back to indexRichard Craib: WallStreetBets, Numerai, and the Future of Stock Trading | Lex Fridman Podcast #159
link |
The following is a conversation with Richard Crabe,
link |
founder of Numeri, which is a crowdsourced hedge fund,
link |
very much in the spirit of Wall Street Bets,
link |
but where the trading is done not directly by humans,
link |
but by artificial intelligence systems
link |
submitted by those humans.
link |
It's a fascinating and extremely difficult
link |
machine learning competition
link |
where the incentives of everybody is aligned,
link |
the code is kept and owned by the people who develop it,
link |
the data, anonymized data is very well organized
link |
and made freely available.
link |
I think this kind of idea has a chance to change
link |
the nature of stock trading
link |
and even just money management in general
link |
by empowering people who are interested in trading stocks
link |
with the modern and quickly advancing tools
link |
of machine learning.
link |
Quick mention of our sponsors,
link |
Audible Audio Books,
link |
Trial Labs Machine Learning Company,
link |
Blinkist app that summarizes books,
link |
and Athletic Greens, all in one nutrition drink.
link |
Click the sponsor links to get a discount
link |
and to support this podcast.
link |
As a side note, let me say that this whole set of events
link |
around GameStop and Wall Street Bets
link |
has been really inspiring to me as a demonstration
link |
that a distributed system,
link |
a large number of regular people
link |
are able to coordinate and collaborate
link |
in taking on the elite centralized power structures,
link |
especially when those elites are misbehaving.
link |
I believe that power in as many cases as possible
link |
should be distributed.
link |
And in this case, the internet, as it is for many cases,
link |
is the fundamental enabler of that power.
link |
And at the core, what the internet
link |
in its distributed nature represents is freedom.
link |
Of course, the thing about freedom
link |
is it enables chaos or progress, or sometimes both.
link |
And that's kind of the point of the thing.
link |
Freedom is empowering, but ultimately unpredictable.
link |
And I think in the end, freedom wins.
link |
If you enjoy this podcast, subscribe on YouTube,
link |
review it on Apple Podcasts, follow on Spotify,
link |
support on Patreon, or connect with me on Twitter
link |
And now, here's my conversation with Richard Crabe.
link |
From your perspective, can you summarize
link |
the important events around this amazing saga
link |
that we've been living through of Wall Street Bets,
link |
the subreddit and GameStop, and in general,
link |
just what are your thoughts about it
link |
from a technical to the philosophical level?
link |
I think it's amazing.
link |
It's like my favorite story ever.
link |
Like when I was reading about it,
link |
I was like, this is the best.
link |
And it's also connected with my company,
link |
which we can talk about.
link |
But what I liked about it is like,
link |
I like decentralized coordination
link |
and looking at the mechanisms
link |
that these are Wall Street Bets users use
link |
to hype each other up, to get excited,
link |
to prove that they bought the stock and they're holding.
link |
And then also to see that how big of an impact
link |
that that decentralized coordination had.
link |
So it really was a big deal.
link |
Were you impressed by the distributed coordination,
link |
the collaboration amongst like,
link |
I don't know what the numbers are.
link |
I know I'm numerized looking at the data.
link |
After all of this is over and done,
link |
it'd be interesting to see like
link |
from a large scale distributed system perspective
link |
to see how everything played out.
link |
But just from your current perspective, what we know,
link |
is it obvious to you that such incredible level
link |
of coordination could happen
link |
where a lot of people come together in a distributed sense,
link |
there's an emergent behavior that happens after that.
link |
No, it's not at all obvious.
link |
And one of the reasons is the lack of credibility.
link |
To coordinate with someone,
link |
you need to make credible contracts or credible claims.
link |
So if you have a username on our Wall Street Bets,
link |
like some of them are, like deep fucking value
link |
That's an actual username.
link |
By the way, we're talking about,
link |
there's a website called Reddit
link |
and there's subreddits on it.
link |
And a lot of people, mostly anonymous,
link |
I think for the most part anonymous,
link |
can create user accounts
link |
and then can then just talk on forum like style boards.
link |
You should know what Reddit is.
link |
If you don't know what Reddit is, check it out.
link |
If you don't know what Reddit is,
link |
maybe go to the awesome subreddit first,
link |
aww with cute pictures of cats and dogs.
link |
That's my recommendation.
link |
Okay, yeah, that would be a good start to Reddit.
link |
When you get into it more, go to our Wall Street Bets.
link |
It gets dark quickly.
link |
We'll probably talk about that too.
link |
So yeah, so there's these users
link |
and there's no contracts, like you were saying.
link |
There's no contracts, the users are anonymous,
link |
but there are little things that do help.
link |
if you've posted a really good investment idea in the past,
link |
that exists on Reddit as well.
link |
And it might have lots of upvotes.
link |
And that's also kind of like giving credibility
link |
to your next thing.
link |
And then they are also putting up screenshots,
link |
like here's the trades I've made and here's a screenshot.
link |
Now you could fake the screenshot,
link |
but still it seems like if you've got a lot of karma
link |
and you've had a good performance on the community,
link |
it somehow becomes credible enough
link |
for other people to be like, you know what?
link |
He actually probably did put a million dollars into this.
link |
And you know what, I can follow that trade easily.
link |
And there's a bunch of people like that.
link |
So you're kind of integrating all that information
link |
together yourself to see like, huh,
link |
there's something happening here.
link |
And then you jump onto this little boat of like behavior,
link |
like we should buy the stock or sell the stock.
link |
And then another person jumps on,
link |
another person jumps on.
link |
And all of a sudden you have just a huge number of people
link |
behaving in the same direction.
link |
It's like flock of whatever birds.
link |
What was strange with this one,
link |
it wasn't just let's all buy Tesla.
link |
We love Elon, we love Tesla, let's all buy Tesla.
link |
Because that we've heard before, right?
link |
Everybody likes Tesla.
link |
Well, now they do.
link |
So what they did with this in this case,
link |
they're buying a stock that was bad.
link |
They're buying it because it was bad.
link |
And that's really weird because that's a little bit
link |
too galaxy brain for a decentralized community.
link |
How did they come up with it?
link |
How did they know that was the right one?
link |
And the reason they liked it
link |
is because it had really, really high short interest.
link |
It had been shorted more than its own float, I believe.
link |
And so they figured out that if they all bought
link |
this bad stock, they could short squeeze some hedge funds.
link |
And those hedge funds would have to capitulate
link |
and buy the stock at really, really high prices.
link |
And we should say that shorted means that
link |
these are a bunch of people, when you short a stock,
link |
you're betting on the, you're predicting
link |
that the stock's going to go down
link |
and then you will make money if it does.
link |
And then what's a short squeeze?
link |
It's really that if you are a hedge fund
link |
and you take a big short position in a company,
link |
there's a certain level at which
link |
you can't sustain holding that position.
link |
There's no limit to how high a stock can go,
link |
but there is a limit to how low it can go, right?
link |
So if you short something, you have infinite loss potential.
link |
And if the stock doubles overnight, like GameStop did,
link |
you're putting a lot of stress on that hedge fund.
link |
And that hedge fund manager might have to say,
link |
you know what, I have to get out of the trade.
link |
And the only way to get out is to buy the bad stock
link |
that they don't want, like they believe will go down.
link |
So it's an interesting situation,
link |
particularly because it's not zero sum.
link |
If you say, let's all get together
link |
and make a bubble in watermelons,
link |
you buy a bunch of watermelons,
link |
the price goes up, it comes down again,
link |
it's a zero sum game.
link |
If someone's already shorted a stock
link |
and you can make them short squeeze,
link |
it's actually a positive sum game.
link |
So yes, some Redditors will make a lot of money,
link |
some will lose a lot,
link |
but actually the whole group will make money.
link |
And that's really why it was such a clever thing
link |
And coupled to the fact that shorting,
link |
I mean, maybe you can push back,
link |
but to me always from an outsider's perspective,
link |
seemed, I hope I'm not using too strong of a word,
link |
but it seemed almost unethical.
link |
Maybe not unethical, maybe it's just a asshole thing to do.
link |
Okay, I'm speaking not from an economics
link |
or financial perspective,
link |
I'm speaking from just somebody who loves,
link |
I'm a fan of a lot of people,
link |
I love celebrating the success of a lot of people.
link |
And this is like the stock market equivalent of like haters.
link |
I know that's not what it is.
link |
I know that there's efficient,
link |
you wanna have an economy efficient mechanism
link |
for punishing sort of overhyped, overvalued things.
link |
That's what shorthand guess is designed for.
link |
But it just always felt like these people are just,
link |
because they're not just betting on the loss of the company.
link |
It feels like they're also using their leverage and power
link |
to manipulate media or just to write articles
link |
or just to hate on you on social media.
link |
Then you get to see that with Elon Musk and so on.
link |
So this is like the man,
link |
so people like hedge funds that were shorting
link |
are like the sort of embodiment of the evil
link |
or just the bad guy, the overpowerful
link |
that's misusing their power.
link |
And here's the crowd,
link |
the people that are standing up and rising up.
link |
So it's not just that they were able to collaborate
link |
on Wall Street bets to sort of effectively
link |
make money for themselves.
link |
It's also that this is like a symbol
link |
of the people getting together
link |
and fighting the centralized elites, the powerful.
link |
And that, I don't know what your thoughts are
link |
about that in general.
link |
At this stage, it feels like that's really exciting
link |
that people have power,
link |
just like regular people have power.
link |
At the same time, it's scary a little bit
link |
because just studying history,
link |
people could be manipulated by charismatic leaders.
link |
And so just like Elon right now is manipulating,
link |
encouraging people to buy Dogecoin or whatever,
link |
there can be good charismatic leaders
link |
and there can be bad charismatic leaders.
