m a fan of a lot of people,
00:09:03 I love celebrating the success of a lot of people.
00:09:07 And this is like the stock market equivalent of like haters.
00:09:13 I know that’s not what it is.
00:09:14 I know that there’s efficient,
00:09:15 you wanna have an economy efficient mechanism
00:09:18 for punishing sort of overhyped, overvalued things.
00:09:23 That’s what shorthand guess is designed for.
00:09:26 But it just always felt like these people are just,
00:09:29 because they’re not just betting on the loss of the company.
00:09:33 It feels like they’re also using their leverage and power
00:09:37 to manipulate media or just to write articles
00:09:40 or just to hate on you on social media.
00:09:43 Then you get to see that with Elon Musk and so on.
00:09:46 So this is like the man,
00:09:50 so people like hedge funds that were shorting
00:09:53 are like the sort of embodiment of the evil
00:09:58 or just the bad guy, the overpowerful
00:10:01 that’s misusing their power.
00:10:02 And here’s the crowd,
00:10:04 the people that are standing up and rising up.
00:10:06 So it’s not just that they were able to collaborate
00:10:10 on Wall Street bets to sort of effectively
00:10:13 make money for themselves.
00:10:14 It’s also that this is like a symbol
00:10:17 of the people getting together
00:10:19 and fighting the centralized elites, the powerful.
00:10:23 And that, I don’t know what your thoughts are
00:10:27 about that in general.
00:10:28 At this stage, it feels like that’s really exciting
00:10:32 that people have power,
00:10:35 just like regular people have power.
00:10:38 At the same time, it’s scary a little bit
00:10:40 because just studying history,
00:10:44 people could be manipulated by charismatic leaders.
00:10:49 And so just like Elon right now is manipulating,
00:10:54 encouraging people to buy Dogecoin or whatever,
00:10:58 there can be good charismatic leaders
00:11:00 and there can be bad charismatic leaders.
00:11:02 And so it’s nerve wracking.
00:11:04 It’s a little bit scary how much power
00:11:06 a subreddit can have to destroy somebody
00:11:11 because right now we’re celebrating
00:11:12 they might be attacking or destroying somebody
00:11:15 that everybody doesn’t like,
00:11:17 but what if they attack somebody
00:11:19 that is actually good for this world?
00:11:21 So that, and that’s kind of the awesomeness
00:11:25 and the price of freedom.
00:11:28 It’s like it could destroy the world
00:11:31 or it can save the world.
00:11:32 But at this stage, it feels like, I don’t know,
00:11:35 overall, when you sit back,
00:11:36 do you think this was just a positive wave
00:11:39 of emergent behavior?
00:11:41 Is there something negative about what happened?
00:11:43 Well, yeah, the cool thing is they weren’t doing anything,
00:11:47 the Reddit people weren’t doing anything exotic.
00:11:50 It was a creative trade, but it wasn’t exotic.
00:11:55 It wasn’t, it was just buying the stock.
00:11:57 Okay, maybe they bought some options too,
00:12:00 but it was the hedge fund that was doing the exotic thing.
00:12:04 So I liked that.
00:12:06 It was, it’s hard to say, well, we’ve got together
00:12:10 and we’ve pulled all our money together
00:12:12 and now there’s a company out there that’s worth more.
00:12:16 What’s wrong with that?
00:12:17 Yeah. Right?
00:12:18 But it doesn’t talk about the motivations, which is,
00:12:21 and then we destroyed some hedge funds in the process.
00:12:25 Is there something to be said about the humor
00:12:28 and the, I don’t know, the edginess,
00:12:31 sometimes viciousness of that subreddit?
00:12:34 I haven’t looked at it too much,
00:12:36 but it feels like people can be quite aggressive on there.
00:12:40 So is there, what is that?
00:12:45 Is that what Freedom looks like?
00:12:49 I think it does, yeah.
00:12:50 You definitely need to let people,
00:12:52 one of the things that people have compared it to
00:12:54 is the Occupy Wall Street, which is, let’s say,
00:12:58 some very sincere liberals, like 23 years old, whatever,
00:13:03 and they go out with signs
00:13:04 and they have some kind of case to make.
00:13:08 But this isn’t sincere, really.
00:13:11 It’s like a little bit more nihilistic,
00:13:14 a little bit more YOLO, and therefore a little bit
00:13:18 more scary because who’s scared of the Occupy Wall Street
00:13:22 people with the signs?
00:13:23 Nobody.
00:13:24 But these hedge funds really are scared.
00:13:26 I was scared of the Wall Street bats people.
00:13:29 I’m still scared of them.
00:13:31 Yeah, the anonymity is a bit terrifying and exciting.
00:13:36 Yeah.
00:13:37 I mean, yeah, I don’t know what to do with this.
00:13:40 I’ve been following events in Russia, for example.
00:13:43 It’s like there’s a struggle between centralized power
00:13:46 and the distributed.
00:13:47 I mean, that’s the struggle of the history
00:13:50 of human civilization, right?
00:13:51 But this on the internet, just that you can multiply people.
00:13:57 Like some of them don’t have to be real.
00:13:59 Like you can probably create bots.
00:14:01 Like it starts getting me, me as a programmer,
00:14:05 I start to think like, hmm, me as one person,
00:14:08 how much chaos can I create by writing some bots?
00:14:12 Yeah.
00:14:13 And I’m sure I’m not the only one thinking that.
00:14:17 There’s, I’m sure there’s hundreds, thousands
00:14:19 of good developers out there listening to this,
00:14:22 thinking the same thing.
00:14:23 And then as that develops further and further
00:14:26 in the next like decade or two,
00:14:28 what impact does that have on financial markets,
00:14:30 on just destruction of reputations of just,
00:14:37 or politics, the bickering of left and right
00:14:41 political discourse, the dynamics of that
00:14:44 being manipulated by, you know,
00:14:46 people talk about like Russian bots or whatever.
00:14:49 We’re probably in a very early stage of that, right?
00:14:52 Yeah, exactly.
00:14:53 And this is a good example.
00:14:56 So do you have a sense that most of WallStreetBets folks
00:15:00 are actually individual people, right?
00:15:02 That’s the feeling I have is they’re just individual,
00:15:05 maybe young investors, just doing a little bit
00:15:07 of an investment, but just on a large scale.
00:15:10 Yeah, exactly.
00:15:11 The reason I found out, I’ve known about WallStreetBets
00:15:13 for a while, but the reason I found out about GameStop
00:15:15 was this, just I met somebody at a party
00:15:17 who told me about it and he was like 21 years old
00:15:20 and he’s like, it’s gonna go up 100% in the next one day.
00:15:23 That we’re talking about in last year?
00:15:26 This was probably, no, this was, yeah, a few days ago.
00:15:29 Yeah, it was like maybe two weeks ago or something.
00:15:34 So it was already high GameStop,
00:15:37 but it was just strange to me that there was someone
00:15:39 telling me at a party how to trade stocks
00:15:42 who was like 21 years old.
00:15:44 And I started to, yeah, I started to look into it.
00:15:49 And yeah, and he did make, he made it, yeah,
00:15:52 he made 140% in one day, he was right.
00:15:55 And now he’s supercharged, he’s a little bit wealthier
00:15:58 and now he’s gonna wait for the next thing.
00:16:01 And this decentralized entity
00:16:03 is just gonna get bigger and bigger.
00:16:04 And they’re gonna together search for the next thing.
00:16:06 So there’s thousands of folks like him
00:16:08 and they’re going to probably search
00:16:10 for the next thing to attack.
00:16:11 People that have power in this world that sit there
00:16:14 with power right now in government and in finance
00:16:18 and any kind of position
00:16:20 are probably a little bit scared right now.
00:16:22 And honestly, that’s probably a little bit good.
00:16:26 It’s dangerous, but it’s good.
00:16:28 Yeah, it certainly makes you think twice about shorting.
00:16:31 It certainly makes you think twice
00:16:32 about putting a lot of money into a short.
00:16:35 Like these funds put a lot into one or two names.
00:16:38 And so it was very, very badly risk managed.
00:16:41 Do you think shorting is, can you speak at a high level
00:16:46 just for your own as a person, is it good for the world?
00:16:50 Is it good for markets?
00:16:52 I do think that the two kinds of shorting,
00:16:55 evil shorting and chill shorting.
00:17:00 Evil shorting is what Melvin Capital was doing.
00:17:04 And it’s like, you put a huge position down,
00:17:08 you get all your buddies to also short it
00:17:10 and you start making press
00:17:12 and trying to bring this company down.
00:17:16 And I don’t think in some cases,
00:17:20 you go out after like fraudulent companies say,
00:17:22 this company is a fraud.
00:17:24 Maybe that’s okay.
00:17:24 Like some, but they weren’t even saying,
00:17:27 they’re just saying it’s a bad company
00:17:29 and we’re going to bring it to the ground,
00:17:30 bring it to its knees.
00:17:32 So a quant fund like Numerai,
00:17:37 we always have lots of positions
00:17:39 and we never have a position
00:17:40 that’s like more than 1% of our fund.
00:17:43 So we actually have right now, 250 shorts.
00:17:48 I don’t know any of them except for one
00:17:52 because it was one of the meme stocks.
00:17:53 But we shorting them not to make them go,
00:17:58 we don’t even want them to go down necessarily.
00:18:01 That doesn’t sound a bit strange that I say that,
00:18:03 but we just want them to not go up as much as our longs.
00:18:09 So by shorting a little bit,
00:18:13 we can actually go long more
00:18:15 in the things we do believe in.
00:18:16 So when we were going long in Tesla,
00:18:19 we could do it with more money than we had
00:18:22 because we would borrow from banks
00:18:24 who would lend us money to go down.
00:18:27 Who would lend us money because we had longs and shorts,
00:18:32 because we didn’t have market exposure,
00:18:34 we didn’t have market risk.
00:18:35 And so I think that’s a good thing
00:18:37 because that means we can short the oil companies
00:18:41 and go long Tesla and make the future come forward faster.
00:18:44 And I do think that’s not a bad thing.
00:18:47 So we’ve talked about this incredible distributed system
00:18:51 created by Wall Street Bets.
00:18:52 And then there’s a platform which is Robinhood,
00:18:56 which allows investors to efficiently as far,
00:18:59 you can correct me if I’m wrong,
00:19:00 but there’s those and there’s others in this new MRI
00:19:04 that allow you to make it accessible for people to invest.
00:19:09 But that said, Robinhood was in a centralized way
00:19:15 applied its power to restrict trading
00:19:17 on the stocks that we’re referring to.
00:19:19 Do you have a thoughts on actually
00:19:21 like the things that happened?
00:19:23 I don’t know how much you were paying attention
00:19:26 to sort of the shadiness around the whole thing.
00:19:30 Do you think it was forced to do it?
00:19:32 Or was there something shady going on?
00:19:34 What are your thoughts in general?
00:19:36 Well, I think I wanna see the alternate history.
00:19:39 Like I wanna see the counterfactual history
00:19:43 of them not doing that.
00:19:44 How bad would it have gotten for hedge funds?
00:19:47 How much more damage could have been done
00:19:50 if the momentum of these short squeezes could continue?
00:19:53 What happens when there are short squeezes,
00:19:57 even if they’re in a few stocks,
00:20:00 they affect kind of all the other shorts too.
00:20:02 And suddenly brokers are saying things like,
00:20:05 you need to put up more collateral.
00:20:07 So we had a short, it wasn’t GameStop luckily,
00:20:12 it was Blackberry and it went up like 100% in a day.
00:20:15 It was one of these meme stocks, super bad company.
00:20:17 The AIs don’t like it, okay?
00:20:19 The AIs think it’s going down.
00:20:21 What’s a meme stock?
00:20:22 A meme stock is kind of a new term for these stocks
00:20:26 that catch memetic momentum on Reddit.
00:20:32 And so the meme stocks were GameStop, the biggest one,
00:20:36 GameStonk, as Elon calls it, AMC.
00:20:39 And Blackberry was one, Nokia was one.
00:20:44 So these are high short interest stocks as well.
00:20:47 So these are targeted stocks.
00:20:50 Some people say, oh, isn’t it adorable
00:20:53 that these people are investing money
00:20:56 in these companies that are nostalgic?
00:20:59 It’s like, you go into the AMC movie theater,
00:21:01 it’s like nostalgic.
00:21:02 It’s like, no, it’s not why they’re doing it.
00:21:05 It’s that they had a lot of short interest.
00:21:07 That was the main thing.
00:21:08 And so there were high chance of short squeeze.
00:21:11 In saying, I would love to see an alternate history,
00:21:14 do you have a sense that that,
00:21:17 what is your prediction of what that history would look like?
00:21:19 Well, you wouldn’t have needed very many more days
00:21:23 of that kind of chaos to hurt hedge funds.
00:21:28 I think it’s underrated how damaging it could have been.
00:21:32 Because when your shorts go up,
00:21:38 your collateral requirements for them go up.
00:21:40 It’s similar to Robinhood.