link |
And so it's nerve wracking.
link |
It's a little bit scary how much power
link |
a subreddit can have to destroy somebody
link |
because right now we're celebrating
link |
they might be attacking or destroying somebody
link |
that everybody doesn't like,
link |
but what if they attack somebody
link |
that is actually good for this world?
link |
So that, and that's kind of the awesomeness
link |
and the price of freedom.
link |
It's like it could destroy the world
link |
or it can save the world.
link |
But at this stage, it feels like, I don't know,
link |
overall, when you sit back,
link |
do you think this was just a positive wave
link |
of emergent behavior?
link |
Is there something negative about what happened?
link |
Well, yeah, the cool thing is they weren't doing anything,
link |
the Reddit people weren't doing anything exotic.
link |
It was a creative trade, but it wasn't exotic.
link |
It wasn't, it was just buying the stock.
link |
Okay, maybe they bought some options too,
link |
but it was the hedge fund that was doing the exotic thing.
link |
It was, it's hard to say, well, we've got together
link |
and we've pulled all our money together
link |
and now there's a company out there that's worth more.
link |
What's wrong with that?
link |
But it doesn't talk about the motivations, which is,
link |
and then we destroyed some hedge funds in the process.
link |
Is there something to be said about the humor
link |
and the, I don't know, the edginess,
link |
sometimes viciousness of that subreddit?
link |
I haven't looked at it too much,
link |
but it feels like people can be quite aggressive on there.
link |
So is there, what is that?
link |
Is that what Freedom looks like?
link |
I think it does, yeah.
link |
You definitely need to let people,
link |
one of the things that people have compared it to
link |
is the Occupy Wall Street, which is, let's say,
link |
some very sincere liberals, like 23 years old, whatever,
link |
and they go out with signs
link |
and they have some kind of case to make.
link |
But this isn't sincere, really.
link |
It's like a little bit more nihilistic,
link |
a little bit more YOLO, and therefore a little bit
link |
more scary because who's scared of the Occupy Wall Street
link |
people with the signs?
link |
But these hedge funds really are scared.
link |
I was scared of the Wall Street bats people.
link |
I'm still scared of them.
link |
Yeah, the anonymity is a bit terrifying and exciting.
link |
I mean, yeah, I don't know what to do with this.
link |
I've been following events in Russia, for example.
link |
It's like there's a struggle between centralized power
link |
and the distributed.
link |
I mean, that's the struggle of the history
link |
of human civilization, right?
link |
But this on the internet, just that you can multiply people.
link |
Like some of them don't have to be real.
link |
Like you can probably create bots.
link |
Like it starts getting me, me as a programmer,
link |
I start to think like, hmm, me as one person,
link |
how much chaos can I create by writing some bots?
link |
And I'm sure I'm not the only one thinking that.
link |
There's, I'm sure there's hundreds, thousands
link |
of good developers out there listening to this,
link |
thinking the same thing.
link |
And then as that develops further and further
link |
in the next like decade or two,
link |
what impact does that have on financial markets,
link |
on just destruction of reputations of just,
link |
or politics, the bickering of left and right
link |
political discourse, the dynamics of that
link |
being manipulated by, you know,
link |
people talk about like Russian bots or whatever.
link |
We're probably in a very early stage of that, right?
link |
And this is a good example.
link |
So do you have a sense that most of WallStreetBets folks
link |
are actually individual people, right?
link |
That's the feeling I have is they're just individual,
link |
maybe young investors, just doing a little bit
link |
of an investment, but just on a large scale.
link |
The reason I found out, I've known about WallStreetBets
link |
for a while, but the reason I found out about GameStop
link |
was this, just I met somebody at a party
link |
who told me about it and he was like 21 years old
link |
and he's like, it's gonna go up 100% in the next one day.
link |
That we're talking about in last year?
link |
This was probably, no, this was, yeah, a few days ago.
link |
Yeah, it was like maybe two weeks ago or something.
link |
So it was already high GameStop,
link |
but it was just strange to me that there was someone
link |
telling me at a party how to trade stocks
link |
who was like 21 years old.
link |
And I started to, yeah, I started to look into it.
link |
And yeah, and he did make, he made it, yeah,
link |
he made 140% in one day, he was right.
link |
And now he's supercharged, he's a little bit wealthier
link |
and now he's gonna wait for the next thing.
link |
And this decentralized entity
link |
is just gonna get bigger and bigger.
link |
And they're gonna together search for the next thing.
link |
So there's thousands of folks like him
link |
and they're going to probably search
link |
for the next thing to attack.
link |
People that have power in this world that sit there
link |
with power right now in government and in finance
link |
and any kind of position
link |
are probably a little bit scared right now.
link |
And honestly, that's probably a little bit good.
link |
It's dangerous, but it's good.
link |
Yeah, it certainly makes you think twice about shorting.
link |
It certainly makes you think twice
link |
about putting a lot of money into a short.
link |
Like these funds put a lot into one or two names.
link |
And so it was very, very badly risk managed.
link |
Do you think shorting is, can you speak at a high level
link |
just for your own as a person, is it good for the world?
link |
Is it good for markets?
link |
I do think that the two kinds of shorting,
link |
evil shorting and chill shorting.
link |
Evil shorting is what Melvin Capital was doing.
link |
And it's like, you put a huge position down,
link |
you get all your buddies to also short it
link |
and you start making press
link |
and trying to bring this company down.
link |
And I don't think in some cases,
link |
you go out after like fraudulent companies say,
link |
this company is a fraud.
link |
Maybe that's okay.
link |
Like some, but they weren't even saying,
link |
they're just saying it's a bad company
link |
and we're going to bring it to the ground,
link |
bring it to its knees.
link |
So a quant fund like Numerai,
link |
we always have lots of positions
link |
and we never have a position
link |
that's like more than 1% of our fund.
link |
So we actually have right now, 250 shorts.
link |
I don't know any of them except for one
link |
because it was one of the meme stocks.
link |
But we shorting them not to make them go,
link |
we don't even want them to go down necessarily.
link |
That doesn't sound a bit strange that I say that,
link |
but we just want them to not go up as much as our longs.
link |
So by shorting a little bit,
link |
we can actually go long more
link |
in the things we do believe in.
link |
So when we were going long in Tesla,
link |
we could do it with more money than we had
link |
because we would borrow from banks
link |
who would lend us money to go down.
link |
Who would lend us money because we had longs and shorts,
link |
because we didn't have market exposure,
link |
we didn't have market risk.
link |
And so I think that's a good thing
link |
because that means we can short the oil companies
link |
and go long Tesla and make the future come forward faster.
link |
And I do think that's not a bad thing.
link |
So we've talked about this incredible distributed system
link |
created by Wall Street Bets.
link |
And then there's a platform which is Robinhood,
link |
which allows investors to efficiently as far,
link |
you can correct me if I'm wrong,
link |
but there's those and there's others in this new MRI
link |
that allow you to make it accessible for people to invest.
link |
But that said, Robinhood was in a centralized way
link |
applied its power to restrict trading
link |
on the stocks that we're referring to.
link |
Do you have a thoughts on actually
link |
like the things that happened?
link |
I don't know how much you were paying attention
link |
to sort of the shadiness around the whole thing.
link |
Do you think it was forced to do it?
link |
Or was there something shady going on?
link |
What are your thoughts in general?
link |
Well, I think I wanna see the alternate history.
link |
Like I wanna see the counterfactual history
link |
of them not doing that.
link |
How bad would it have gotten for hedge funds?
link |
How much more damage could have been done
link |
if the momentum of these short squeezes could continue?
link |
What happens when there are short squeezes,
link |
even if they're in a few stocks,
link |
they affect kind of all the other shorts too.
link |
And suddenly brokers are saying things like,
link |
you need to put up more collateral.
link |
So we had a short, it wasn't GameStop luckily,
link |
it was Blackberry and it went up like 100% in a day.
link |
It was one of these meme stocks, super bad company.
link |
The AIs don't like it, okay?
link |
The AIs think it's going down.
link |
What's a meme stock?
link |
A meme stock is kind of a new term for these stocks
link |
that catch memetic momentum on Reddit.
link |
And so the meme stocks were GameStop, the biggest one,
link |
GameStonk, as Elon calls it, AMC.
link |
And Blackberry was one, Nokia was one.
link |
So these are high short interest stocks as well.
link |
So these are targeted stocks.
link |
Some people say, oh, isn't it adorable
link |
that these people are investing money
link |
in these companies that are nostalgic?
link |
It's like, you go into the AMC movie theater,
link |
it's like nostalgic.
link |
It's like, no, it's not why they're doing it.
link |
It's that they had a lot of short interest.
link |
That was the main thing.
link |
And so there were high chance of short squeeze.
link |
In saying, I would love to see an alternate history,
link |
do you have a sense that that,
link |
what is your prediction of what that history would look like?
link |
Well, you wouldn't have needed very many more days
link |
of that kind of chaos to hurt hedge funds.
link |
I think it's underrated how damaging it could have been.
link |
Because when your shorts go up,
link |
your collateral requirements for them go up.
link |
It's similar to Robinhood.
link |
Like we have a prime broker that says, said to us,
link |
you need to put up like $40 per $100 of short exposure.
link |
And then the next day they said,
link |
actually you have to put up all of it, 100%.
link |
And we were like, what?
link |
But if that happens to all the short,
link |
all the commonly held hedge fund shorts,
link |
because they're all kind of holding the same things.
link |
If that happens, not only do you have to cover the short,
link |
which means you're buying the bad companies,
link |
you need to sell your good companies
link |
in order to cover the short.
link |
So suddenly like all the good companies,
link |
all the ones that the hedge funds like are coming down
link |
and all the ones that the hedge funds hate are going up
link |
in a cascading way.
link |
So I believe that if you could have had a few more days
link |
of GameStop doubling, AMC doubling,
link |
you would have had more and more hedge fund deleveraging.
link |
But so hedge funds, I mean, they get a lot of shit,
link |
but they, do you have a sense
link |
that they do some good for the world?
link |
I mean, ultimately, so, okay.
link |
First of all, Wall Street bets itself
link |
is a kind of distributed hedge fund.
link |
Numeri is a kind of hedge fund.
link |
So I got, hedge fund is a very broad category.
link |
I mean, like if some of those were destroyed,
link |
would that be good for the world?
link |
Or would there be coupled
link |
with the destroying the evil shorting,
link |
would there be just a lot of pain
link |
in terms of investment in good companies?
link |
Yeah, a thing I like to tell people
link |
if they hate hedge funds is,
link |
I don't think you want to rerun American economic history
link |
without hedge funds.
link |
So on mass they're, yeah, they're good.
link |
Yeah, you really wouldn't want to.
link |
Because hedge funds are kind of like picking up,
link |
they're making liquidity, right, in stocks.
link |
And so if you love venture capitalists,
link |
they're investing in new technology, it's so good.
link |
You have to also kind of like hedge funds
link |
because they're the reason venture capitalists exist
link |
because their companies can have a liquidity event
link |
when they go to the public markets.
link |
So it's kind of essential that we have them.
link |
There are many different kinds of them.
link |
I believe we could maybe get away
link |
with only having an AI hedge fund.
link |
But we don't necessarily need
link |
these evil billions type hedge funds
link |
that make the media and try to kill companies.
link |
But we definitely need hedge funds.
link |
Maybe from your perspective,
link |
because you run such an organization
link |
and Vlad, the CEO of Robinhood,
link |
sort of had to make decisions really quickly,
link |
probably had to wake up
link |
in the middle of the night kind of thing.
link |
And he also had a conversation with Elon Musk on Clubhouse,
link |
which I just signed up for.
link |
It was a fascinating,
link |
one of the great journalistic performances of our time
link |
Pull a surprise for Elon.
link |
How hilarious would it be if he gets a pull a surprise?
link |
And then his Wikipedia would be like journalist
link |
and part time entrepreneur.
link |
As you know, I don't know if you can comment
link |
on any aspects of that,
link |
but like, if you were Vlad,
link |
how would you do things differently?
link |
What are your thoughts about his interaction with Elon?
link |
How he should have played it differently?
link |
Like, I guess there's a lot of aspects to this interaction.
link |
One is about transparency.
link |
Like how much do you want to tell people
link |
about really what went down?
link |
There's NDAs potentially involved.
link |
How much in private do you want to push back
link |
and say, no, fuck you, to centralize power?
link |
Whatever the phone calls you're getting,
link |
which I'm sure he was getting some kind of phone calls
link |
that might not be contractual.
link |
Like it's not contracts that are forcing him,
link |
but he was being, what do you call it?
link |
Like pressured to behave in certain kinds of ways
link |
from all kinds of directions.
link |
Like what do you take from this whole situation?
link |
I was very excited to see Vlad's response.
link |
I mean, it's pretty cool to have him talk to Elon.
link |
And one of the things that like struck me
link |
in the first like few seconds of Vlad speaking was like,
link |
I was like, is Vlad like a boomer?
link |
Like, but hear me out.
link |
Like he seemed like a 55 year old man
link |
talking to a 20 year old.
link |
Elon was like the 20 year old.
link |
And he's like the 55 year old man.
link |
You can see why Citadel are NMR buddies, right?
link |
Like you can, you can see why.
link |
It's like, this is a nice, it's not a bad thing.
link |
It's like, he's got a respectable professional attitude.
link |
Well, he also tried to do like a jokey thing.
link |
Like, no, we're not being ageist here.
link |
Boomer, but like a 60 year old CEO of Bank of America
link |
would try to make a joke for the kids.
link |
That's what Vlad's like.
link |
Yeah, I was like, what is this?
link |
This guy's like, what is he, 30?
link |
And I'm like, this is weird.
link |
But I think, and maybe that's also what I like
link |
about Elon's kind of influence on American business
link |
is like, he's super like anti the professional.
link |
Like why say, you know, a hundred words about nothing?
link |
And so I liked how he was cutting in and saying,
link |
Vlad, what do you mean?
link |
Spill the beans, bro.
link |
Yeah, so you don't have to be courteous.
link |
It's like the first principles thinking.
link |
It's like, what the hell happened?
link |
Let's just talk like normal people.
link |
The problem of course is, you know, for Elon,
link |
it's cost them, what is it?
link |
Tens of millions of dollars is tweeting like that.
link |
But perhaps it's a worthy price to pay
link |
because ultimately there's something magical
link |
about just being real and honest
link |
and just going off the cuff and making the mistakes
link |
and paying for them, but just being real.
link |
And then moments like this,
link |
that was an opportunity for Vlad to be that.
link |
And it felt like he wasn't.
link |
Do you think we'll ever find out what really went down
link |
if there was something shady underneath it all?
link |
Yeah, I mean, it would be sad if nothing shady happened,
link |
but his presence made it shady.
link |
Sometimes I feel like that would mark Zuckerberg,
link |
the CEO of Facebook.
link |
Sometimes I feel like, yeah,
link |
there's a lot of shitty things that Facebook is doing,
link |
but sometimes I think he makes it look worse
link |
by the way he presents himself about those things.
link |
Like I honestly think that a large amount of people
link |
at Facebook just have a huge unstable chaotic system
link |
and they're all, not all, but mass are trying to do good
link |
with this chaotic system.
link |
But the presentation is like,
link |
it sounds like there's a lot of back room conversations
link |
that are trying to manipulate people.
link |
And there's something about the realness that Elon has
link |
that it feels like CEO should have
link |
and Vlad had that opportunity.
link |
I think Mark Zuckerberg had that too when he was younger.
link |
And somebody said, you gotta be more professional, man.
link |
You can't say, you know, lol to an interview.
link |
And then suddenly he became like this distant person
link |
Like you'd rather have him make mistakes,
link |
but be honest than be like professional
link |
and never make mistakes.
link |
Yeah, one of the difficult hires I think
link |
is like marketing people or like PR people
link |
is you have to hire people that get the fact
link |
that you can say lol on an interview.
link |
Or, you know, take risks as opposed to what the PR,
link |
I've talked to quite a few big CEOs
link |
and the people around them are trying to constantly
link |
minimize risk of like, what if he says the wrong thing?
link |
What if she says the wrong thing?
link |
It's like, what, be careful.
link |
It's constantly like, ooh, like, I don't know.
link |
And there's this nervous energy that builds up over time
link |
with larger and larger teams where the whole thing,
link |
like I visited YouTube, for example.
link |
Everybody I talked at YouTube, incredible engineering
link |
and incredible system, but everybody's scared.
link |
Like, let's be honest about this like madness
link |
that we have going on of huge amounts of video
link |
that we can't possibly ever handle.
link |
There's a bunch of hate on YouTube.
link |
There's this chaos of comments,
link |
bunch of conspiracy theories, some of which might be true.
link |
And then just like this mess that we're dealing with
link |
and it's exciting, it's beautiful.
link |
It's a place where like democratizes education,
link |
all that kind of stuff.
link |
And instead they're all like sitting in like,
link |
trying to be very polite and saying like,
link |
well, we just want to improve the health of our platforms.
link |
Like, it's like this discussion like,
link |
all right, man, let's just be real.
link |
Let's both advertise how amazing this fricking thing is,
link |
but also to say like, we don't know what we're doing.
link |
We have all these Nazis posting videos on YouTube.
link |
We don't know how to like handle it.
link |
And just being real like that,
link |
because I suppose that's just a skill.
link |
Maybe it can't be taught, but over time,
link |
the whatever the dynamics of the company is,
link |
it does seem like Zuckerberg and others get worn down.
link |
They just get tired.
link |
They get tired of.
link |
Of not being real, which is sad.
link |
So let's talk about Numerai,
link |
which is an incredible company system idea, I think,
link |
but good place to start.
link |
What is Numerai and how does it work?
link |
So Numerai is the first hedge fund
link |
that gives away all of its data.
link |
So this is like probably the last thing
link |
a hedge fund would do, right?
link |
Why would we give away a data?
link |
It's like giving away your edge.
link |
But the reason we do it is because we're looking for people
link |
to model our data.
link |
And the way we do it is by obfuscating the data.
link |
So when you look at Numerai's data
link |
that you can download for free,
link |
it just looks like a million rows
link |
of numbers between zero and one.
link |
And you have no idea what the columns mean,
link |
but you do know that if you're good at machine learning
link |
or have done regressions before,
link |
you know that I can still find patterns in this data,
link |
even though I don't know what the features mean.
link |
And the data itself is a time series data.
link |
And even though it's obfuscated, anonymized,
link |
what is the source data like approximately?
link |
What are we talking about?
link |
So we are buying data from lots of different data vendors
link |
and they would also never want us to share that data.
link |
So we have strict contracts with them.
link |
but that's the kind of data you could never buy yourself
link |
unless you had maybe a million dollars a year
link |
of budget to buy data.
link |
So what's happened with the hedge fund industry
link |
is you have a lot of talented people
link |
who used to be able to trade and still can trade,
link |
but now they have such a data disadvantage,
link |
it would never make sense for them to trade themselves.
link |
But Numerai, by giving away this obfuscated data,
link |
we can give them a really, really high quality data set
link |
that would otherwise be very expensive.
link |
And they can use whatever new machine learning technique
link |
they want to find patterns in that data
link |
that we can use in our hedge fund.
link |
And so how much variety is there in underlying data?