00:21:41 Like we have a prime broker that says, said to us,
00:21:46 you need to put up like $40 per $100 of short exposure.
00:21:51 And then the next day they said,
00:21:52 actually you have to put up all of it, 100%.
00:21:56 And we were like, what?
00:21:58 But if that happens to all the short,
00:22:02 all the commonly held hedge fund shorts,
00:22:05 because they’re all kind of holding the same things.
00:22:08 If that happens, not only do you have to cover the short,
00:22:13 which means you’re buying the bad companies,
00:22:16 you need to sell your good companies
00:22:18 in order to cover the short.
00:22:20 So suddenly like all the good companies,
00:22:23 all the ones that the hedge funds like are coming down
00:22:26 and all the ones that the hedge funds hate are going up
00:22:29 in a cascading way.
00:22:33 So I believe that if you could have had a few more days
00:22:37 of GameStop doubling, AMC doubling,
00:22:40 you would have had more and more hedge fund deleveraging.
00:22:45 But so hedge funds, I mean, they get a lot of shit,
00:22:48 but they, do you have a sense
00:22:51 that they do some good for the world?
00:22:53 I mean, ultimately, so, okay.
00:22:55 First of all, Wall Street bets itself
00:22:57 is a kind of distributed hedge fund.
00:22:59 Numeri is a kind of hedge fund.
00:23:01 So I got, hedge fund is a very broad category.
00:23:04 I mean, like if some of those were destroyed,
00:23:07 would that be good for the world?
00:23:08 Or would there be coupled
00:23:11 with the destroying the evil shorting,
00:23:15 would there be just a lot of pain
00:23:16 in terms of investment in good companies?
00:23:18 Yeah, a thing I like to tell people
00:23:21 if they hate hedge funds is,
00:23:23 I don’t think you want to rerun American economic history
00:23:27 without hedge funds.
00:23:29 So on mass they’re, yeah, they’re good.
00:23:34 Yeah, you really wouldn’t want to.
00:23:36 Because hedge funds are kind of like picking up,
00:23:38 they’re making liquidity, right, in stocks.
00:23:41 And so if you love venture capitalists,
00:23:45 they’re investing in new technology, it’s so good.
00:23:48 You have to also kind of like hedge funds
00:23:50 because they’re the reason venture capitalists exist
00:23:54 because their companies can have a liquidity event
00:23:56 when they go to the public markets.
00:23:58 So it’s kind of essential that we have them.
00:24:01 There are many different kinds of them.
00:24:04 I believe we could maybe get away
00:24:06 with only having an AI hedge fund.
00:24:11 But we don’t necessarily need
00:24:13 these evil billions type hedge funds
00:24:15 that make the media and try to kill companies.
00:24:18 But we definitely need hedge funds.
00:24:20 Maybe from your perspective,
00:24:21 because you run such an organization
00:24:27 and Vlad, the CEO of Robinhood,
00:24:30 sort of had to make decisions really quickly,
00:24:32 probably had to wake up
00:24:33 in the middle of the night kind of thing.
00:24:36 And he also had a conversation with Elon Musk on Clubhouse,
00:24:40 which I just signed up for.
00:24:42 It was a fascinating,
00:24:43 one of the great journalistic performances of our time
00:24:47 with Elon Musk.
00:24:49 Pull a surprise for Elon.
00:24:51 How hilarious would it be if he gets a pull a surprise?
00:24:55 And then his Wikipedia would be like journalist
00:24:58 and part time entrepreneur.
00:25:00 Business magnate.
00:25:01 Business magnate.
00:25:02 As you know, I don’t know if you can comment
00:25:05 on any aspects of that,
00:25:06 but like, if you were Vlad,
00:25:08 how would you do things differently?
00:25:10 What are your thoughts about his interaction with Elon?
00:25:13 How he should have played it differently?
00:25:15 Like, I guess there’s a lot of aspects to this interaction.
00:25:19 One is about transparency.
00:25:20 Like how much do you want to tell people
00:25:23 about really what went down?
00:25:24 There’s NDAs potentially involved.
00:25:27 How much in private do you want to push back
00:25:32 and say, no, fuck you, to centralize power?
00:25:36 Whatever the phone calls you’re getting,
00:25:37 which I’m sure he was getting some kind of phone calls
00:25:40 that might not be contractual.
00:25:42 Like it’s not contracts that are forcing him,
00:25:44 but he was being, what do you call it?
00:25:47 Like pressured to behave in certain kinds of ways
00:25:49 from all kinds of directions.
00:25:51 Like what do you take from this whole situation?
00:25:55 I was very excited to see Vlad’s response.
00:25:58 I mean, it’s pretty cool to have him talk to Elon.
00:26:00 And one of the things that like struck me
00:26:02 in the first like few seconds of Vlad speaking was like,
00:26:06 I was like, is Vlad like a boomer?
00:26:10 Like, but hear me out.
00:26:13 Like he seemed like a 55 year old man
00:26:16 talking to a 20 year old.
00:26:18 Elon was like the 20 year old.
00:26:19 And he’s like the 55 year old man.
00:26:21 You can see why Citadel are NMR buddies, right?
00:26:25 Like you can, you can see why.
00:26:27 It’s like, this is a nice, it’s not a bad thing.
00:26:30 It’s like, he’s got a respectable professional attitude.
00:26:35 Well, he also tried to do like a jokey thing.
00:26:38 Like, no, we’re not being ageist here.
00:26:41 Boomer, but like a 60 year old CEO of Bank of America
00:26:46 would try to make a joke for the kids.
00:26:48 That’s what Vlad’s like.
00:26:49 Yeah, I was like, what is this?
00:26:51 This guy’s like, what is he, 30?
00:26:53 Yeah.
00:26:55 And I’m like, this is weird.
00:26:56 Yeah.
00:26:58 But I think, and maybe that’s also what I like
00:27:00 about Elon’s kind of influence on American business
00:27:03 is like, he’s super like anti the professional.
00:27:07 Like why say, you know, a hundred words about nothing?
00:27:13 And so I liked how he was cutting in and saying,
00:27:15 Vlad, what do you mean?
00:27:16 Spill the beans, bro.
00:27:17 Yeah, so you don’t have to be courteous.
00:27:19 It’s like the first principles thinking.
00:27:21 It’s like, what the hell happened?
00:27:23 Yes.
00:27:24 Let’s just talk like normal people.
00:27:26 The problem of course is, you know, for Elon,
00:27:31 it’s cost them, what is it?
00:27:33 Tens of millions of dollars is tweeting like that.
00:27:36 But perhaps it’s a worthy price to pay
00:27:39 because ultimately there’s something magical
00:27:42 about just being real and honest
00:27:45 and just going off the cuff and making the mistakes
00:27:48 and paying for them, but just being real.
00:27:50 And then moments like this,
00:27:52 that was an opportunity for Vlad to be that.
00:27:55 And it felt like he wasn’t.
00:27:57 Do you think we’ll ever find out what really went down
00:28:02 if there was something shady underneath it all?
00:28:05 Yeah, I mean, it would be sad if nothing shady happened,
00:28:09 but his presence made it shady.
00:28:12 Sometimes I feel like that would mark Zuckerberg,
00:28:15 the CEO of Facebook.
00:28:18 Sometimes I feel like, yeah,
00:28:19 there’s a lot of shitty things that Facebook is doing,
00:28:22 but sometimes I think he makes it look worse
00:28:25 by the way he presents himself about those things.
00:28:28 Like I honestly think that a large amount of people
00:28:31 at Facebook just have a huge unstable chaotic system
00:28:36 and they’re all, not all, but mass are trying to do good
00:28:40 with this chaotic system.
00:28:42 But the presentation is like,
00:28:43 it sounds like there’s a lot of back room conversations
00:28:47 that are trying to manipulate people.
00:28:49 And there’s something about the realness that Elon has
00:28:53 that it feels like CEO should have
00:28:55 and Vlad had that opportunity.
00:28:57 I think Mark Zuckerberg had that too when he was younger.
00:28:59 Younger.
00:29:00 And somebody said, you gotta be more professional, man.
00:29:02 You can’t say, you know, lol to an interview.
00:29:06 And then suddenly he became like this distant person
00:29:10 that was hot.
00:29:11 Like you’d rather have him make mistakes,
00:29:13 but be honest than be like professional
00:29:16 and never make mistakes.
00:29:18 Yeah, one of the difficult hires I think
00:29:22 is like marketing people or like PR people
00:29:25 is you have to hire people that get the fact
00:29:28 that you can say lol on an interview.
00:29:31 Or, you know, take risks as opposed to what the PR,
00:29:37 I’ve talked to quite a few big CEOs
00:29:39 and the people around them are trying to constantly
00:29:44 minimize risk of like, what if he says the wrong thing?
00:29:48 What if she says the wrong thing?
00:29:49 It’s like, what, be careful.
00:29:51 It’s constantly like, ooh, like, I don’t know.
00:29:53 And there’s this nervous energy that builds up over time
00:29:56 with larger and larger teams where the whole thing,
00:29:59 like I visited YouTube, for example.
00:30:01 Everybody I talked at YouTube, incredible engineering
00:30:05 and incredible system, but everybody’s scared.
00:30:08 Like, let’s be honest about this like madness
00:30:13 that we have going on of huge amounts of video
00:30:15 that we can’t possibly ever handle.
00:30:17 There’s a bunch of hate on YouTube.
00:30:19 There’s this chaos of comments,
00:30:21 bunch of conspiracy theories, some of which might be true.
00:30:24 And then just like this mess that we’re dealing with
00:30:27 and it’s exciting, it’s beautiful.
00:30:30 It’s a place where like democratizes education,
00:30:32 all that kind of stuff.
00:30:34 And instead they’re all like sitting in like,
00:30:36 trying to be very polite and saying like,
00:30:38 well, we just want to improve the health of our platforms.
00:30:41 Like, it’s like this discussion like,
00:30:44 all right, man, let’s just be real.
00:30:46 Let’s both advertise how amazing this fricking thing is,
00:30:50 but also to say like, we don’t know what we’re doing.
00:30:53 We have all these Nazis posting videos on YouTube.
00:30:56 We don’t know how to like handle it.
00:30:58 And just being real like that,
00:31:00 because I suppose that’s just a skill.
00:31:02 Maybe it can’t be taught, but over time,
00:31:05 the whatever the dynamics of the company is,
00:31:06 it does seem like Zuckerberg and others get worn down.
00:31:09 They just get tired.
00:31:10 Yeah.
00:31:11 They get tired of.
00:31:12 Not being real.
00:31:13 Of not being real, which is sad.
00:31:16 So let’s talk about Numerai,
00:31:17 which is an incredible company system idea, I think,
00:31:24 but good place to start.
00:31:26 What is Numerai and how does it work?
00:31:30 So Numerai is the first hedge fund
00:31:32 that gives away all of its data.
00:31:35 So this is like probably the last thing
00:31:37 a hedge fund would do, right?
00:31:39 Why would we give away a data?
00:31:40 It’s like giving away your edge.
00:31:42 But the reason we do it is because we’re looking for people
00:31:46 to model our data.
00:31:49 And the way we do it is by obfuscating the data.
00:31:52 So when you look at Numerai’s data
00:31:55 that you can download for free,
00:31:57 it just looks like a million rows
00:32:00 of numbers between zero and one.
00:32:02 And you have no idea what the columns mean,
00:32:05 but you do know that if you’re good at machine learning
00:32:09 or have done regressions before,
00:32:11 you know that I can still find patterns in this data,
00:32:14 even though I don’t know what the features mean.
00:32:17 And the data itself is a time series data.
00:32:20 And even though it’s obfuscated, anonymized,
00:32:24 what is the source data like approximately?
00:32:26 What are we talking about?
00:32:27 So we are buying data from lots of different data vendors
00:32:32 and they would also never want us to share that data.
00:32:36 So we have strict contracts with them.
00:32:38 So we only can,
00:32:41 but that’s the kind of data you could never buy yourself
00:32:43 unless you had maybe a million dollars a year
00:32:46 of budget to buy data.
00:32:48 So what’s happened with the hedge fund industry
00:32:50 is you have a lot of talented people
00:32:54 who used to be able to trade and still can trade,
00:32:59 but now they have such a data disadvantage,
00:33:02 it would never make sense for them to trade themselves.
00:33:07 But Numerai, by giving away this obfuscated data,
00:33:09 we can give them a really, really high quality data set
00:33:12 that would otherwise be very expensive.
00:33:14 And they can use whatever new machine learning technique
00:33:18 they want to find patterns in that data
00:33:22 that we can use in our hedge fund.
00:33:24 And so how much variety is there in underlying data?
00:33:27 We’re talking about,
00:33:29 I apologize if I’m using the wrong terms,
00:33:31 but one is just like the stock price.
00:33:34 The other, there’s like options and all that kind of stuff,
00:33:36 like the, what are they called, order books or whatever.
00:33:41 Is there maybe other totally unrelated
00:33:45 directly to the stock market data,
00:33:47 like natural language as well, all that kind of stuff?
00:33:51 Yeah, we were really focused on stock data
00:33:55 that’s specific to stocks.
00:33:56 So things like you can have like a,
00:33:59 every stock has like a PE ratio.
00:34:01 For some stocks, it’s not as meaningful,
00:34:03 but every stock has that.
00:34:05 Every stock has one year momentum,
00:34:07 how much they went up in the last year,
00:34:10 but those are very common factors.
00:34:12 But we try to get lots and lots of those factors
00:34:15 that we have for many, many years,
00:34:17 like 15, 20 years history.
00:34:21 And then the setup of the problem is commonly in quant
00:34:25 called like cross sectional global equity.
00:34:28 You’re not really trying to say,
00:34:29 I believe this stock will go up.
00:34:31 You’re trying to say the relative position of this stock
00:34:36 in feature space makes it not a bad buy in a portfolio.
00:34:41 So it captures some period of time
00:34:44 and you’re trying to find the patterns,
00:34:46 the dynamics captured by the data of that period of time
00:34:49 in order to make short term predictions
00:34:51 about what’s going to happen.
00:34:53 Yeah, so our predictions are also not that short.
00:34:55 We’re not really caring about things like order books
00:34:58 and tech data, not high frequency at all.
00:35:02 We’re actually holding things for quite a bit longer.
00:35:05 So our prediction time horizon is about one month.
00:35:07 We end up holding stocks
00:35:08 for maybe like three or four months.
00:35:10 So I kind of believe that’s a little bit more
00:35:12 like investing than kind of plumbing,
00:35:17 like to go long a stock that’s mispriced on one exchange
00:35:21 and short on another exchange, that’s just arbitrage.
00:35:26 But what we’re trying to do is really know something more
00:35:30 about the longer term future of the stock.
00:35:31 Yeah, so from the patterns,
00:35:33 from these like periods of time series data,
00:35:37 you’re trying to understand something fundamental
00:35:39 about the stock, not like about deep value,
00:35:43 about like it’s big in the context of the market,
00:35:46 is it underpriced, overpriced, all that kind of stuff.
00:35:48 So like, this is about investing.
00:35:50 It’s not about like, just like you said,
00:35:52 high frequency trading,
00:35:54 which I think is a fascinating open question
00:35:57 from a machine learning perspective,
00:35:58 but just to like sort of build on that.
00:36:00 So you’ve anonymized the data
00:36:02 and now you’re giving away the data.
00:36:05 And then now anyone can try to build algorithms
00:36:12 that make investing decisions on top of that data
00:36:14 or predictions on the top of that data.
00:36:16 Exactly.
00:36:17 And so that’s, what does that look like?
00:36:21 What’s the goal of that?
00:36:22 What are the underlying principles of that?
00:36:24 So the first thing is,
00:36:26 we could obviously model that data in house, right?
00:36:29 We can make an XGBoost model on the data
00:36:33 and that would be quite good too.
00:36:36 But what we’re trying to do is by opening it up
00:36:40 and letting anybody participate,
00:36:43 we can do quite a lot better than if we modeled it ourselves
00:36:47 and a lot better on the stock market
00:36:50 doesn’t need to be very much.
00:36:52 Like it really matters the difference
00:36:54 between if you can make 10 and 12%
00:36:56 in an equity market neutral hedge fund
00:36:58 because usually you’re charging 2% fees.
00:37:02 So if you can do 2% better,
00:37:05 that’s like all your fees, it’s worth it.
00:37:07 So we’re trying to make sure
00:37:09 that we always have the best possible model
00:37:11 as new machine learning libraries come out,
00:37:13 new techniques come out,
00:37:14 they get automatically synthesized.
00:37:16 Like if there’s a great paper on supervised learning,
00:37:19 someone on Numerai will figure out
00:37:21 how to use it on Numerai’s data.
00:37:23 And is there an ensemble of models going on
00:37:28 or is it more towards kind of like one or two or three
00:37:33 like best performing models?
00:37:35 So the way we decide on how to weight
00:37:37 all of the predictions together
00:37:40 is by how much the users are staking on them.
00:37:44 How much of the cryptocurrency
00:37:46 that they’re putting behind their models.
00:37:48 So they’re saying, I believe in my model.
00:37:51 You can trust me because I’m gonna put skin in the game.
00:37:55 And so we can take the stake weighted predictions
00:37:57 from all our users, add those together,
00:38:01 average those together,
00:38:02 and that’s a much better model than any one model
00:38:05 in the sum because ensembling a lot of models together
00:38:08 is kind of the key thing you need to do in investing too.
00:38:12 Yeah, so you’re putting,
00:38:14 so there’s a kind of duality from the user,
00:38:16 from the perspective of a machine learning engineer
00:38:19 where it’s both a competition, just a really interesting,
00:38:22 difficult machine learning problem,
00:38:24 and it’s a way to invest algorithmically.
00:38:29 So like, but the way to invest algorithmically
00:38:33 also is a way to put skin in the game
00:38:37 that communicates to you that the quality of the algorithm
00:38:42 and also forces you to really be serious
00:38:46 about the models that you build.
00:38:49 So it’s like, everything just works nicely together.
00:38:52 Like, I guess one way to say that
00:38:54 is the interests are aligned.
00:38:58 Okay, so it’s just like poker is not fun
00:39:02 when it’s like for very low stakes.
00:39:04 The higher the stakes,
00:39:05 the more the dynamics of the system
00:39:07 starts playing out correctly.
00:39:10 Like as a small side note,
00:39:12 is there something you can say about which kind,
00:39:16 looking at the big broad view of machine learning today
00:39:19 or AI, what kind of algorithms seem to do good
00:39:24 in these kinds of competitions at this time?
00:39:27 Is there some universal thing you can say,
00:39:29 like neural networks suck,
00:39:32 recurrent neural networks suck, transformers suck,
00:39:34 or they’re awesome, like old school,
00:39:37 sort of more basic kind of classifiers are better,
00:39:40 all that, is there some kind of conclusions
00:39:42 so far that you can say?
00:39:44 There is definitely something pretty nice about tree models,
00:39:47 like XGBoost,
00:39:50 and they just seem to work pretty nicely
00:39:53 on this type of data.
00:39:54 So out of the box,
00:39:56 if you’re trying to come a hundredth
00:39:59 in the competition, in the tournament,
00:40:01 maybe you would try to use that.
00:40:04 But what’s particularly interesting about the problem
00:40:09 that not many people understand,
00:40:11 if you’re familiar with machine learning,
00:40:14 this typically will surprise you when you model our data.
00:40:18 So one of the things that you look at in finance
00:40:23 is you don’t wanna be too exposed to any one risk.
00:40:28 Like, even if the best sector in the world
00:40:33 to invest in over the last 10 years was tech,
00:40:37 does not mean you should put all of your money into tech.
00:40:40 So if you train a model,
00:40:42 it would say, put all your money in tech, it’s super good.
00:40:45 But what you wanna do is actually be very careful
00:40:49 of how much of this exposure you have to certain features.
00:40:53 So on Numeri, what a lot of people figure out is,
00:40:58 actually, if you train a model on this kind of data,
00:41:02 you wanna somehow neutralize or minimize your exposure
00:41:05 to these certain features, which is unusual,
00:41:08 because if you did train a stoplight
00:41:11 or stop street detection on computer vision,
00:41:18 your favorite feature, let’s say you have an auto encoder
00:41:21 and it’s figuring out, okay, it’s gotta be red
00:41:23 and it’s gotta be white,
00:41:25 that’s the last thing you wanna reduce your exposure to.
00:41:30 Why would you reduce your exposure
00:41:31 to the thing that’s helping your model the most?
00:41:34 And that’s actually this counterintuitive thing
00:41:36 you have to do with machine learning on financial data.
00:41:38 So reducing your exposure
00:41:41 would help you generalize the things that are…
00:41:44 So basically, a financial data has a large amount
00:41:48 of patterns that appeared in the past
00:41:51 and also a large amount of patterns
00:41:53 that have not appeared in the past.
00:41:55 And so like in that sense,
00:41:56 you have to reduce the exposure to red lights,
00:41:59 to the color red.
00:42:02 That’s interesting, but how much of this is art
00:42:05 and how much of it is science from your perspective so far
00:42:09 in terms of as you start to climb
00:42:11 from the 100th position to the 95th in the competition?
00:42:16 Yeah, well, if you do make yourself super exposed
00:42:20 to one or two features,
00:42:23 you can have a lot of volatility
00:42:25 when you’re playing Numerai.
00:42:26 You could maybe very rapidly rise to be high
00:42:31 if you were getting lucky.
00:42:33 Yes.
00:42:34 And that’s a bit like the stock market.
00:42:36 Sure, take on massive risk exposure,
00:42:38 put all your money into one stock
00:42:41 and you might make 100%,
00:42:43 but it doesn’t in the long run work out very well.
00:42:47 And so the best users are trying to stay high
00:42:53 for as long as possible,
00:42:55 not necessarily try to be first for a little bit.
00:43:00 So me, a developer, machine learning researcher,
00:43:04 how do I, Lex Friedman, participate in this competition
00:43:07 and how do others,
00:43:09 which I’m sure there’ll be a lot of others
00:43:10 interested in participating in this competition,
00:43:12 what are, let’s see, there’s like a million questions,
00:43:15 but like first one is how do I get started?
00:43:19 Well, you can go to numero.ai,
00:43:22 sign up, download the data.
00:43:24 And on the data is pretty small.
00:43:30 In the data pack you download,
00:43:32 there’s like an example script,
00:43:33 Python script that just builds a XGBoost model
00:43:37 very quickly from the data.
00:43:40 And so in a very short time,
00:43:44 you can have an example model.
00:43:46 Is that a particular structure?
00:43:47 Like what, is this model then submitted somewhere?
00:43:50 So there needs to be some kind of structure
00:43:52 that communicates with some kind of API.
00:43:54 Like how does the whole,
00:43:56 how does your model, once you’ve built,
00:43:58 once you create a little baby Frankenstein,
00:44:01 how does it then live in the world?
00:44:02 Okay, well, we want you to keep your baby Frankenstein
00:44:05 at home and take care of it.
00:44:06 We don’t want it.
00:44:07 So you never upload your model to us.
00:44:10 You always only giving us predictions.
00:44:14 So we never see the code that wrote your model,
00:44:17 which is pretty cool,
00:44:18 that our whole hedge fund is built from models
00:44:20 where we’ve never, ever seen the code.
00:44:23 But it’s important for the users because it’s their IP,
00:44:27 why do they want to give it to us?
00:44:28 That’s brilliant.
00:44:28 So they’ve got it themselves,
00:44:31 but they can basically almost like license
00:44:34 the predictions from that model to us.
00:44:36 License the predictions, yeah.
00:44:38 So. Think about it.
00:44:40 What some users do is they set up a compute server
00:44:43 and we call it Numeric Compute.
00:44:45 It’s like a little AWS kind of image
00:44:47 and you can automate this process.
00:44:50 So we can ping you.
00:44:51 We can be like, we need more predictions now.
00:44:53 And then you send it to us.
00:44:55 Okay, cool.
00:44:56 So that’s, is that described somewhere,
00:45:00 like what the preferred is, the AWS,
00:45:02 or whether another cloud platform,
00:45:04 is there, I mean, is there sort of specific technical things
00:45:07 you want to say that comes to mind
00:45:09 that is a good path for getting started?
00:45:12 So download the data, maybe play around,
00:45:16 see if you can modify the basic algorithm provided
00:45:22 in the example.
00:45:25 And then you, what, set up a little server on the AWS
00:45:28 that then runs this model and takes pings
00:45:31 and then makes predictions.
00:45:34 And so how does your own money actually come into play
00:45:37 doing the stake of a cryptocurrency?
00:45:41 Yeah, so you don’t have to stake.
00:45:44 You can start without staking.
00:45:45 And many users might try for months
00:45:49 without staking anything at all
00:45:50 to see if their model works on the real life data, right?
00:45:54 And is not overfit.
00:45:57 But then you can get Numeraire many different ways.
00:46:01 You can buy it on, you can buy some on Coinbase.
00:46:04 You can buy some on Uniswap.
00:46:06 You can buy some on Binance.
00:46:09 So what did you say this is?
00:46:11 How do you pronounce it?
00:46:12 So this is the Numeraire cryptocurrency.
00:46:15 Yeah, NMR.
00:46:16 NMR, you just say NMR?
00:46:19 It is technically called Numeraire.
00:46:21 Numeraire, I like it.
00:46:23 Yeah, but NMR is simple.
00:46:26 NMR, Numeraire.
00:46:27 Okay, so, and you could buy it basically anywhere.
00:46:31 Yeah, so it’s a bit strange
00:46:33 because sometimes people are like,
00:46:34 is this like pay to play?
00:46:36 Right.
00:46:37 And it’s like, yeah, you need to put some money down
00:46:40 to show us you believe in your model.
00:46:42 But weirdly, we’re not selling you the,
00:46:45 like you can’t buy the cryptocurrency from us.