link |
We're talking about,
link |
I apologize if I'm using the wrong terms,
link |
but one is just like the stock price.
link |
The other, there's like options and all that kind of stuff,
link |
like the, what are they called, order books or whatever.
link |
Is there maybe other totally unrelated
link |
directly to the stock market data,
link |
like natural language as well, all that kind of stuff?
link |
Yeah, we were really focused on stock data
link |
that's specific to stocks.
link |
So things like you can have like a,
link |
every stock has like a PE ratio.
link |
For some stocks, it's not as meaningful,
link |
but every stock has that.
link |
Every stock has one year momentum,
link |
how much they went up in the last year,
link |
but those are very common factors.
link |
But we try to get lots and lots of those factors
link |
that we have for many, many years,
link |
like 15, 20 years history.
link |
And then the setup of the problem is commonly in quant
link |
called like cross sectional global equity.
link |
You're not really trying to say,
link |
I believe this stock will go up.
link |
You're trying to say the relative position of this stock
link |
in feature space makes it not a bad buy in a portfolio.
link |
So it captures some period of time
link |
and you're trying to find the patterns,
link |
the dynamics captured by the data of that period of time
link |
in order to make short term predictions
link |
about what's going to happen.
link |
Yeah, so our predictions are also not that short.
link |
We're not really caring about things like order books
link |
and tech data, not high frequency at all.
link |
We're actually holding things for quite a bit longer.
link |
So our prediction time horizon is about one month.
link |
We end up holding stocks
link |
for maybe like three or four months.
link |
So I kind of believe that's a little bit more
link |
like investing than kind of plumbing,
link |
like to go long a stock that's mispriced on one exchange
link |
and short on another exchange, that's just arbitrage.
link |
But what we're trying to do is really know something more
link |
about the longer term future of the stock.
link |
Yeah, so from the patterns,
link |
from these like periods of time series data,
link |
you're trying to understand something fundamental
link |
about the stock, not like about deep value,
link |
about like it's big in the context of the market,
link |
is it underpriced, overpriced, all that kind of stuff.
link |
So like, this is about investing.
link |
It's not about like, just like you said,
link |
high frequency trading,
link |
which I think is a fascinating open question
link |
from a machine learning perspective,
link |
but just to like sort of build on that.
link |
So you've anonymized the data
link |
and now you're giving away the data.
link |
And then now anyone can try to build algorithms
link |
that make investing decisions on top of that data
link |
or predictions on the top of that data.
link |
And so that's, what does that look like?
link |
What's the goal of that?
link |
What are the underlying principles of that?
link |
So the first thing is,
link |
we could obviously model that data in house, right?
link |
We can make an XGBoost model on the data
link |
and that would be quite good too.
link |
But what we're trying to do is by opening it up
link |
and letting anybody participate,
link |
we can do quite a lot better than if we modeled it ourselves
link |
and a lot better on the stock market
link |
doesn't need to be very much.
link |
Like it really matters the difference
link |
between if you can make 10 and 12%
link |
in an equity market neutral hedge fund
link |
because usually you're charging 2% fees.
link |
So if you can do 2% better,
link |
that's like all your fees, it's worth it.
link |
So we're trying to make sure
link |
that we always have the best possible model
link |
as new machine learning libraries come out,
link |
new techniques come out,
link |
they get automatically synthesized.
link |
Like if there's a great paper on supervised learning,
link |
someone on Numerai will figure out
link |
how to use it on Numerai's data.
link |
And is there an ensemble of models going on
link |
or is it more towards kind of like one or two or three
link |
like best performing models?
link |
So the way we decide on how to weight
link |
all of the predictions together
link |
is by how much the users are staking on them.
link |
How much of the cryptocurrency
link |
that they're putting behind their models.
link |
So they're saying, I believe in my model.
link |
You can trust me because I'm gonna put skin in the game.
link |
And so we can take the stake weighted predictions
link |
from all our users, add those together,
link |
average those together,
link |
and that's a much better model than any one model
link |
in the sum because ensembling a lot of models together
link |
is kind of the key thing you need to do in investing too.
link |
Yeah, so you're putting,
link |
so there's a kind of duality from the user,
link |
from the perspective of a machine learning engineer
link |
where it's both a competition, just a really interesting,
link |
difficult machine learning problem,
link |
and it's a way to invest algorithmically.
link |
So like, but the way to invest algorithmically
link |
also is a way to put skin in the game
link |
that communicates to you that the quality of the algorithm
link |
and also forces you to really be serious
link |
about the models that you build.
link |
So it's like, everything just works nicely together.
link |
Like, I guess one way to say that
link |
is the interests are aligned.
link |
Okay, so it's just like poker is not fun
link |
when it's like for very low stakes.
link |
The higher the stakes,
link |
the more the dynamics of the system
link |
starts playing out correctly.
link |
Like as a small side note,
link |
is there something you can say about which kind,
link |
looking at the big broad view of machine learning today
link |
or AI, what kind of algorithms seem to do good
link |
in these kinds of competitions at this time?
link |
Is there some universal thing you can say,
link |
like neural networks suck,
link |
recurrent neural networks suck, transformers suck,
link |
or they're awesome, like old school,
link |
sort of more basic kind of classifiers are better,
link |
all that, is there some kind of conclusions
link |
so far that you can say?
link |
There is definitely something pretty nice about tree models,
link |
and they just seem to work pretty nicely
link |
on this type of data.
link |
So out of the box,
link |
if you're trying to come a hundredth
link |
in the competition, in the tournament,
link |
maybe you would try to use that.
link |
But what's particularly interesting about the problem
link |
that not many people understand,
link |
if you're familiar with machine learning,
link |
this typically will surprise you when you model our data.
link |
So one of the things that you look at in finance
link |
is you don't wanna be too exposed to any one risk.
link |
Like, even if the best sector in the world
link |
to invest in over the last 10 years was tech,
link |
does not mean you should put all of your money into tech.
link |
So if you train a model,
link |
it would say, put all your money in tech, it's super good.
link |
But what you wanna do is actually be very careful
link |
of how much of this exposure you have to certain features.
link |
So on Numeri, what a lot of people figure out is,
link |
actually, if you train a model on this kind of data,
link |
you wanna somehow neutralize or minimize your exposure
link |
to these certain features, which is unusual,
link |
because if you did train a stoplight
link |
or stop street detection on computer vision,
link |
your favorite feature, let's say you have an auto encoder
link |
and it's figuring out, okay, it's gotta be red
link |
and it's gotta be white,
link |
that's the last thing you wanna reduce your exposure to.
link |
Why would you reduce your exposure
link |
to the thing that's helping your model the most?
link |
And that's actually this counterintuitive thing
link |
you have to do with machine learning on financial data.
link |
So reducing your exposure
link |
would help you generalize the things that are...
link |
So basically, a financial data has a large amount
link |
of patterns that appeared in the past
link |
and also a large amount of patterns
link |
that have not appeared in the past.
link |
And so like in that sense,
link |
you have to reduce the exposure to red lights,
link |
That's interesting, but how much of this is art
link |
and how much of it is science from your perspective so far
link |
in terms of as you start to climb
link |
from the 100th position to the 95th in the competition?
link |
Yeah, well, if you do make yourself super exposed
link |
to one or two features,
link |
you can have a lot of volatility
link |
when you're playing Numerai.
link |
You could maybe very rapidly rise to be high
link |
if you were getting lucky.
link |
And that's a bit like the stock market.
link |
Sure, take on massive risk exposure,
link |
put all your money into one stock
link |
and you might make 100%,
link |
but it doesn't in the long run work out very well.
link |
And so the best users are trying to stay high
link |
for as long as possible,
link |
not necessarily try to be first for a little bit.
link |
So me, a developer, machine learning researcher,
link |
how do I, Lex Friedman, participate in this competition
link |
and how do others,
link |
which I'm sure there'll be a lot of others
link |
interested in participating in this competition,
link |
what are, let's see, there's like a million questions,
link |
but like first one is how do I get started?
link |
Well, you can go to numero.ai,
link |
sign up, download the data.
link |
And on the data is pretty small.
link |
In the data pack you download,
link |
there's like an example script,
link |
Python script that just builds a XGBoost model
link |
very quickly from the data.
link |
And so in a very short time,
link |
you can have an example model.
link |
Is that a particular structure?
link |
Like what, is this model then submitted somewhere?
link |
So there needs to be some kind of structure
link |
that communicates with some kind of API.
link |
Like how does the whole,
link |
how does your model, once you've built,
link |
once you create a little baby Frankenstein,
link |
how does it then live in the world?
link |
Okay, well, we want you to keep your baby Frankenstein
link |
at home and take care of it.
link |
So you never upload your model to us.
link |
You always only giving us predictions.
link |
So we never see the code that wrote your model,
link |
which is pretty cool,
link |
that our whole hedge fund is built from models
link |
where we've never, ever seen the code.
link |
But it's important for the users because it's their IP,
link |
why do they want to give it to us?
link |
So they've got it themselves,
link |
but they can basically almost like license
link |
the predictions from that model to us.
link |
License the predictions, yeah.
link |
So. Think about it.
link |
What some users do is they set up a compute server
link |
and we call it Numeric Compute.
link |
It's like a little AWS kind of image
link |
and you can automate this process.
link |
So we can ping you.
link |
We can be like, we need more predictions now.
link |
And then you send it to us.
link |
So that's, is that described somewhere,
link |
like what the preferred is, the AWS,
link |
or whether another cloud platform,
link |
is there, I mean, is there sort of specific technical things
link |
you want to say that comes to mind
link |
that is a good path for getting started?
link |
So download the data, maybe play around,
link |
see if you can modify the basic algorithm provided
link |
And then you, what, set up a little server on the AWS
link |
that then runs this model and takes pings
link |
and then makes predictions.