00:46:48 Right.
00:46:49 It’s like, it’s also, we never,
00:46:51 if you do badly, we destroy your cryptocurrency.
00:46:57 Okay, that’s not good, right?
00:46:58 You don’t want it to be destroyed.
00:47:00 But what’s good about it is it’s also not coming to us.
00:47:03 Right.
00:47:04 So it’s not like we win when you lose or something,
00:47:06 like we’re the house.
00:47:08 Like we’re definitely on the same team.
00:47:10 Yes.
00:47:11 Helping us make a hedge fund that’s never been done before.
00:47:14 Yeah, so again, interests are aligned.
00:47:15 There’s no, there’s no tension there at all,
00:47:18 which is really fascinating.
00:47:19 You’re giving away everything
00:47:21 and then the IP is owned by sort of the code.
00:47:24 You never share the code.
00:47:25 That’s fascinating.
00:47:27 So since I have you here and you said a hundred,
00:47:31 I didn’t ask out of how many, so we’ll just,
00:47:34 but if I then once you get started
00:47:37 and you find this interesting, how do you then win
00:47:43 or do well, but also how do you potentially try to win
00:47:47 if this is something you want to take on seriously
00:47:49 from the machine learning perspective,
00:47:51 not from a financial perspective?
00:47:53 Yeah, I think that first of all,
00:47:55 you would want to talk to the community.
00:47:57 People are pretty open.
00:47:58 We give out really interesting scripts and ideas
00:48:01 for things you might want to try.
00:48:03 And, but you’re also going to need a lot of compute probably.
00:48:09 And so some of the best users are, you know,
00:48:12 actually the very first time someone won on Numerai,
00:48:15 I would, I wrote them a personal email.
00:48:17 It’s like, you know, you’ve won some money.
00:48:18 We’re so excited to give you $300.
00:48:21 And then they said, I spend way more on the compute,
00:48:26 but.
00:48:26 So this is fundamentally a machine learning problem first,
00:48:29 I think is this is one of the exciting things.
00:48:31 I don’t know if we’ll, in how many ways we can approach this,
00:48:34 but really this is less about kind of no offense,
00:48:40 but like finance people, finance minded people,
00:48:43 they’re also, I’m sure great people,
00:48:45 but it feels like from the community that I’ve experienced,
00:48:49 these are people who see finance
00:48:52 as a fascinating problem space, source of data,
00:48:57 but ultimately they’re machine learning people or AI people,
00:49:01 which is a very different kind of flavor of community.
00:49:03 And I mean, I should say to that,
00:49:07 I’d love to participate in this and I will participate in this
00:49:10 and I’d love to hear from other people.
00:49:12 If you’re listening to this,
00:49:13 if you’re a machine learning person,
00:49:15 you should participate in it and tell me,
00:49:17 give me some hints how I can do well at this thing.
00:49:21 Cause this boomer, I’m not sure I still got it,
00:49:24 but cause some of it is, it’s like a Kaggle competitions.
00:49:28 Like some of it is certainly set of ideas,
00:49:33 like research ideas, like fundamental innovation,
00:49:37 but I’m sure some of it is like deeply understanding,
00:49:40 getting like an intuition about the data.
00:49:42 And then like a lot of it will be like figuring out
00:49:45 like what works, like tricks.
00:49:47 I mean, you could argue most of deep learning research
00:49:50 is just tricks on top of tricks,
00:49:51 but there’s some of it is just the art
00:49:55 of getting to know how to work in a really difficult
00:49:58 machine learning problem.
00:50:00 And I think what’s important,
00:50:02 the important difference with something
00:50:03 like a Kaggle competition,
00:50:04 where they’ll set up this kind of toy problem
00:50:08 and then there will be an out of sample test,
00:50:10 like, Hey, you did well out of sample.
00:50:12 And this is like, okay, cool.
00:50:14 But what’s cool with Numeri is the out of sample
00:50:19 is the real life stock market.
00:50:22 We don’t even know,
00:50:23 like we don’t know the onset of the problem.
00:50:25 We don’t, like you’ll have to find out live.
00:50:28 And so we’ve had users who’ve like submitted every week
00:50:31 for like four years because it’s kind of,
00:50:38 we say it’s the hardest data science problem
00:50:40 on the planet, right?
00:50:41 And it sounds maybe sounds like maybe
00:50:44 but too much for like a marketing thing,
00:50:45 but it’s the hardest because it’s the stock market.
00:50:48 It’s like literally there are like billions of dollars
00:50:51 at stake and like no one’s like letting it be inefficient
00:50:55 on purpose.
00:50:56 So if you can find something that works at Numeri,
00:50:58 you really have something that is like working
00:51:01 on the real stock market.
00:51:03 Yeah, because there’s like humans involved
00:51:05 in the stock market.
00:51:06 I mean, you could argue there might be harder data sets
00:51:09 like maybe predicting the weather,
00:51:11 all those kinds of things.
00:51:12 But the fundamental statement here is, which I like,
00:51:16 I was thinking like,
00:51:16 is this really the hardest data science problem?
00:51:19 And you start thinking about that,
00:51:21 but ultimately it also boils down to a problem
00:51:24 where the data is accessible.
00:51:26 It’s made accessible, made really easy and efficient
00:51:31 at like submitting algorithms.
00:51:33 So it’s not just, you know,
00:51:35 it’s not about the data being out there, like the weather.
00:51:38 It’s about making the data super accessible,
00:51:40 making the ability of community around it.
00:51:43 Like this is what ImageNet did.
00:51:45 Exactly.
00:51:46 Like it’s not just, there’s always images.
00:51:49 The point is you aggregate them together.
00:51:51 You give it a little title.
00:51:52 This is a community and that was one of the hardest,
00:51:56 right, for a time.
00:51:58 And most important data science problems in the world
00:52:03 because it was accessible, because it was made sort of,
00:52:08 like there was mechanisms by which like standards
00:52:12 and mechanisms by which you judge your performance,
00:52:14 all those kinds of things.
00:52:14 And Numerize actually step up from that.
00:52:17 Is there something more you can say about why
00:52:19 from your perspective it’s the hardest problem in the world?
00:52:24 I mean, you said it’s connected to the market.
00:52:26 So if you could find a pattern in the market,
00:52:29 that’s a really difficult thing to do
00:52:31 because a lot of people are trying to do it.
00:52:33 Exactly.
00:52:34 But there’s also the biggest one is
00:52:37 it’s non stationary time series.
00:52:40 We’ve tried to regularize the data
00:52:42 so you can find patterns by doing certain things
00:52:46 to the features and the target.
00:52:48 But ultimately you’re in a space where you don’t,
00:52:51 there’s no guarantees that the out of sample distributions
00:52:55 will conform to any of the training data.
00:52:59 And every single era, which we call on the website,
00:53:04 like every single era in the data,
00:53:07 which is like sort of showing you the order of the time.
00:53:12 Even the training data has the same dislocations.
00:53:16 And so, yeah, and then there’s so many things
00:53:22 that you might wanna try.
00:53:25 There’s unlimited possible number of models, right?
00:53:30 And so by having it be open,
00:53:37 we can at least search that space.
00:53:40 Zooming back out to the philosophical,
00:53:42 you said that Numerai is very much like Wall Street Bets.
00:53:46 Is there, I think it’d be interesting
00:53:51 to dig in why you think so.
00:53:52 I think you’re speaking to the distributed nature of the two
00:53:56 and the power of the people nature of the two.
00:53:59 So maybe can you speak to the similarities
00:54:02 and the differences and in which way is Numerai more powerful
00:54:06 in which way is Wall Street Bets more powerful?
00:54:09 Yeah, this is why the Wall Street Bets story
00:54:11 is so interesting to me because it’s like,
00:54:12 feels like we’re connected.
00:54:15 And looking at how,
00:54:16 just looking at the form of Wall Street Bets,
00:54:19 I was talking earlier about how,
00:54:21 how can you make credible claims?
00:54:23 You’re anonymous.
00:54:24 Okay, well, maybe you can take a screenshot.
00:54:26 Or maybe you can upvote someone.
00:54:29 Maybe you can have karma on Reddit.
00:54:31 And those kinds of things make this emerging thing possible.
00:54:35 Numerai, it didn’t work at all when we started.
00:54:40 It didn’t work at all.
00:54:41 Why?
00:54:42 People made multiple accounts.
00:54:43 They made really random models
00:54:45 and hoped they would get lucky.
00:54:47 And some of them did.
00:54:48 Yes.
00:54:49 Staking was our solution to,
00:54:53 could we make it so that we could trust?
00:54:56 We could know which model people believed in the most.
00:55:00 And we could weight models that had high stake more
00:55:04 and effectively coordinate this group of people
00:55:07 to be like, well, actually there’s no incentive
00:55:10 to creating bot accounts anymore.
00:55:12 Either I stake my accounts,
00:55:14 in which case I should believe in them
00:55:16 because I could lose my stake or I don’t.
00:55:18 And that’s a very powerful thing
00:55:20 that having a negative incentive
00:55:23 and a positive incentive can make things a lot better.
00:55:26 And staking is like this,
00:55:28 is this really nice like key thing about blockchain.
00:55:31 It’s like something special you can do
00:55:33 where they’re not even trusting us
00:55:35 with their stake in some ways.
00:55:37 They’re trusting the blockchain, right?
00:55:40 So the incentives, like you say,
00:55:42 it’s about making these perfect incentives
00:55:44 so that you can have coordination to solve one problem.
00:55:47 And nowadays I sleep easy
00:55:52 because I have less money in my own hedge fund
00:55:56 than our users are staking on their models.
00:56:00 That’s powerful.
00:56:01 In some sense, from a human psychology perspective,
00:56:04 it’s fascinating that the WallStreetBets worked at all, right?
00:56:08 That amidst that chaos, emerging behavior,
00:56:12 like behavior that made sense emerged.
00:56:15 It would be fascinating to think if numerized style staking
00:56:20 could then be transferred to places like Reddit, you know?
00:56:24 And not necessarily for financial investments,
00:56:27 but I wish sometimes people would have to stake something
00:56:34 in the comments they make on the internet.
00:56:36 Yeah.
00:56:37 That’s the problem with anonymity is like,
00:56:40 anonymity is freedom and power
00:56:42 that you don’t have to, you can speak your mind,
00:56:44 but it’s too easy to just be shitty.
00:56:47 Yeah, exactly.
00:56:49 So this, I mean, you’re making me realize
00:56:52 from a profoundly philosophical aspect, numerized staking
00:56:58 is a really clean way to solve this problem.
00:57:01 It’s a really beautiful way.
00:57:02 Of course, it only with Numerai currently works
00:57:05 for a very particular problem, right?
00:57:07 Not for human interaction on the internet,
00:57:10 but that’s fascinating.
00:57:11 Yeah, there’s nothing to stop people.
00:57:13 In fact, we’ve open sourced the code we use for staking
00:57:17 in a protocol we call Erasure.
00:57:19 And if Reddit wanted to, they could even use that code
00:57:23 to enable staking on our Wall Street pets.
00:57:29 And they’re actually researching now,
00:57:30 they’ve had some Ethereum grants
00:57:32 on how could they have more crypto stuff in there
00:57:37 in Ethereum, because wouldn’t that be interesting?
00:57:40 Like, imagine you could, instead of seeing a screenshot,
00:57:43 like, guys, I promise I will not sell my GameStop.
00:57:49 We’re just going to go huge.
00:57:50 We’re not going to sell at all.
00:57:53 And here is a smart contract, which no one in the world,
00:57:58 including me, can undo, that says,
00:58:02 I have staked millions against this claim.
00:58:08 That’s powerful.
00:58:09 And then what could you do?
00:58:11 And of course, it doesn’t have to be millions.
00:58:12 It could be just a very small amount,
00:58:14 but then just a huge number of users doing that kind of stake.
00:58:17 Exactly.
00:58:20 That could change the internet.
00:58:21 It would change, and then Wall Street.
00:58:23 It would change Wall Street.
00:58:25 They would never have been able to,
00:58:26 they would still be short squeezing one day
00:58:29 after the next, every single hedge fund collapsing.
00:58:32 If we look into the future, do you think it’s possible
00:58:35 that numerae style infrastructure,
00:58:39 where AI systems backed by humans are doing the trading,
00:58:44 is what the entirety of the stock market is,
00:58:48 or the entirety of the economy, is
00:58:50 run by basically this army of AI systems
00:58:54 with high level human supervision?
00:58:57 Yeah, the thing is that some of them could be bad actors.
00:59:02 Some of the humans?
00:59:03 No, well, these systems could be tricky.
00:59:06 So actually, I once met a hedge fund manager,
00:59:08 and this is kind of interesting.
00:59:10 He said, very famous one, and he said,
00:59:14 we can see, sometimes we can see things in the market
00:59:17 where we know we can make money, but it will mess shit up.
00:59:22 We know we can make money, but it will mess things up.
00:59:25 And we choose not to do those things.