link |
And so how does your own money actually come into play
link |
doing the stake of a cryptocurrency?
link |
Yeah, so you don't have to stake.
link |
You can start without staking.
link |
And many users might try for months
link |
without staking anything at all
link |
to see if their model works on the real life data, right?
link |
And is not overfit.
link |
But then you can get Numeraire many different ways.
link |
You can buy it on, you can buy some on Coinbase.
link |
You can buy some on Uniswap.
link |
You can buy some on Binance.
link |
So what did you say this is?
link |
How do you pronounce it?
link |
So this is the Numeraire cryptocurrency.
link |
NMR, you just say NMR?
link |
It is technically called Numeraire.
link |
Numeraire, I like it.
link |
Yeah, but NMR is simple.
link |
Okay, so, and you could buy it basically anywhere.
link |
Yeah, so it's a bit strange
link |
because sometimes people are like,
link |
is this like pay to play?
link |
And it's like, yeah, you need to put some money down
link |
to show us you believe in your model.
link |
But weirdly, we're not selling you the,
link |
like you can't buy the cryptocurrency from us.
link |
It's like, it's also, we never,
link |
if you do badly, we destroy your cryptocurrency.
link |
Okay, that's not good, right?
link |
You don't want it to be destroyed.
link |
But what's good about it is it's also not coming to us.
link |
So it's not like we win when you lose or something,
link |
like we're the house.
link |
Like we're definitely on the same team.
link |
Helping us make a hedge fund that's never been done before.
link |
Yeah, so again, interests are aligned.
link |
There's no, there's no tension there at all,
link |
which is really fascinating.
link |
You're giving away everything
link |
and then the IP is owned by sort of the code.
link |
You never share the code.
link |
That's fascinating.
link |
So since I have you here and you said a hundred,
link |
I didn't ask out of how many, so we'll just,
link |
but if I then once you get started
link |
and you find this interesting, how do you then win
link |
or do well, but also how do you potentially try to win
link |
if this is something you want to take on seriously
link |
from the machine learning perspective,
link |
not from a financial perspective?
link |
Yeah, I think that first of all,
link |
you would want to talk to the community.
link |
People are pretty open.
link |
We give out really interesting scripts and ideas
link |
for things you might want to try.
link |
And, but you're also going to need a lot of compute probably.
link |
And so some of the best users are, you know,
link |
actually the very first time someone won on Numerai,
link |
I would, I wrote them a personal email.
link |
It's like, you know, you've won some money.
link |
We're so excited to give you $300.
link |
And then they said, I spend way more on the compute,
link |
So this is fundamentally a machine learning problem first,
link |
I think is this is one of the exciting things.
link |
I don't know if we'll, in how many ways we can approach this,
link |
but really this is less about kind of no offense,
link |
but like finance people, finance minded people,
link |
they're also, I'm sure great people,
link |
but it feels like from the community that I've experienced,
link |
these are people who see finance
link |
as a fascinating problem space, source of data,
link |
but ultimately they're machine learning people or AI people,
link |
which is a very different kind of flavor of community.
link |
And I mean, I should say to that,
link |
I'd love to participate in this and I will participate in this
link |
and I'd love to hear from other people.
link |
If you're listening to this,
link |
if you're a machine learning person,
link |
you should participate in it and tell me,
link |
give me some hints how I can do well at this thing.
link |
Cause this boomer, I'm not sure I still got it,
link |
but cause some of it is, it's like a Kaggle competitions.
link |
Like some of it is certainly set of ideas,
link |
like research ideas, like fundamental innovation,
link |
but I'm sure some of it is like deeply understanding,
link |
getting like an intuition about the data.
link |
And then like a lot of it will be like figuring out
link |
like what works, like tricks.
link |
I mean, you could argue most of deep learning research
link |
is just tricks on top of tricks,
link |
but there's some of it is just the art
link |
of getting to know how to work in a really difficult
link |
machine learning problem.
link |
And I think what's important,
link |
the important difference with something
link |
like a Kaggle competition,
link |
where they'll set up this kind of toy problem
link |
and then there will be an out of sample test,
link |
like, Hey, you did well out of sample.
link |
And this is like, okay, cool.
link |
But what's cool with Numeri is the out of sample
link |
is the real life stock market.
link |
We don't even know,
link |
like we don't know the onset of the problem.
link |
We don't, like you'll have to find out live.
link |
And so we've had users who've like submitted every week
link |
for like four years because it's kind of,
link |
we say it's the hardest data science problem
link |
on the planet, right?
link |
And it sounds maybe sounds like maybe
link |
but too much for like a marketing thing,
link |
but it's the hardest because it's the stock market.
link |
It's like literally there are like billions of dollars
link |
at stake and like no one's like letting it be inefficient
link |
So if you can find something that works at Numeri,
link |
you really have something that is like working
link |
on the real stock market.
link |
Yeah, because there's like humans involved
link |
in the stock market.
link |
I mean, you could argue there might be harder data sets
link |
like maybe predicting the weather,
link |
all those kinds of things.
link |
But the fundamental statement here is, which I like,
link |
I was thinking like,
link |
is this really the hardest data science problem?
link |
And you start thinking about that,
link |
but ultimately it also boils down to a problem
link |
where the data is accessible.
link |
It's made accessible, made really easy and efficient
link |
at like submitting algorithms.
link |
So it's not just, you know,
link |
it's not about the data being out there, like the weather.
link |
It's about making the data super accessible,
link |
making the ability of community around it.
link |
Like this is what ImageNet did.
link |
Like it's not just, there's always images.
link |
The point is you aggregate them together.
link |
You give it a little title.
link |
This is a community and that was one of the hardest,
link |
right, for a time.
link |
And most important data science problems in the world
link |
because it was accessible, because it was made sort of,
link |
like there was mechanisms by which like standards
link |
and mechanisms by which you judge your performance,
link |
all those kinds of things.
link |
And Numerize actually step up from that.
link |
Is there something more you can say about why
link |
from your perspective it's the hardest problem in the world?
link |
I mean, you said it's connected to the market.
link |
So if you could find a pattern in the market,
link |
that's a really difficult thing to do
link |
because a lot of people are trying to do it.
link |
But there's also the biggest one is
link |
it's non stationary time series.
link |
We've tried to regularize the data
link |
so you can find patterns by doing certain things
link |
to the features and the target.
link |
But ultimately you're in a space where you don't,
link |
there's no guarantees that the out of sample distributions
link |
will conform to any of the training data.
link |
And every single era, which we call on the website,
link |
like every single era in the data,
link |
which is like sort of showing you the order of the time.
link |
Even the training data has the same dislocations.
link |
And so, yeah, and then there's so many things
link |
that you might wanna try.
link |
There's unlimited possible number of models, right?
link |
And so by having it be open,
link |
we can at least search that space.
link |
Zooming back out to the philosophical,
link |
you said that Numerai is very much like Wall Street Bets.
link |
Is there, I think it'd be interesting
link |
to dig in why you think so.
link |
I think you're speaking to the distributed nature of the two
link |
and the power of the people nature of the two.
link |
So maybe can you speak to the similarities
link |
and the differences and in which way is Numerai more powerful
link |
in which way is Wall Street Bets more powerful?
link |
Yeah, this is why the Wall Street Bets story
link |
is so interesting to me because it's like,
link |
feels like we're connected.
link |
And looking at how,
link |
just looking at the form of Wall Street Bets,
link |
I was talking earlier about how,
link |
how can you make credible claims?
link |
Okay, well, maybe you can take a screenshot.
link |
Or maybe you can upvote someone.
link |
Maybe you can have karma on Reddit.
link |
And those kinds of things make this emerging thing possible.
link |
Numerai, it didn't work at all when we started.
link |
It didn't work at all.
link |
People made multiple accounts.
link |
They made really random models
link |
and hoped they would get lucky.
link |
And some of them did.
link |
Staking was our solution to,
link |
could we make it so that we could trust?
link |
We could know which model people believed in the most.
link |
And we could weight models that had high stake more
link |
and effectively coordinate this group of people
link |
to be like, well, actually there's no incentive
link |
to creating bot accounts anymore.
link |
Either I stake my accounts,
link |
in which case I should believe in them
link |
because I could lose my stake or I don't.
link |
And that's a very powerful thing
link |
that having a negative incentive
link |
and a positive incentive can make things a lot better.
link |
And staking is like this,
link |
is this really nice like key thing about blockchain.
link |
It's like something special you can do
link |
where they're not even trusting us
link |
with their stake in some ways.
link |
They're trusting the blockchain, right?
link |
So the incentives, like you say,
link |
it's about making these perfect incentives
link |
so that you can have coordination to solve one problem.
link |
And nowadays I sleep easy
link |
because I have less money in my own hedge fund
link |
than our users are staking on their models.
link |
In some sense, from a human psychology perspective,
link |
it's fascinating that the WallStreetBets worked at all, right?
link |
That amidst that chaos, emerging behavior,
link |
like behavior that made sense emerged.
link |
It would be fascinating to think if numerized style staking
link |
could then be transferred to places like Reddit, you know?
link |
And not necessarily for financial investments,
link |
but I wish sometimes people would have to stake something
link |
in the comments they make on the internet.
link |
That's the problem with anonymity is like,
link |
anonymity is freedom and power
link |
that you don't have to, you can speak your mind,
link |
but it's too easy to just be shitty.
link |
So this, I mean, you're making me realize
link |
from a profoundly philosophical aspect, numerized staking
link |
is a really clean way to solve this problem.
link |
It's a really beautiful way.
link |
Of course, it only with Numerai currently works
link |
for a very particular problem, right?
link |
Not for human interaction on the internet,
link |
but that's fascinating.
link |
Yeah, there's nothing to stop people.