00:59:28 And on the one hand, maybe this is like, oh, you’re
00:59:30 being super arrogant.
00:59:31 Of course you can’t do this, but maybe he can.
00:59:35 And maybe he really isn’t doing things
00:59:38 he knows he could do, but would be pretty bad.
00:59:43 Would the Reddit army have that kind of morality or concern
00:59:51 for what they’re doing?
00:59:53 Probably not, based on what we’ve seen.
00:59:55 The madness of crowds.
00:59:57 There’ll be one person that says, hey, maybe,
01:00:00 and then they get trampled over.
01:00:03 That’s the terrifying thing, actually.
01:00:07 A lot of people have written about this,
01:00:08 is somehow that little voice that’s human morality
01:00:13 gets silenced when we get into groups and start chanting.
01:00:17 And that’s terrifying.
01:00:18 But I think maybe I misunderstood.
01:00:22 I thought that you’re saying AI systems can be dangerous,
01:00:25 but you just describe how humans can be dangerous.
01:00:28 So which is safer?
01:00:30 So one thing is, so Wall Street bets these kinds of attacks.
01:00:38 It’s not possible to model, numerize data,
01:00:42 and then come up with the idea from the model,
01:00:45 let’s short squeak, just game stop.
01:00:46 It’s not even framed in that way.
01:00:49 It’s not possible to have that idea.
01:00:52 But it is possible for a bunch of humans.
01:00:54 So I think this, it’s, numera could get very powerful
01:00:59 without it being dangerous.
01:01:01 But Wall Street bets needs to get a little bit more powerful,
01:01:05 and it’ll be pretty dangerous.
01:01:08 Yeah, well, I mean, this is a good place
01:01:11 to think about numera data today and numera signals
01:01:17 and what that looks like in 10, 20, 30, 50, 100 years.
01:01:22 Right now, I guess, maybe you can correct me,
01:01:24 but the data that we’re working with is like a window.
01:01:28 It’s an anonymized, obfuscated window
01:01:32 into a particular aspect, time period of the market.
01:01:36 And you can expand that more and more and more and more,
01:01:40 potentially.
01:01:41 You can imagine in different dimensions
01:01:43 to where it encapsulates all the things that,
01:01:47 where you could include kind of human to human communication
01:01:52 that was available to buy GameStop, for example,
01:01:55 on Wall Street bets.
01:01:57 So maybe as a step back, can you speak
01:02:00 to what is numera signals and what are the different data
01:02:05 sets that are involved?
01:02:07 So with numera signals, you’re still
01:02:10 providing predictions to us, but you can do it
01:02:15 from your own data sets.
01:02:18 So numera, it’s all you have to model our data
01:02:21 to come up with predictions.
01:02:22 Numa signals is whatever data you can find out there,
01:02:26 you can turn it into a signal and give it to us.
01:02:29 So it’s a way for us to import signals
01:02:32 on data we don’t yet have.
01:02:36 And that’s why it’s particularly valuable,
01:02:38 because it’s going to be signals.
01:02:42 You’re only rewarded for signals that are
01:02:44 orthogonal to our core signal.
01:02:47 So you have to be doing something uncorrelated.
01:02:50 And so strange alternative data tends to have that property.
01:02:55 There isn’t too many other signals
01:02:57 that are correlated with what’s happening on Wall Street
01:03:02 bets.
01:03:02 That’s not going to be correlated with the price
01:03:05 to earnings ratio.
01:03:07 And we have some users, as of recently, as of a week ago,
01:03:11 there was a user that created, I think he’s in India,
01:03:14 he created a signal that is scraped from Wall Street bets.
01:03:21 And now we have that signal as one
01:03:24 of our signals in thousands that we use at Numerai.
01:03:28 And the structure of the signal is similar,
01:03:30 so it’s just numbers and time series data.
01:03:33 It’s exactly.
01:03:34 And it’s just like, you’re providing a ranking of stocks.
01:03:37 So you just say, a one means you like the stock,
01:03:41 zero means you don’t like the stock,
01:03:43 and you provide that for 5,000 stocks in the world.
01:03:46 And they somehow converted the natural language
01:03:49 that’s in the Wall Street bet.
01:03:51 Exactly.
01:03:52 And they open sourced this Colab notebook.
01:03:55 You can go and see it.
01:03:57 And so, yeah, it’s making a sentiment score.
01:04:00 And then turning it into a rank of stocks.
01:04:02 A sentiment score.
01:04:03 Yeah.
01:04:04 Like, this stock sucks, or this stock is awesome.
01:04:07 And then converting.
01:04:08 That’s fascinating.
01:04:08 Just even looking at that data would be fascinating.
01:04:11 So on the signal side, what’s the vision?
01:04:15 This long term, what do you see that becoming?
01:04:18 So we want to manage all the money in the world.
01:04:21 That’s Numerai’s mission.
01:04:23 And to get that, we need to have all the data
01:04:27 and have all of the talent.
01:04:30 Like, there’s no way, first principles,
01:04:32 if you had really good modeling and really good data
01:04:35 that you would lose, right?
01:04:37 It’s just a question of how much do you need to get really good.
01:04:41 So Numerai already has some really nice data
01:04:43 that we give out.
01:04:45 This year, we are 10xing that.
01:04:48 And I actually think we’ll 10x the amount of data
01:04:50 we have on Numerai every year for at least the next 10 years.
01:04:55 Wow.
01:04:55 So it’s going to get very big, the data we give out.
01:04:59 And signals is more data.
01:05:03 People with any other random data set
01:05:06 can turn that into a signal and give it to us.
01:05:09 And in some sense, that kind of data
01:05:10 is the edge cases, the weirdness is the,
01:05:12 so you’re focused on the bulk, the main data.
01:05:16 And then there’s just weirdness from all over the place
01:05:18 that just can enter through this back door of Numerai signals.
01:05:22 Exactly.
01:05:22 And it’s also a little bit shorter term.
01:05:26 So the signals are about a seven day time horizon.
01:05:31 And on Numerai, it’s like a 30 day.
01:05:33 So it’s often for faster situations.
01:05:38 You’ve written about a master plan.
01:05:40 And you’ve mentioned, which I love,
01:05:43 in a similar sort of style of big style thinking,
01:05:46 you would like Numerai to manage all of the world’s money.
01:05:52 So how do we get there from yesterday
01:05:56 to several years from now?
01:05:59 Like what is the plan?
01:06:02 So you’ve already started to allure to get all the data
01:06:06 and get all the talent, humans, models.
01:06:11 Exactly.
01:06:12 I mean, the important thing to note there is,
01:06:14 what would that mean?
01:06:16 And I think the biggest thing it means
01:06:18 is if there was one hedge fund, you
01:06:22 would have not so much talent wasted
01:06:26 on all the other hedge funds.
01:06:27 Like it’s super weird how the industry works.
01:06:30 It’s like one hedge fund gets a data source and hires a PhD.
01:06:34 And another hedge fund has to buy the same data source
01:06:36 and hire a PhD.
01:06:37 And suddenly, a third of American PhDs
01:06:40 are working at hedge funds.
01:06:41 And we’re not even on Mars.
01:06:43 And so in some ways, Numerai, it’s
01:06:46 all about freeing up people who work at hedge funds
01:06:49 to go work for Elon.
01:06:52 Yeah.
01:06:53 And also, the people who are working on Numerai problem,
01:06:58 it feels like a lot of the knowledge
01:07:00 there is also transferable to other domains.
01:07:02 Exactly.
01:07:03 One of our top users, he works at NASA Jet Propulsion Lab.
01:07:08 And he’s amazing.
01:07:10 I went to go visit him there.
01:07:11 And he’s got Numerai posters.
01:07:13 And it looks like the movies.
01:07:16 It looks like Apollo 11 or whatever.
01:07:19 Yeah, the point is he didn’t quit his job to join full time.
01:07:26 He’s working on getting us to Jupiter’s moon.
01:07:29 That’s his mission, the Europa Klippa mission.
01:07:31 Actually, literally what you’re saying.
01:07:33 Literally.
01:07:34 He’s smart enough that we really want his intelligence
01:07:37 to reach the stock market.
01:07:38 Because the stock market’s a good thing.
01:07:39 Hedge funds are a good thing.
01:07:40 All kinds of hedge funds, especially.
01:07:43 But we don’t want him to quit his job.
01:07:45 So he can just do Numerai on the weekends.
01:07:47 And that’s what he does.
01:07:47 He just made a model and it just automatically submits to us.
01:07:50 And he’s like one of our best users.
01:07:53 You mentioned briefly that stock markets are good.
01:07:57 From my sort of outsider perspective, is there a sense,
01:08:01 do you think trading stocks is closer to gambling?
01:08:06 Or is it closer to investing?
01:08:09 Sometimes it feels like it’s gambling as opposed to betting
01:08:14 on companies that succeed.
01:08:15 And this is maybe connected to our discussion of shorting
01:08:17 in general, but from your sense, the way you think about it,
01:08:21 is it fundamentally still investing?
01:08:23 I do think, I mean, it’s a good question.
01:08:29 I’ve also seen lately people say, this is like speculation.
01:08:33 Is there too much speculation in the market?
01:08:35 And it’s like, but all the trades are speculative.
01:08:38 All the trades have a horizon.
01:08:40 People want them to work.
01:08:44 So I would say that there’s certainly
01:08:48 a lot of aspects of gambling math that applies to investing.
01:08:54 Like one thing you don’t do in gambling
01:08:57 is put all your money in one bet.
01:09:00 You have bankroll management, and it’s a key part of it.
01:09:04 And small alterations to your bankroll management
01:09:07 might be better than improvements to your skill.
01:09:10 And then there are things we care about in our fund.
01:09:13 Like we want to make a lot of independent bets.
01:09:16 We talk about it, like we want to make
01:09:18 a lot of independent bets, because that’s
01:09:20 going to be a higher sharp than if you have a lot of bets that
01:09:23 depend on each other, like all in one sector.
01:09:27 But yeah, I mean, the point is that you
01:09:31 want the prices of the stocks to be reflective of their value.
01:09:39 Of the underlying value of the company.
01:09:40 Yeah, you shouldn’t have there be like a hedge fund that’s
01:09:45 able to say, well, I’ve looked at some data,
01:09:48 and all of this stuff’s super mispriced.
01:09:51 That’s super bad for society if it looks like that to someone.
01:09:55 I guess the underlying question then
01:09:57 is, do you see that the market often drifts away
01:10:01 from the underlying value of companies,
01:10:04 and it becomes a game in itself?
01:10:06 Like with these, whatever they’re called,
01:10:09 like derivatives, like the options, and shorting,
01:10:17 and all that kind of stuff.
01:10:18 It’s like layers of game on top of the actual,
01:10:22 like what you said, which is like the basic thing
01:10:25 that the Wall Street Bets was doing,
01:10:26 which is like just buying stocks.
01:10:28 Yeah.
01:10:29 There are a lot of games that people
01:10:31 play that are in the derivatives market.
01:10:36 And I think a lot of the stuff people dislike when they look
01:10:40 at the history of what’s happened,
01:10:42 they hate like credit default swaps,
01:10:45 or collateralized debt obligations.
01:10:48 Like these are the enemies of 2008.
01:10:52 And then the long term capital management thing,
01:10:54 it was like they had 30 times leverage, or something.
01:11:00 You could just go to a gas station
01:11:03 and ask anybody at the gas station,
01:11:05 is it a good idea to have 30 times leverage?
01:11:08 And they just say no.
01:11:09 It’s like common sense just like went out the window.
01:11:12 So yeah, I don’t respect long term capital management.
01:11:20 OK.
01:11:21 But Numerai doesn’t actually use any derivatives
01:11:24 unless you call shorting derivative.
01:11:26 We just we do put money into companies.
01:11:29 And that does help the companies we’re investing in.
01:11:32 It’s just in little ways.
01:11:34 We really did buy Tesla.
01:11:36 And it did.
01:11:37 And we played some role in its success.
01:11:44 Super small, make no mistake.
01:11:46 But still, I think that’s important.
01:11:48 Can I ask you a pothead question,
01:11:51 which is what is money, man?
01:11:55 So if we just kind of zoom out and look at,
01:11:59 because I’d love to talk to you about cryptocurrency, which
01:12:02 perhaps could be the future of money.
01:12:04 In general, how do you think about money?
01:12:07 You said Numerai, the vision, the goal
01:12:10 is to run, to manage the world’s money.
01:12:15 What is money in your view?
01:12:19 I don’t have a good answer to that.
01:12:22 But it’s definitely in my personal life,
01:12:25 it’s become more and more warped.
01:12:29 And you start to care about the real thing,
01:12:33 like what’s really going on here.
01:12:37 Elon talks about things like this,
01:12:39 like what is a company, really?
01:12:40 It’s a bunch of people who show up to work together
01:12:43 and they solve a problem.
01:12:45 And there might not be a stock out there
01:12:47 that’s trading that represents what they’re doing,
01:12:49 but it’s not the real thing.