link |
In fact, we've open sourced the code we use for staking
link |
in a protocol we call Erasure.
link |
And if Reddit wanted to, they could even use that code
link |
to enable staking on our Wall Street pets.
link |
And they're actually researching now,
link |
they've had some Ethereum grants
link |
on how could they have more crypto stuff in there
link |
in Ethereum, because wouldn't that be interesting?
link |
Like, imagine you could, instead of seeing a screenshot,
link |
like, guys, I promise I will not sell my GameStop.
link |
We're just going to go huge.
link |
We're not going to sell at all.
link |
And here is a smart contract, which no one in the world,
link |
including me, can undo, that says,
link |
I have staked millions against this claim.
link |
And then what could you do?
link |
And of course, it doesn't have to be millions.
link |
It could be just a very small amount,
link |
but then just a huge number of users doing that kind of stake.
link |
That could change the internet.
link |
It would change, and then Wall Street.
link |
It would change Wall Street.
link |
They would never have been able to,
link |
they would still be short squeezing one day
link |
after the next, every single hedge fund collapsing.
link |
If we look into the future, do you think it's possible
link |
that numerae style infrastructure,
link |
where AI systems backed by humans are doing the trading,
link |
is what the entirety of the stock market is,
link |
or the entirety of the economy, is
link |
run by basically this army of AI systems
link |
with high level human supervision?
link |
Yeah, the thing is that some of them could be bad actors.
link |
Some of the humans?
link |
No, well, these systems could be tricky.
link |
So actually, I once met a hedge fund manager,
link |
and this is kind of interesting.
link |
He said, very famous one, and he said,
link |
we can see, sometimes we can see things in the market
link |
where we know we can make money, but it will mess shit up.
link |
We know we can make money, but it will mess things up.
link |
And we choose not to do those things.
link |
And on the one hand, maybe this is like, oh, you're
link |
being super arrogant.
link |
Of course you can't do this, but maybe he can.
link |
And maybe he really isn't doing things
link |
he knows he could do, but would be pretty bad.
link |
Would the Reddit army have that kind of morality or concern
link |
for what they're doing?
link |
Probably not, based on what we've seen.
link |
The madness of crowds.
link |
There'll be one person that says, hey, maybe,
link |
and then they get trampled over.
link |
That's the terrifying thing, actually.
link |
A lot of people have written about this,
link |
is somehow that little voice that's human morality
link |
gets silenced when we get into groups and start chanting.
link |
And that's terrifying.
link |
But I think maybe I misunderstood.
link |
I thought that you're saying AI systems can be dangerous,
link |
but you just describe how humans can be dangerous.
link |
So which is safer?
link |
So one thing is, so Wall Street bets these kinds of attacks.
link |
It's not possible to model, numerize data,
link |
and then come up with the idea from the model,
link |
let's short squeak, just game stop.
link |
It's not even framed in that way.
link |
It's not possible to have that idea.
link |
But it is possible for a bunch of humans.
link |
So I think this, it's, numera could get very powerful
link |
without it being dangerous.
link |
But Wall Street bets needs to get a little bit more powerful,
link |
and it'll be pretty dangerous.
link |
Yeah, well, I mean, this is a good place
link |
to think about numera data today and numera signals
link |
and what that looks like in 10, 20, 30, 50, 100 years.
link |
Right now, I guess, maybe you can correct me,
link |
but the data that we're working with is like a window.
link |
It's an anonymized, obfuscated window
link |
into a particular aspect, time period of the market.
link |
And you can expand that more and more and more and more,
link |
You can imagine in different dimensions
link |
to where it encapsulates all the things that,
link |
where you could include kind of human to human communication
link |
that was available to buy GameStop, for example,
link |
on Wall Street bets.
link |
So maybe as a step back, can you speak
link |
to what is numera signals and what are the different data
link |
sets that are involved?
link |
So with numera signals, you're still
link |
providing predictions to us, but you can do it
link |
from your own data sets.
link |
So numera, it's all you have to model our data
link |
to come up with predictions.
link |
Numa signals is whatever data you can find out there,
link |
you can turn it into a signal and give it to us.
link |
So it's a way for us to import signals
link |
on data we don't yet have.
link |
And that's why it's particularly valuable,
link |
because it's going to be signals.
link |
You're only rewarded for signals that are
link |
orthogonal to our core signal.
link |
So you have to be doing something uncorrelated.
link |
And so strange alternative data tends to have that property.
link |
There isn't too many other signals
link |
that are correlated with what's happening on Wall Street
link |
That's not going to be correlated with the price
link |
to earnings ratio.
link |
And we have some users, as of recently, as of a week ago,
link |
there was a user that created, I think he's in India,
link |
he created a signal that is scraped from Wall Street bets.
link |
And now we have that signal as one
link |
of our signals in thousands that we use at Numerai.
link |
And the structure of the signal is similar,
link |
so it's just numbers and time series data.
link |
And it's just like, you're providing a ranking of stocks.
link |
So you just say, a one means you like the stock,
link |
zero means you don't like the stock,
link |
and you provide that for 5,000 stocks in the world.
link |
And they somehow converted the natural language
link |
that's in the Wall Street bet.
link |
And they open sourced this Colab notebook.
link |
You can go and see it.
link |
And so, yeah, it's making a sentiment score.
link |
And then turning it into a rank of stocks.
link |
A sentiment score.
link |
Like, this stock sucks, or this stock is awesome.
link |
And then converting.
link |
That's fascinating.
link |
Just even looking at that data would be fascinating.
link |
So on the signal side, what's the vision?
link |
This long term, what do you see that becoming?
link |
So we want to manage all the money in the world.
link |
That's Numerai's mission.
link |
And to get that, we need to have all the data
link |
and have all of the talent.
link |
Like, there's no way, first principles,
link |
if you had really good modeling and really good data
link |
that you would lose, right?
link |
It's just a question of how much do you need to get really good.
link |
So Numerai already has some really nice data
link |
This year, we are 10xing that.
link |
And I actually think we'll 10x the amount of data
link |
we have on Numerai every year for at least the next 10 years.
link |
So it's going to get very big, the data we give out.
link |
And signals is more data.
link |
People with any other random data set
link |
can turn that into a signal and give it to us.
link |
And in some sense, that kind of data
link |
is the edge cases, the weirdness is the,
link |
so you're focused on the bulk, the main data.
link |
And then there's just weirdness from all over the place
link |
that just can enter through this back door of Numerai signals.
link |
And it's also a little bit shorter term.
link |
So the signals are about a seven day time horizon.
link |
And on Numerai, it's like a 30 day.
link |
So it's often for faster situations.
link |
You've written about a master plan.
link |
And you've mentioned, which I love,
link |
in a similar sort of style of big style thinking,
link |
you would like Numerai to manage all of the world's money.
link |
So how do we get there from yesterday
link |
to several years from now?
link |
Like what is the plan?
link |
So you've already started to allure to get all the data
link |
and get all the talent, humans, models.
link |
I mean, the important thing to note there is,
link |
what would that mean?
link |
And I think the biggest thing it means
link |
is if there was one hedge fund, you
link |
would have not so much talent wasted
link |
on all the other hedge funds.
link |
Like it's super weird how the industry works.
link |
It's like one hedge fund gets a data source and hires a PhD.
link |
And another hedge fund has to buy the same data source
link |
And suddenly, a third of American PhDs
link |
are working at hedge funds.
link |
And we're not even on Mars.
link |
And so in some ways, Numerai, it's
link |
all about freeing up people who work at hedge funds
link |
to go work for Elon.
link |
And also, the people who are working on Numerai problem,
link |
it feels like a lot of the knowledge
link |
there is also transferable to other domains.
link |
One of our top users, he works at NASA Jet Propulsion Lab.
link |
I went to go visit him there.
link |
And he's got Numerai posters.
link |
And it looks like the movies.
link |
It looks like Apollo 11 or whatever.
link |
Yeah, the point is he didn't quit his job to join full time.
link |
He's working on getting us to Jupiter's moon.
link |
That's his mission, the Europa Klippa mission.
link |
Actually, literally what you're saying.
link |
He's smart enough that we really want his intelligence
link |
to reach the stock market.
link |
Because the stock market's a good thing.
link |
Hedge funds are a good thing.
link |
All kinds of hedge funds, especially.
link |
But we don't want him to quit his job.
link |
So he can just do Numerai on the weekends.
link |
And that's what he does.
link |
He just made a model and it just automatically submits to us.
link |
And he's like one of our best users.
link |
You mentioned briefly that stock markets are good.
link |
From my sort of outsider perspective, is there a sense,
link |
do you think trading stocks is closer to gambling?
link |
Or is it closer to investing?
link |
Sometimes it feels like it's gambling as opposed to betting
link |
on companies that succeed.
link |
And this is maybe connected to our discussion of shorting
link |
in general, but from your sense, the way you think about it,
link |
is it fundamentally still investing?
link |
I do think, I mean, it's a good question.
link |
I've also seen lately people say, this is like speculation.
link |
Is there too much speculation in the market?
link |
And it's like, but all the trades are speculative.
link |
All the trades have a horizon.
link |
People want them to work.
link |
So I would say that there's certainly
link |
a lot of aspects of gambling math that applies to investing.
link |
Like one thing you don't do in gambling
link |
is put all your money in one bet.
link |
You have bankroll management, and it's a key part of it.
link |
And small alterations to your bankroll management
link |
might be better than improvements to your skill.
link |
And then there are things we care about in our fund.
link |
Like we want to make a lot of independent bets.
link |
We talk about it, like we want to make
link |
a lot of independent bets, because that's
link |
going to be a higher sharp than if you have a lot of bets that
link |
depend on each other, like all in one sector.