01:12:52 And being involved in crypto, I put
01:12:57 in crowdsale of Ethereum and all these other things
01:13:03 and different crypto hedge funds and things
01:13:06 that I’ve invested in.
01:13:07 And it’s just kind of like, it feels
01:13:09 like how I used to think about money stuff
01:13:13 is just totally warped.
01:13:15 Because you stop caring about the price
01:13:23 and you care about the product.
01:13:26 So by the product, you mean the different mechanisms
01:13:29 that money is exchanged.
01:13:30 I mean, money is ultimately a kind of a little,
01:13:33 one is a store of wealth, but it’s also
01:13:36 a mechanism of exchanging wealth.
01:13:38 But what wealth means becomes a totally different thing,
01:13:42 especially with cryptocurrency, to where it’s almost
01:13:45 like these little contracts, these little agreements,
01:13:48 these transactions between human beings
01:13:50 that represent something that’s bigger than just cash being
01:13:56 exchanged at 7.11, it feels like.
01:13:58 Yeah.
01:13:58 Maybe I’ll answer what is finance.
01:14:03 It’s what are you doing when you have the ability
01:14:06 to take out a loan?
01:14:08 You can bring a whole new future into being with finance.
01:14:15 If you couldn’t get a student loan to get a college degree,
01:14:20 you couldn’t get a college degree
01:14:22 if you didn’t have the money.
01:14:23 But now, weirdly, you can get it with, and all you have
01:14:29 is this loan, which is like, so now you
01:14:32 can bring a different future into the world.
01:14:34 And that’s how when I was saying earlier about if you rerun
01:14:36 American economic history without these things,
01:14:40 like you’re not allowed to take out loans,
01:14:42 you’re not allowed to have derivatives,
01:14:44 you’re not allowed to have money,
01:14:47 it just doesn’t really work.
01:14:49 And it’s a really magic thing how much
01:14:51 you can do with finance by bringing the future forward.
01:14:56 Finance is empowering.
01:14:58 We sometimes forget this, but it enables innovation.
01:15:01 It enables big risk takers and bold builders that ultimately
01:15:04 make this world better.
01:15:06 You said you were early in on cryptocurrency.
01:15:09 Can you give your high level overview
01:15:12 of just your thoughts about the past, present, and future
01:15:15 of cryptocurrency?
01:15:17 Yeah, so my friends told me about Bitcoin,
01:15:19 and I was interested in equities a lot.
01:15:23 And I was like, well, it has no net present value.
01:15:27 It has no future cash flows.
01:15:29 Bitcoin pays no dividends.
01:15:33 So I really couldn’t get my head around it,
01:15:36 that this could be valuable.
01:15:39 And then I didn’t feel like I was early in cryptocurrency,
01:15:44 in fact, because it was like 2014.
01:15:46 It felt like a long time after Bitcoin.
01:15:50 But then I really liked some of the things
01:15:52 that Ethereum was doing.
01:15:54 It seemed like a super visionary thing.
01:15:57 I was reading something that was just
01:15:59 going to change the world when I was reading the white paper.
01:16:03 And I liked the different constructs
01:16:06 you could have inside of Ethereum
01:16:08 that you couldn’t have on Bitcoin.
01:16:10 Like smart contracts and all that kind of stuff?
01:16:11 Exactly, yeah.
01:16:12 And even spoke about different constructions you could have.
01:16:18 Yeah, that’s a cool dance between Bitcoin and Ethereum.
01:16:21 It’s in the space of ideas.
01:16:23 It feels so young.
01:16:25 Like, I wonder what cryptocurrencies will look like
01:16:29 in the future.
01:16:30 If Bitcoin or Ethereum 2.0 or some version
01:16:33 will stick around or any of those,
01:16:35 who’s going to win out?
01:16:37 Or if there’s even a concept of winning out at all?
01:16:39 Is there a cryptocurrency that you especially
01:16:44 find interesting that technically, financially,
01:16:48 philosophically, you think is something
01:16:52 you’re keeping your eye on?
01:16:54 Well, I don’t really.
01:16:55 I’m not looking to invest in cryptocurrencies anymore.
01:16:59 But I mean, and many are almost identical.
01:17:05 I mean, there wasn’t too much difference
01:17:09 between even Ethereum and Bitcoin in some ways, right?
01:17:13 But there are some that I like the privacy ones.
01:17:15 I mean, I like Zcash for its coolness.
01:17:19 It’s actually a different kind of invention
01:17:23 compared to some of the other things.
01:17:25 OK, can you speak just briefly to privacy?
01:17:29 Is there some mechanism of preserving
01:17:31 some privacy of the universe?
01:17:32 So I guess everything is public.
01:17:34 Yeah.
01:17:35 Is that the problem?
01:17:36 Yeah, none of the transactions are private.
01:17:40 And so I have some numeraire.
01:17:46 And you can just see it.
01:17:48 In fact, you can go to a website and it says like,
01:17:50 you can go to like, etherscan.
01:17:51 And it’ll say like, numeraire founder.
01:17:54 And I’m like, how the hell do you guys know this?
01:17:57 So they can reverse the near, whatever that’s called.
01:18:00 Yeah, and so they can see me move it, too.
01:18:02 They can see me.
01:18:02 Oh, why is he moving it?
01:18:04 Yeah.
01:18:06 So but yeah, Zcash.
01:18:10 Also, when you can make private transactions,
01:18:12 you can also play different games.
01:18:14 Yes.
01:18:15 And it’s unclear.
01:18:17 It’s like what’s quite cool about Zcash
01:18:19 is I wonder what games are being played there.
01:18:21 No one will know.
01:18:23 So from a deeply analytical perspective,
01:18:27 can you describe why Dogecoin is going to win?
01:18:31 Which it surely will.
01:18:32 Like it very likely will take over the world.
01:18:34 And once we expand out into the universe,
01:18:37 we’ll take over the universe.
01:18:40 Or on a more serious note, like what
01:18:42 are your thoughts on the recent success of Dogecoin
01:18:45 where you’ve spoken to sort of the meme stocks,
01:18:49 the memetics of the whole thing, that it feels like the joke can
01:18:55 become the reality.
01:18:58 Like the meme, the joke has power in this world.
01:19:02 Yeah.
01:19:02 It’s fascinating.
01:19:04 Exactly.
01:19:05 It’s like why is it correlated with Elon tweeting about it?
01:19:12 It’s not just Elon alone tweeting, right?
01:19:15 It’s like Elon tweeting and that becomes
01:19:17 a catalyst for everybody on the internet kind of like spreading
01:19:22 the joke, right?
01:19:22 Exactly.
01:19:23 The joke of it.
01:19:24 So it’s the initial spark of the fire for Wall Street
01:19:28 bets type of situation.
01:19:30 Yeah.
01:19:31 And that’s fascinating because jokes
01:19:33 seem to spread faster than other mechanisms.
01:19:37 Like funny shit is very effective at captivating
01:19:43 the discourse on the internet.
01:19:47 Yeah, and I think you can have, like I like the one meme,
01:19:51 like Doge, I haven’t heard that name in a long time.
01:19:57 Like I think back to that meme often.
01:20:00 That’s like funny.
01:20:01 And every time I think back to it,
01:20:04 there’s a little probability that I might buy it, right?
01:20:08 And so imagine you have millions of people who have had
01:20:12 all these great jokes, told them,
01:20:14 and every now and then they reminisce,
01:20:16 oh, that was really funny.
01:20:17 And then they’re like, let me buy some.
01:20:21 Wouldn’t that be interesting if we travel in time
01:20:25 like multiple centuries where the entirety
01:20:28 of the communication of the human species is like humor?
01:20:33 Like it’s all just jokes.
01:20:35 Like we’re high on probably some really advanced drugs
01:20:39 and we’re all just laughing nonstop.
01:20:42 It’s a weird like dystopian future of just humor.
01:20:47 Elon has made me realize how like good it feels
01:20:53 to just not take shit seriously every once in a while
01:20:55 and just relieve like the pressure of the world.
01:20:58 At the same time, the reason I don’t always like
01:21:03 when people finish their sentences with lol
01:21:06 is like when you don’t take anything seriously.
01:21:11 When everything becomes a joke,
01:21:13 then it feels like that way of thinking
01:21:20 feels like it will destroy the world.
01:21:22 It’s like, I often think of like,
01:21:24 will memes save the world or destroy it?
01:21:25 Because I think both are possible directions.
01:21:28 Yeah, I think this is a big problem.
01:21:30 I mean, I always felt that about America,
01:21:33 a lot of people are telling jokes kind of all the time
01:21:36 and they’re kind of good at it.
01:21:37 And you take someone aside, an American,
01:21:42 you’re like, I really wanna have a sincere conversation.
01:21:44 It’s like hard to even keep a straight face
01:21:46 because everything is so, there’s so much levity.
01:21:50 So it’s complicated.
01:21:51 I like how sincere actually like your Twitter can be.
01:21:54 You’re like, I am in love with the world today.
01:21:57 I get so much shit for it, it’s hilarious.
01:22:00 I’m never gonna stop because I realize like,
01:22:03 you have to be able to sometimes just be real
01:22:05 and be positive and just be, say the cliche things,
01:22:09 which ultimately those things actually capture
01:22:11 some fundamental truths about life.
01:22:15 But it’s a dance.
01:22:16 And I think Elon does a good job of that
01:22:20 from an engineering perspective of being able to joke,
01:22:22 but everyone’s mostly to pull back and be like,
01:22:26 here’s real problems, let’s solve them and so on.
01:22:29 And then be able to jump back to a joke.
01:22:31 So it’s ultimately, I think, I guess a skill that we
01:22:36 have to learn.
01:22:39 But I guess your advice is to invest everything
01:22:41 anyone listening owns into Dogecoin.
01:22:44 That’s what I heard from this interaction.
01:22:46 Yeah, no, exactly.
01:22:46 Yeah, our hedge fund is unavailable.
01:22:50 Just go straight to Dogecoin.
01:22:52 You’re running a successful company.
01:22:55 It’s just interesting because my mind has been in that space
01:22:58 of potentially just being one of the millions of other
01:23:01 entrepreneurs.
01:23:03 What’s your advice on how to build a successful startup,
01:23:08 how to build a successful company?
01:23:10 I think that one thing I do like,
01:23:13 and it might be a particular thing about America,
01:23:16 but there is something about playing,
01:23:20 tell people what you really want to happen in the world.
01:23:23 Don’t stop.
01:23:25 It’s not gonna make it,
01:23:28 like if you’re asking someone to invest in your company,
01:23:31 don’t say, I think maybe one day we might make
01:23:33 a million dollars.
01:23:35 When you actually believe something else,
01:23:38 you actually believe you’re actually more optimistic,
01:23:41 but you’re toning down your optimism because you want
01:23:45 to appear like low risk.
01:23:50 But actually it’s super high risk if your company
01:23:53 becomes mediocre because no one wants to work
01:23:56 in a mediocre company.
01:23:57 No one wants to invest in a mediocre company.
01:24:00 So you should play the real game.
01:24:02 And obviously this doesn’t apply to all businesses,
01:24:04 but if you play a venture backed startup kind of game,
01:24:07 like play for keeps, play to win, go big.
01:24:11 And it’s very hard to do that.
01:24:13 I’ve always feel like, yeah,
01:24:18 you can start narrowing your focus because 10 people
01:24:22 are telling you, you gotta care about this boring thing
01:24:26 that won’t matter five years from now.
01:24:28 And you should push back and do the real,
01:24:31 play the real game.
01:24:32 So be bold.
01:24:33 So both, I mean, there’s an interesting duality there.
01:24:37 So there’s the way you speak to other people
01:24:41 about like your plans and what you are like privately
01:24:45 just in your own mind.
01:24:48 And maybe it’s connected with what you were saying about,
01:24:50 yeah, sincerity as well.
01:24:51 Like if you appear to be sincerely optimistic
01:24:55 about something that’s big or crazy,
01:24:59 it’s putting yourself up to be kind of like
01:25:01 ridiculed or something.
01:25:03 And so if you say, my mission is to, yeah, go to Mars,
01:25:08 it’s just so bonkers that it’s hard to say.
01:25:12 It is, but one powerful thing, just like you said,
01:25:17 is if you say it and you believe it,
01:25:20 then actually amazing people come and work with you.
01:25:25 Exactly.
01:25:26 It’s not just skill, but the dreams.
01:25:28 There’s something about optimism that,
01:25:31 like that fire that you have when you’re optimistic
01:25:33 of actually having the hope of building
01:25:36 something totally cool, something totally new,
01:25:38 that when those people get in a room together,
01:25:41 like they can actually do it.
01:25:42 Yeah.
01:25:43 Yeah, there’s, yeah, that’s,
01:25:47 and also makes life really fun when you’re in that room.
01:25:50 So all of that together, ultimately,
01:25:55 I don’t know, that’s what makes this crazy ride
01:25:57 of a startup really look fun.
01:25:59 And Elon is an example of a person who succeeded at that.
01:26:02 There’s not many other inspiring figures, which is sad.
01:26:06 I used to be at Google and there’s something that happens
01:26:11 that sometimes when the company grows
01:26:13 bigger and bigger and bigger,
01:26:14 where that kind of ambition kind of quiets down a little bit.
01:26:18 Yeah.
01:26:19 Google had this ambition, still does,
01:26:21 of making the world’s information accessible to everyone.
01:26:24 And I remember, I don’t know, that’s beautiful.
01:26:28 I still love that dream of that they used to scan books,
01:26:33 but just in every way possible
01:26:34 make the world’s information accessible.
01:26:37 Same with Wikipedia.
01:26:38 Every time I open up Wikipedia,
01:26:40 I’m just awe inspired by how awesome humans are, man.
01:26:47 And creating this together,
01:26:48 I don’t know what the meanings are over there,
01:26:50 but it’s just beautiful.
01:26:52 Like what they’ve created is incredible.
01:26:55 And I’d love to be able to be part of something like that.
01:26:58 And you’re right, for that, you have to be bold.
01:27:01 And there’s, and strange to me also,
01:27:03 I think you’re right that there’s
01:27:04 how many boring companies there are.
01:27:06 Something I always talk about, especially in FinTech,
01:27:08 it’s like, why am I excited about, this is so lame.
01:27:13 Like what is, this isn’t even important.
01:27:16 Even if you succeed, this is gonna be like terrible.
01:27:19 This is not good.
01:27:21 And it’s just strange how people can kind of
01:27:23 get fake enthusiastic about like boring ideas
01:27:27 when there’s so many bigger ideas that,
01:27:32 yeah, I mean, you read these things,
01:27:33 like this company raises money,
01:27:35 and it’s just like, that’s a lot of money
01:27:36 for the worst idea I’ve ever heard.
01:27:38 Some ideas are really big.
01:27:41 So like I worked on autonomous vehicles quite a bit.
01:27:44 And there’s so many ways in which you can present
01:27:48 that idea to yourself, to the team you work with,
01:27:50 to just, yeah, like to yourself when you’re quietly
01:27:53 looking in the mirror in the morning,
01:27:55 that’s really boring or really exciting.
01:27:58 Like if you’re really ambitious with autonomous vehicles,
01:28:01 it changes the nature of like human robot interaction,
01:28:06 it changes the nature of how we move.
01:28:08 Forget money, forget all that stuff.
01:28:09 It changes like everything about robotics and AI,
01:28:13 machine learning, it changes everything about manufacture.
01:28:16 I mean, cars, transportation is so fundamentally connected
01:28:20 to cars, and if that changes,
01:28:22 it’s changing the fabric of society,
01:28:24 of movies, of everything.
01:28:27 And if you go bold and take risks
01:28:29 and be willing to go bankrupt with your company,
01:28:33 as opposed to cautiously, you can really change the world.
01:28:37 And it’s so sad for me to see all these autonomous companies,
01:28:40 autonomous vehicle companies,
01:28:41 they’re like really more focused about fundraising
01:28:45 and kind of like smoke and mirrors,
01:28:46 they’re really afraid,
01:28:48 the entirety of their marketing is grounded in fear
01:28:51 and presenting enough smoke to where they keep raising funds
01:28:54 so they can cautiously use technology of a previous decade
01:28:59 or previous two decades to kind of test vehicles
01:29:02 here and there, as opposed to do crazy things
01:29:04 and bold and go huge at scale, do huge data collection.
01:29:10 And yeah, so that’s just an example.
01:29:12 Like the idea can be big,
01:29:14 but if you don’t allow yourself to take that idea
01:29:17 and think really big with it,
01:29:20 then you’re not gonna make anything happen.
01:29:22 Yeah, you’re absolutely right in that.
01:29:24 So you’ve been connected in your work
01:29:28 with a bunch of amazing people.
01:29:31 How much interaction do you have with investors,
01:29:34 that whole process is an entire mystery to me.
01:29:36 Is there some people that just have influence
01:29:38 on the trajectory of your thinking completely,
01:29:42 or is it just this collective energy behind the company?
01:29:46 Yeah, I mean, I came here and I was amazed how,
01:29:52 yeah, people would, I was only here for a few months
01:29:54 and I met some incredible investors
01:29:57 and I’d almost run out of money.
01:29:59 And once they invested, I was like,
01:30:04 I am not gonna let you down.
01:30:06 And I was like, okay, I’m gonna send them
01:30:08 like an email update every like three minutes.
01:30:11 And then they don’t care at all.
01:30:14 So they kind of wanna, I don’t know, like,
01:30:15 so for some, I like it when it’s just like,
01:30:18 they’re always available to talk,
01:30:20 but a lot of building a business,
01:30:23 especially a high tech business,
01:30:26 there’s a little for them to add, right?
01:30:29 There’s little for them to add on product.
01:30:31 There’s a lot for them to add on like business development.
01:30:34 And if we are doing product research,
01:30:36 which is for us research into the market,
01:30:39 research into how to make a great hedge fund,
01:30:41 and we do that for years,
01:30:44 there’s not much to tell the investors.
01:30:47 So that basically is like, I believe in you.
01:30:49 There’s something, I like the cut of your jib.
01:30:52 There’s something in your idea, in your ambition,
01:30:55 in your plans that I like.
01:30:57 And it’s almost like a pat on the back.
01:30:59 It’s like, go get them kid.
01:31:01 Yeah, it is a bit like that.
01:31:02 And that’s cool.
01:31:04 That’s a good way to do it.
01:31:05 I’m glad they do it that way.
01:31:07 Like the one meeting I had,
01:31:08 which was like really good with this
01:31:10 was meeting Howard Morgan,
01:31:13 who’s actually a co founder of Renaissance Technologies
01:31:16 in the like 1980s and worked with Jim Simons.
01:31:21 And he was in the room
01:31:25 and I was meeting some other guy and he was in the room
01:31:28 and I was explaining how quantitative finance works.
01:31:33 I was like, so they use mathematical models.
01:31:36 And then he was like, yeah, I started Renaissance.
01:31:40 I know a bit about this.
01:31:43 And then I was like, oh my God.
01:31:46 So yeah, but then, and then I think he kind of said, well,
01:31:50 yeah, he said, well, cause I was talking,
01:31:52 he was working at first round capital as a partner
01:31:55 and they kind of said, they didn’t want to invest.
01:31:59 And then I wrote a blog post describing the idea
01:32:01 and I was like, I really think you guys should invest.
01:32:03 And then they end up.
01:32:04 Oh, interesting.
01:32:05 You convinced them on that.
01:32:06 That must be good.
01:32:07 Yeah, cause they’re like,
01:32:08 we don’t really invest in hedge funds.
01:32:09 And I was like, you don’t see like what I’m doing.
01:32:11 This is so a tech company, not a hedge fund, right?
01:32:14 Yeah, and Numerai is brilliant.
01:32:15 It’s, when it caught my eye,
01:32:18 there’s something special there.
01:32:19 So I really do hope you succeed in the,
01:32:22 obviously it’s a risky thing you’re taking on,
01:32:24 the ambition of it, the size of it,
01:32:26 but I do hope you succeed.
01:32:28 You mentioned Jim Simons.
01:32:30 He comes up in another world of mine really often on the,
01:32:34 he’s just a brilliant guy on the mathematics side
01:32:37 as a mathematician,
01:32:38 but he’s also brilliant finance hedge fund manager guy.
01:32:44 Have you gotten a chance to interact with him at all?
01:32:47 Have you learned anything from him on the math,
01:32:51 on the finance, on the philosophy, life side, things?
01:32:54 I’ve played poker with him.
01:32:56 It was pretty cool.
01:32:57 It was like, actually in the show, Billions,
01:32:59 they kind of do a little thing about this poker tournament
01:33:02 thing with all the hedge fund managers.
01:33:04 And that’s real life thing.
01:33:06 And they have a lot of like world series of bracelet,
01:33:09 world series poker bracelets holders,
01:33:11 but it’s kind of Jim’s thing.
01:33:13 And I met him there and yeah, it was kind of brief,
01:33:19 but I was just like, he’s like, oh, how do you,
01:33:21 why are you here?
01:33:22 And I was like, oh, Howard sent me, you know,
01:33:23 he’s like, go play this tournament,
01:33:25 meet some of the other players.
01:33:27 And then…
01:33:29 Was it Texas Holdem?
01:33:30 Yeah, Texas Holdem tournament, yeah.
01:33:32 Do you play poker yourself or was it?
01:33:33 Yeah, I do.
01:33:34 I mean, it was crazy.
01:33:36 On my right was the CEO,
01:33:39 who’s the current CEO of Renaissance, Peter Brown.
01:33:42 And Peter Muller, who’s a hedge fund manager at PDT.
01:33:49 And yeah, I mean, it was just like,
01:33:50 and then, you know, just everyone.
01:33:51 And then all these bracelet world series,
01:33:53 like people that I know from like TV.
01:33:56 And Robert Mercer, who’s fucking crazy.
01:34:01 Who’s that?
01:34:02 He’s the guy who donated the most money to Trump.
01:34:08 And he’s just like…
01:34:09 It’s a lot of personality.
01:34:10 Character, yeah, geez, it’s crazy.
01:34:13 So it’s quite cool how, yeah, like the, it was really fun.
01:34:17 And then I managed to knock out Peter Muller.
01:34:19 I have a, I got a little trophy for knocking him out
01:34:22 because he was a previous champion.
01:34:24 In fact, I think he’s won the most.
01:34:25 I think he’s won three times.
01:34:27 Super smart guy.
01:34:30 But I will say Jim outlasted me in the tournament.
01:34:35 And they’re all extremely good at poker,
01:34:41 but they’re also, so it was a $10,000 buy in.
01:34:45 And I was like, this is kind of expensive,
01:34:50 but it all goes to charity, Jim’s math charity.
01:34:54 But then the way they play, they have like rebis
01:34:58 and like they all do a shit ton of rebis
01:35:01 because it’s for charity.
01:35:02 So immediately they’re like going all in
01:35:06 and I’m like, man, like, so I ended up adding more as well.
01:35:12 So it’s like you couldn’t play at all without doing that.
01:35:15 Yeah, the stakes are high.
01:35:16 But you’re connected to a lot of these folks
01:35:18 that are kind of titans of just of economics
01:35:25 and tech in general.
01:35:27 Do you feel a burden from this?
01:35:28 You’re a young guy.
01:35:30 I did feel a bit out of place there.
01:35:33 The company was quite new
01:35:35 and they also don’t speak about things, right?
01:35:39 So it’s not like going to meet a famous rocket engineer
01:35:44 who will tell you how to make a rocket.
01:35:46 They do not want to tell you anything
01:35:48 about how to make a hedge fund.
01:35:49 It’s like all secretive and that part I didn’t like.
01:35:54 And they were also kind of making fun of me a little bit.
01:35:57 Like they would say, like they’d call me like,
01:36:00 I don’t know, the Bitcoin kid or.
01:36:02 Yeah, yeah, yeah.
01:36:02 And then they would say, even things like,
01:36:05 member Peter, yeah, said to me something like,
01:36:08 I don’t think AI is gonna have a big role in finance.
01:36:12 And I was like, hearing this from the CEO of Renaissance
01:36:16 was like weird to hear because I was like,
01:36:17 of course it will.
01:36:19 And he’s like, but he can see,
01:36:21 I can see it having a really big impact
01:36:23 on things like self driving cars.
01:36:25 But finance, it’s too noisy and whatever.
01:36:28 And so I don’t think it’s like the perfect application.
01:36:30 And I was like, that was interesting to hear
01:36:32 because it’s like, and I think it was that same day
01:36:35 that Libra, I think it is, the poker playing AI
01:36:40 started to beat like the human.
01:36:42 So it was kind of funny hearing them like say,
01:36:44 oh, I’m not sure AI could ever attack that problem.
01:36:47 And then that very day it’s attacking the problem
01:36:49 of the game we’re playing.
01:36:51 Well, there’s a kind of a magic to somebody
01:36:55 who’s exceptionally successful looking at you,
01:36:59 giving you respect, but also saying that what you’re doing
01:37:03 is not going to succeed in a sense.
01:37:06 Like they’re not really saying it,
01:37:08 but I tend to believe from my interactions with people
01:37:11 that it’s a kind of prod to say like, prove me wrong.
01:37:14 Yeah.
01:37:15 That’s ultimately, that’s how those guys talk.
01:37:18 They see good talent and they’re like.
01:37:20 Yeah.
01:37:21 And I think they’re also saying
01:37:22 it’s not gonna succeed quickly in some way.
01:37:25 They’re like, this is gonna take a long time
01:37:29 and maybe that’s good to know.
01:37:32 And certainly AI in trading,
01:37:36 that’s one of the most philosophically interesting questions
01:37:42 about artificial intelligence and the nature of money.
01:37:45 Because it’s like, how much can you extract
01:37:48 in terms of patterns from all of these millions
01:37:52 of humans interacting using this methodology of money?