link |
But yeah, I mean, the point is that you
link |
want the prices of the stocks to be reflective of their value.
link |
Of the underlying value of the company.
link |
Yeah, you shouldn't have there be like a hedge fund that's
link |
able to say, well, I've looked at some data,
link |
and all of this stuff's super mispriced.
link |
That's super bad for society if it looks like that to someone.
link |
I guess the underlying question then
link |
is, do you see that the market often drifts away
link |
from the underlying value of companies,
link |
and it becomes a game in itself?
link |
Like with these, whatever they're called,
link |
like derivatives, like the options, and shorting,
link |
and all that kind of stuff.
link |
It's like layers of game on top of the actual,
link |
like what you said, which is like the basic thing
link |
that the Wall Street Bets was doing,
link |
which is like just buying stocks.
link |
There are a lot of games that people
link |
play that are in the derivatives market.
link |
And I think a lot of the stuff people dislike when they look
link |
at the history of what's happened,
link |
they hate like credit default swaps,
link |
or collateralized debt obligations.
link |
Like these are the enemies of 2008.
link |
And then the long term capital management thing,
link |
it was like they had 30 times leverage, or something.
link |
You could just go to a gas station
link |
and ask anybody at the gas station,
link |
is it a good idea to have 30 times leverage?
link |
And they just say no.
link |
It's like common sense just like went out the window.
link |
So yeah, I don't respect long term capital management.
link |
But Numerai doesn't actually use any derivatives
link |
unless you call shorting derivative.
link |
We just we do put money into companies.
link |
And that does help the companies we're investing in.
link |
It's just in little ways.
link |
We really did buy Tesla.
link |
And we played some role in its success.
link |
Super small, make no mistake.
link |
But still, I think that's important.
link |
Can I ask you a pothead question,
link |
which is what is money, man?
link |
So if we just kind of zoom out and look at,
link |
because I'd love to talk to you about cryptocurrency, which
link |
perhaps could be the future of money.
link |
In general, how do you think about money?
link |
You said Numerai, the vision, the goal
link |
is to run, to manage the world's money.
link |
What is money in your view?
link |
I don't have a good answer to that.
link |
But it's definitely in my personal life,
link |
it's become more and more warped.
link |
And you start to care about the real thing,
link |
like what's really going on here.
link |
Elon talks about things like this,
link |
like what is a company, really?
link |
It's a bunch of people who show up to work together
link |
and they solve a problem.
link |
And there might not be a stock out there
link |
that's trading that represents what they're doing,
link |
but it's not the real thing.
link |
And being involved in crypto, I put
link |
in crowdsale of Ethereum and all these other things
link |
and different crypto hedge funds and things
link |
that I've invested in.
link |
And it's just kind of like, it feels
link |
like how I used to think about money stuff
link |
is just totally warped.
link |
Because you stop caring about the price
link |
and you care about the product.
link |
So by the product, you mean the different mechanisms
link |
that money is exchanged.
link |
I mean, money is ultimately a kind of a little,
link |
one is a store of wealth, but it's also
link |
a mechanism of exchanging wealth.
link |
But what wealth means becomes a totally different thing,
link |
especially with cryptocurrency, to where it's almost
link |
like these little contracts, these little agreements,
link |
these transactions between human beings
link |
that represent something that's bigger than just cash being
link |
exchanged at 7.11, it feels like.
link |
Maybe I'll answer what is finance.
link |
It's what are you doing when you have the ability
link |
to take out a loan?
link |
You can bring a whole new future into being with finance.
link |
If you couldn't get a student loan to get a college degree,
link |
you couldn't get a college degree
link |
if you didn't have the money.
link |
But now, weirdly, you can get it with, and all you have
link |
is this loan, which is like, so now you
link |
can bring a different future into the world.
link |
And that's how when I was saying earlier about if you rerun
link |
American economic history without these things,
link |
like you're not allowed to take out loans,
link |
you're not allowed to have derivatives,
link |
you're not allowed to have money,
link |
it just doesn't really work.
link |
And it's a really magic thing how much
link |
you can do with finance by bringing the future forward.
link |
Finance is empowering.
link |
We sometimes forget this, but it enables innovation.
link |
It enables big risk takers and bold builders that ultimately
link |
make this world better.
link |
You said you were early in on cryptocurrency.
link |
Can you give your high level overview
link |
of just your thoughts about the past, present, and future
link |
of cryptocurrency?
link |
Yeah, so my friends told me about Bitcoin,
link |
and I was interested in equities a lot.
link |
And I was like, well, it has no net present value.
link |
It has no future cash flows.
link |
Bitcoin pays no dividends.
link |
So I really couldn't get my head around it,
link |
that this could be valuable.
link |
And then I didn't feel like I was early in cryptocurrency,
link |
in fact, because it was like 2014.
link |
It felt like a long time after Bitcoin.
link |
But then I really liked some of the things
link |
that Ethereum was doing.
link |
It seemed like a super visionary thing.
link |
I was reading something that was just
link |
going to change the world when I was reading the white paper.
link |
And I liked the different constructs
link |
you could have inside of Ethereum
link |
that you couldn't have on Bitcoin.
link |
Like smart contracts and all that kind of stuff?
link |
And even spoke about different constructions you could have.
link |
Yeah, that's a cool dance between Bitcoin and Ethereum.
link |
It's in the space of ideas.
link |
It feels so young.
link |
Like, I wonder what cryptocurrencies will look like
link |
If Bitcoin or Ethereum 2.0 or some version
link |
will stick around or any of those,
link |
who's going to win out?
link |
Or if there's even a concept of winning out at all?
link |
Is there a cryptocurrency that you especially
link |
find interesting that technically, financially,
link |
philosophically, you think is something
link |
you're keeping your eye on?
link |
Well, I don't really.
link |
I'm not looking to invest in cryptocurrencies anymore.
link |
But I mean, and many are almost identical.
link |
I mean, there wasn't too much difference
link |
between even Ethereum and Bitcoin in some ways, right?
link |
But there are some that I like the privacy ones.
link |
I mean, I like Zcash for its coolness.
link |
It's actually a different kind of invention
link |
compared to some of the other things.
link |
OK, can you speak just briefly to privacy?
link |
Is there some mechanism of preserving
link |
some privacy of the universe?
link |
So I guess everything is public.
link |
Is that the problem?
link |
Yeah, none of the transactions are private.
link |
And so I have some numeraire.
link |
And you can just see it.
link |
In fact, you can go to a website and it says like,
link |
you can go to like, etherscan.
link |
And it'll say like, numeraire founder.
link |
And I'm like, how the hell do you guys know this?
link |
So they can reverse the near, whatever that's called.
link |
Yeah, and so they can see me move it, too.
link |
Oh, why is he moving it?
link |
So but yeah, Zcash.
link |
Also, when you can make private transactions,
link |
you can also play different games.
link |
It's like what's quite cool about Zcash
link |
is I wonder what games are being played there.
link |
So from a deeply analytical perspective,
link |
can you describe why Dogecoin is going to win?
link |
Which it surely will.
link |
Like it very likely will take over the world.
link |
And once we expand out into the universe,
link |
we'll take over the universe.
link |
Or on a more serious note, like what
link |
are your thoughts on the recent success of Dogecoin
link |
where you've spoken to sort of the meme stocks,
link |
the memetics of the whole thing, that it feels like the joke can
link |
become the reality.
link |
Like the meme, the joke has power in this world.
link |
It's like why is it correlated with Elon tweeting about it?
link |
It's not just Elon alone tweeting, right?
link |
It's like Elon tweeting and that becomes
link |
a catalyst for everybody on the internet kind of like spreading
link |
So it's the initial spark of the fire for Wall Street
link |
bets type of situation.
link |
And that's fascinating because jokes
link |
seem to spread faster than other mechanisms.
link |
Like funny shit is very effective at captivating
link |
the discourse on the internet.
link |
Yeah, and I think you can have, like I like the one meme,
link |
like Doge, I haven't heard that name in a long time.
link |
Like I think back to that meme often.
link |
That's like funny.
link |
And every time I think back to it,
link |
there's a little probability that I might buy it, right?
link |
And so imagine you have millions of people who have had
link |
all these great jokes, told them,
link |
and every now and then they reminisce,
link |
oh, that was really funny.
link |
And then they're like, let me buy some.
link |
Wouldn't that be interesting if we travel in time
link |
like multiple centuries where the entirety
link |
of the communication of the human species is like humor?
link |
Like it's all just jokes.
link |
Like we're high on probably some really advanced drugs
link |
and we're all just laughing nonstop.
link |
It's a weird like dystopian future of just humor.
link |
Elon has made me realize how like good it feels
link |
to just not take shit seriously every once in a while
link |
and just relieve like the pressure of the world.
link |
At the same time, the reason I don't always like
link |
when people finish their sentences with lol
link |
is like when you don't take anything seriously.
link |
When everything becomes a joke,
link |
then it feels like that way of thinking
link |
feels like it will destroy the world.
link |
It's like, I often think of like,
link |
will memes save the world or destroy it?
link |
Because I think both are possible directions.
link |
Yeah, I think this is a big problem.
link |
I mean, I always felt that about America,
link |
a lot of people are telling jokes kind of all the time
link |
and they're kind of good at it.
link |
And you take someone aside, an American,
link |
you're like, I really wanna have a sincere conversation.
link |
It's like hard to even keep a straight face
link |
because everything is so, there's so much levity.
link |
So it's complicated.
link |
I like how sincere actually like your Twitter can be.
link |
You're like, I am in love with the world today.
link |
I get so much shit for it, it's hilarious.
link |
I'm never gonna stop because I realize like,
link |
you have to be able to sometimes just be real
link |
and be positive and just be, say the cliche things,
link |
which ultimately those things actually capture
link |
some fundamental truths about life.