01:37:57 It’s like one of the open questions
01:37:58 in the artificial intelligence.
01:37:59 In that sense, you converting into a data set
01:38:02 is one of like the biggest gifts to the research community,
01:38:07 to the whole, anyone who loves data science and AI,
01:38:11 this is kind of fascinating.
01:38:14 I’d love to see where this goes actually.
01:38:15 I think I say sometimes long before AGI destroys the world,
01:38:19 a narrow intelligence will win all the money
01:38:21 in the stock market.
01:38:23 Way, like just a narrow AI.
01:38:25 Yeah.
01:38:26 And I don’t know if I’m gonna be the one who invents that.
01:38:29 So I’m building Numerai to make sure
01:38:31 that that narrow AI uses our data.
01:38:35 So you’re giving a platform
01:38:37 to where millions of people can participate
01:38:38 and do build that narrow AI themselves.
01:38:43 People love it when I ask this kind of question
01:38:45 about books, about ideas and philosophers and so on.
01:38:50 I was wondering if you had books or ideas,
01:38:56 philosophers, thinkers that had an influence
01:38:59 on your life when you were growing up
01:39:01 or just today that you would recommend
01:39:04 that people check out blog posts, podcasts, videos,
01:39:08 all that kind of stuff.
01:39:09 Is there something that just kind of had an impact on you
01:39:12 that you couldn’t recommend?
01:39:13 A super kind of obvious one that I really,
01:39:19 I was reading Zero to One while coming up with Numerai.
01:39:22 It was like halfway through the book.
01:39:24 And I really do like a lot of the ideas there.
01:39:27 And it’s also about kind of thinking big
01:39:29 and also it’s like peculiar little book.
01:39:34 It’s like why, like there’s a little picture
01:39:36 of the hipster versus Unabomber.
01:39:38 And it’s a weird little book.
01:39:40 So I like, there’s kind of like some depth there.
01:39:42 In terms of a book on a, if you’re thinking
01:39:44 of doing a startup, that’s a good book.
01:39:47 A book I like a lot is maybe my favorite book
01:39:52 is David Deutsch’s Beginning of Infinity.
01:39:56 I just found that so optimistic.
01:40:00 It puts you, everything you read in science,
01:40:03 it like makes the world feel like kind of colder
01:40:06 because like it’s like we’re just coming from evolution
01:40:10 and coming from nothing should be this way or whatever.
01:40:15 And humans are not very powerful.
01:40:16 We’re just like scum on the earth.
01:40:19 And the way David Deutsch sees things
01:40:20 and argues, he argues them with the same rigor
01:40:23 that the cynics often use
01:40:26 and then has a much better conclusion.
01:40:29 That’s some of the statements of things like,
01:40:33 anything that doesn’t violate the laws
01:40:35 of physics can be solved.
01:40:39 So ultimately arriving at a hopeful,
01:40:41 like a hopeful path forward.
01:40:42 Yeah, without being like a hippie.
01:40:45 You’ve mentioned kind of advice for startups.
01:40:47 Is there, in general, whether you do a startup or not,
01:40:50 do you have advice for young people today?
01:40:52 You’re like an example of somebody
01:40:54 who’s paved their own path
01:40:56 and were, I would say exceptionally successful.
01:40:59 Is there advice, somebody who’s like 20 today, 18,
01:41:02 undergrad or thinking about going to college,
01:41:05 or in college and so on that you would give them?
01:41:09 I think I often tell young people don’t start companies.
01:41:13 Is it not, don’t start a company
01:41:16 unless you’re prepared to make it your life’s work.
01:41:19 Like that’s a really good way of putting it.
01:41:22 And a lot of people think, well, this semester
01:41:25 I’m gonna take a semester off.
01:41:27 And in that one semester,
01:41:28 I’m gonna start a company and sell it or whatever.
01:41:31 And it’s just like, what are you talking about?
01:41:33 It doesn’t really work that way.
01:41:34 You should be like super into the idea,
01:41:37 so into it that you wanna spend a really long time on it.
01:41:41 Is that more about psychology or actual time allocation?
01:41:44 Like, is it literally the fact that you need to give 100%
01:41:47 for potentially years for it to succeed?
01:41:49 Or is it more about just the mindset that’s required?
01:41:53 Yeah, I mean, I think, well, any, I think, yeah,
01:41:55 you don’t wanna have,
01:41:56 certainly don’t wanna have a plan to sell the company
01:42:00 like quickly or something,
01:42:02 or it’s like a company that has a very,
01:42:05 it’s like a big fashion component.
01:42:07 Like it’ll only work now.
01:42:08 It’s like an app or something.
01:42:12 So yeah, that’s a big one.
01:42:14 And then I also think something I’ve thought about recently
01:42:18 is I had a job as a quant at a fund
01:42:23 for about two and a half years.
01:42:25 And part of me thinks if I had spent
01:42:29 another two years there,
01:42:31 I would have learned a lot more
01:42:34 and had even more knowledge to be where,
01:42:38 to basically accelerate how long Numerai took.
01:42:41 So the idea that you can sit in an air conditioned room
01:42:44 and get free food,
01:42:46 or even sit at home now in your underwear
01:42:48 and make a huge amount of money and learn whatever you want
01:42:53 and get, it’s just crazy.
01:42:55 It’s such a good deal.
01:42:56 Yeah, oh, that’s interesting.
01:42:57 That’s the case for, I was terrified of that.
01:43:00 Like at Google, I thought I would become really comfortable
01:43:04 in that air conditioned room.
01:43:06 And that, I was afraid the quant situation is,
01:43:10 I mean, what you present is really brilliant
01:43:13 that it’s exceptionally valuable, the lessons you learn,
01:43:17 because you get to get paid while you learn from others.
01:43:21 If you see that, if you see jobs
01:43:24 in the space of your passion that way,
01:43:27 that it’s just an education.
01:43:29 It’s like the best kind of education.
01:43:31 But of course you have, from my perspective,
01:43:34 you have to be really careful on that to get comfortable.
01:43:37 Again, in a relationship, then you buy a house
01:43:39 or whatever the hell it is, and then you get,
01:43:42 and then you convince yourself like,
01:43:44 well, I have to pay these fees for the car,
01:43:46 for the house, blah, blah, blah.
01:43:48 And then there’s momentum and all of a sudden
01:43:50 you’re on your death bed and there’s grandchildren
01:43:53 and you’re drinking whiskey
01:43:55 and complaining about kids these days.
01:43:56 So I’m afraid of that momentum, but you’re right.
01:44:00 Like there’s something special about the education
01:44:04 you get working at these companies.
01:44:06 Yeah, and I remember on my desk,
01:44:08 I had like a bunch of papers on quant finance,
01:44:11 a bunch of papers on optimization,
01:44:13 and then the paper on Ethereum, just on my desk as well,
01:44:16 and the white paper, and it’s like,
01:44:19 it’s been amazing how kind of, and you can learn about,
01:44:23 so that, I also thought, I think this idea
01:44:25 of like learning about intersections of things,
01:44:28 I don’t think there are too many people that know
01:44:30 like as much about crypto and quant finance
01:44:33 and machine learning as I do.
01:44:36 And that’s a really nice set of three things
01:44:39 to know stuff about.
01:44:40 And that was because I had like free time in my job.
01:44:45 Okay, let me ask the perfectly impractical,
01:44:48 but the most important question.
01:44:49 What’s the meaning of all the things you’re trying to do
01:44:53 so many amazing things, why?
01:44:56 What’s the meaning of this life of yours or ours?
01:45:00 I don’t know.
01:45:01 Humans.
01:45:03 Yeah, so I have yet had some people say,
01:45:06 asking what meaning of life is,
01:45:07 is like asking the wrong question or something.
01:45:10 The question is wrong.
01:45:11 Yeah.
01:45:12 No, usually people get too nervous to be able to say that
01:45:14 because it’s like your question sucks.
01:45:17 I don’t think there’s an answer.
01:45:18 It’s like the searching for it.
01:45:21 It’s like sometimes asking it.
01:45:22 It’s like sometimes sitting back and looking up at the stars
01:45:25 and being like, huh, I wonder if there’s aliens up there.
01:45:29 There’s a useful like a palette cleanser aspect to it
01:45:36 because it kind of wakes you up to like all the little busy,
01:45:39 hurried day to day activities, all the meetings,
01:45:42 all the things you’d like a part of.
01:45:45 We’re just like ants, a part of a system,
01:45:47 a part of another system.
01:45:48 And then asking this bigger question
01:45:52 allows you to kind of zoom out and think about it.
01:45:53 But there’s ultimately,
01:45:56 I think it’s an impossible thing for a limited capacity,
01:45:58 like cognitive capacity to capture.
01:46:01 But it’s fun to listen to somebody
01:46:03 who’s exceptionally successful, exceptionally busy now,
01:46:07 who’s also young like you,
01:46:09 to ask these kinds of questions about like death.
01:46:14 You know, do you consider your own mortality kind of thing
01:46:18 and life, whether that enters your mind?
01:46:21 Because it often doesn’t,
01:46:23 it kind of almost gets in the way.
01:46:24 Yeah.
01:46:26 It’s amazing how many things you can like that are trivial
01:46:28 that could occupy a lot of your mind
01:46:31 until something bad happens or something flips you.
01:46:36 And then you start thinking about the people you love
01:46:38 that are in your life.
01:46:39 Then you started thinking about like,
01:46:41 holy shit, this ride ends.
01:46:42 Exactly.
01:46:43 Yeah, I just had COVID and I had it quite bad.
01:46:48 What wasn’t really bad is just like,
01:46:51 I also got a simultaneous like lung infection.
01:46:55 So I had like almost like bronchitis or whatever.
01:46:59 I don’t even, I don’t understand that stuff,
01:47:01 but I started, and then you’re forced to be isolated.
01:47:06 Right.
01:47:07 And so it’s actually kind of nice
01:47:08 because it’s very depressing.
01:47:12 And then I’ve heard stories of, I think it’s Sean Parker.
01:47:15 He had like all these diseases as a child
01:47:18 and he had to like just stay in bed for years.
01:47:20 And then he like made Napster.
01:47:23 It’s like pretty cool.
01:47:24 So yeah, I had about 15 days of this recently,
01:47:27 just last month.
01:47:28 And it feels like it did shock me
01:47:30 into a new kind of energy and ambition.
01:47:34 Were there moments when you were just like terrified
01:47:37 at the combination of loneliness?
01:47:39 And like, you know, the thing about COVID is like,
01:47:43 there’s some degree of uncertainty.
01:47:45 Like it feels like it’s a new thing, a new monster
01:47:48 that’s arrived on this earth.
01:47:50 And so, you know, dealing with it alone,
01:47:54 a lot of people are dying.
01:47:55 It’s like wondering like.
01:47:57 Yeah, you do wonder, I mean, for sure.
01:47:59 And then there are the even new strains in South Africa,
01:48:03 which is where I was.
01:48:04 And maybe the new strain had some interaction
01:48:06 with my genes and I’m just gonna die.
01:48:09 But ultimately it was liberating somehow.
01:48:11 I loved it.
01:48:12 Oh, I loved that I got out of it.
01:48:15 Okay.
01:48:16 Because it also affects your mind.
01:48:17 You get confused, you get confusion
01:48:18 and kind of a lot of fatigue
01:48:21 and you can’t do your usual tricks
01:48:23 of psyching yourself out of it.
01:48:24 So, you know, sometimes it’s like, oh man, I feel tired.
01:48:27 Okay, I’m just gonna go have coffee and then I’ll be fine.
01:48:29 It’s like, now it’s like, I feel tired.
01:48:31 I don’t even wanna get out of bed to get coffee
01:48:33 because I feel so tired.
01:48:34 And then you have to confront,
01:48:37 there’s no like quick fix cure and you’re trapped at home.
01:48:40 But that, so now you have this little thing
01:48:43 that happened to you that was a reminder
01:48:44 that you’re mortal and you get to carry that flag
01:48:48 in trying to create something special in this world.
01:48:53 Right?
01:48:54 With Numerai.
01:48:54 Listen, this was like one of my favorite conversation
01:48:58 because the way you think about this world of money
01:49:03 and just this world in general is so clear
01:49:05 and you’re able to explain it so eloquently.
01:49:08 Richard, that was really fun.
01:49:09 Really appreciate you talking to me.
01:49:11 Thank you.
01:49:11 Thank you.
01:49:13 Thanks for listening to this conversation with Richard Crave
01:49:15 and thank you to our sponsors,
01:49:17 Audible Audio Books, Trial Labs,
01:49:20 Machine Learning Company, Blinkist app
01:49:22 that summarizes books and Athletic Greens
01:49:25 all in one nutrition drink.
01:49:27 Click the sponsor links to get a discount
01:49:29 and to support this podcast.
01:49:32 And now let me leave you with some words from Warren Buffett.
01:49:36 Games are won by players who focus on the playing field,
01:49:40 not by those whose eyes are glued to the scoreboard.
01:49:44 Thank you for listening and hope to see you next time.