link |
And I think Elon does a good job of that
link |
from an engineering perspective of being able to joke,
link |
but everyone's mostly to pull back and be like,
link |
here's real problems, let's solve them and so on.
link |
And then be able to jump back to a joke.
link |
So it's ultimately, I think, I guess a skill that we
link |
But I guess your advice is to invest everything
link |
anyone listening owns into Dogecoin.
link |
That's what I heard from this interaction.
link |
Yeah, no, exactly.
link |
Yeah, our hedge fund is unavailable.
link |
Just go straight to Dogecoin.
link |
You're running a successful company.
link |
It's just interesting because my mind has been in that space
link |
of potentially just being one of the millions of other
link |
What's your advice on how to build a successful startup,
link |
how to build a successful company?
link |
I think that one thing I do like,
link |
and it might be a particular thing about America,
link |
but there is something about playing,
link |
tell people what you really want to happen in the world.
link |
It's not gonna make it,
link |
like if you're asking someone to invest in your company,
link |
don't say, I think maybe one day we might make
link |
a million dollars.
link |
When you actually believe something else,
link |
you actually believe you're actually more optimistic,
link |
but you're toning down your optimism because you want
link |
to appear like low risk.
link |
But actually it's super high risk if your company
link |
becomes mediocre because no one wants to work
link |
in a mediocre company.
link |
No one wants to invest in a mediocre company.
link |
So you should play the real game.
link |
And obviously this doesn't apply to all businesses,
link |
but if you play a venture backed startup kind of game,
link |
like play for keeps, play to win, go big.
link |
And it's very hard to do that.
link |
I've always feel like, yeah,
link |
you can start narrowing your focus because 10 people
link |
are telling you, you gotta care about this boring thing
link |
that won't matter five years from now.
link |
And you should push back and do the real,
link |
play the real game.
link |
So both, I mean, there's an interesting duality there.
link |
So there's the way you speak to other people
link |
about like your plans and what you are like privately
link |
just in your own mind.
link |
And maybe it's connected with what you were saying about,
link |
yeah, sincerity as well.
link |
Like if you appear to be sincerely optimistic
link |
about something that's big or crazy,
link |
it's putting yourself up to be kind of like
link |
ridiculed or something.
link |
And so if you say, my mission is to, yeah, go to Mars,
link |
it's just so bonkers that it's hard to say.
link |
It is, but one powerful thing, just like you said,
link |
is if you say it and you believe it,
link |
then actually amazing people come and work with you.
link |
It's not just skill, but the dreams.
link |
There's something about optimism that,
link |
like that fire that you have when you're optimistic
link |
of actually having the hope of building
link |
something totally cool, something totally new,
link |
that when those people get in a room together,
link |
like they can actually do it.
link |
Yeah, there's, yeah, that's,
link |
and also makes life really fun when you're in that room.
link |
So all of that together, ultimately,
link |
I don't know, that's what makes this crazy ride
link |
of a startup really look fun.
link |
And Elon is an example of a person who succeeded at that.
link |
There's not many other inspiring figures, which is sad.
link |
I used to be at Google and there's something that happens
link |
that sometimes when the company grows
link |
bigger and bigger and bigger,
link |
where that kind of ambition kind of quiets down a little bit.
link |
Google had this ambition, still does,
link |
of making the world's information accessible to everyone.
link |
And I remember, I don't know, that's beautiful.
link |
I still love that dream of that they used to scan books,
link |
but just in every way possible
link |
make the world's information accessible.
link |
Same with Wikipedia.
link |
Every time I open up Wikipedia,
link |
I'm just awe inspired by how awesome humans are, man.
link |
And creating this together,
link |
I don't know what the meanings are over there,
link |
but it's just beautiful.
link |
Like what they've created is incredible.
link |
And I'd love to be able to be part of something like that.
link |
And you're right, for that, you have to be bold.
link |
And there's, and strange to me also,
link |
I think you're right that there's
link |
how many boring companies there are.
link |
Something I always talk about, especially in FinTech,
link |
it's like, why am I excited about, this is so lame.
link |
Like what is, this isn't even important.
link |
Even if you succeed, this is gonna be like terrible.
link |
And it's just strange how people can kind of
link |
get fake enthusiastic about like boring ideas
link |
when there's so many bigger ideas that,
link |
yeah, I mean, you read these things,
link |
like this company raises money,
link |
and it's just like, that's a lot of money
link |
for the worst idea I've ever heard.
link |
Some ideas are really big.
link |
So like I worked on autonomous vehicles quite a bit.
link |
And there's so many ways in which you can present
link |
that idea to yourself, to the team you work with,
link |
to just, yeah, like to yourself when you're quietly
link |
looking in the mirror in the morning,
link |
that's really boring or really exciting.
link |
Like if you're really ambitious with autonomous vehicles,
link |
it changes the nature of like human robot interaction,
link |
it changes the nature of how we move.
link |
Forget money, forget all that stuff.
link |
It changes like everything about robotics and AI,
link |
machine learning, it changes everything about manufacture.
link |
I mean, cars, transportation is so fundamentally connected
link |
to cars, and if that changes,
link |
it's changing the fabric of society,
link |
of movies, of everything.
link |
And if you go bold and take risks
link |
and be willing to go bankrupt with your company,
link |
as opposed to cautiously, you can really change the world.
link |
And it's so sad for me to see all these autonomous companies,
link |
autonomous vehicle companies,
link |
they're like really more focused about fundraising
link |
and kind of like smoke and mirrors,
link |
they're really afraid,
link |
the entirety of their marketing is grounded in fear
link |
and presenting enough smoke to where they keep raising funds
link |
so they can cautiously use technology of a previous decade
link |
or previous two decades to kind of test vehicles
link |
here and there, as opposed to do crazy things
link |
and bold and go huge at scale, do huge data collection.
link |
And yeah, so that's just an example.
link |
Like the idea can be big,
link |
but if you don't allow yourself to take that idea
link |
and think really big with it,
link |
then you're not gonna make anything happen.
link |
Yeah, you're absolutely right in that.
link |
So you've been connected in your work
link |
with a bunch of amazing people.
link |
How much interaction do you have with investors,
link |
that whole process is an entire mystery to me.
link |
Is there some people that just have influence
link |
on the trajectory of your thinking completely,
link |
or is it just this collective energy behind the company?
link |
Yeah, I mean, I came here and I was amazed how,
link |
yeah, people would, I was only here for a few months
link |
and I met some incredible investors
link |
and I'd almost run out of money.
link |
And once they invested, I was like,
link |
I am not gonna let you down.
link |
And I was like, okay, I'm gonna send them
link |
like an email update every like three minutes.
link |
And then they don't care at all.
link |
So they kind of wanna, I don't know, like,
link |
so for some, I like it when it's just like,
link |
they're always available to talk,
link |
but a lot of building a business,
link |
especially a high tech business,
link |
there's a little for them to add, right?
link |
There's little for them to add on product.
link |
There's a lot for them to add on like business development.
link |
And if we are doing product research,
link |
which is for us research into the market,
link |
research into how to make a great hedge fund,
link |
and we do that for years,
link |
there's not much to tell the investors.
link |
So that basically is like, I believe in you.
link |
There's something, I like the cut of your jib.
link |
There's something in your idea, in your ambition,
link |
in your plans that I like.
link |
And it's almost like a pat on the back.
link |
It's like, go get them kid.
link |
Yeah, it is a bit like that.
link |
That's a good way to do it.
link |
I'm glad they do it that way.
link |
Like the one meeting I had,
link |
which was like really good with this
link |
was meeting Howard Morgan,
link |
who's actually a co founder of Renaissance Technologies
link |
in the like 1980s and worked with Jim Simons.
link |
And he was in the room
link |
and I was meeting some other guy and he was in the room
link |
and I was explaining how quantitative finance works.
link |
I was like, so they use mathematical models.
link |
And then he was like, yeah, I started Renaissance.
link |
I know a bit about this.
link |
And then I was like, oh my God.
link |
So yeah, but then, and then I think he kind of said, well,
link |
yeah, he said, well, cause I was talking,
link |
he was working at first round capital as a partner
link |
and they kind of said, they didn't want to invest.
link |
And then I wrote a blog post describing the idea
link |
and I was like, I really think you guys should invest.
link |
And then they end up.
link |
You convinced them on that.
link |
That must be good.
link |
Yeah, cause they're like,
link |
we don't really invest in hedge funds.
link |
And I was like, you don't see like what I'm doing.
link |
This is so a tech company, not a hedge fund, right?
link |
Yeah, and Numerai is brilliant.
link |
It's, when it caught my eye,
link |
there's something special there.
link |
So I really do hope you succeed in the,
link |
obviously it's a risky thing you're taking on,
link |
the ambition of it, the size of it,
link |
but I do hope you succeed.
link |
You mentioned Jim Simons.
link |
He comes up in another world of mine really often on the,
link |
he's just a brilliant guy on the mathematics side
link |
as a mathematician,
link |
but he's also brilliant finance hedge fund manager guy.
link |
Have you gotten a chance to interact with him at all?
link |
Have you learned anything from him on the math,
link |
on the finance, on the philosophy, life side, things?
link |
I've played poker with him.
link |
It was pretty cool.
link |
It was like, actually in the show, Billions,
link |
they kind of do a little thing about this poker tournament
link |
thing with all the hedge fund managers.
link |
And that's real life thing.
link |
And they have a lot of like world series of bracelet,
link |
world series poker bracelets holders,
link |
but it's kind of Jim's thing.
link |
And I met him there and yeah, it was kind of brief,
link |
but I was just like, he's like, oh, how do you,
link |
And I was like, oh, Howard sent me, you know,
link |
he's like, go play this tournament,
link |
meet some of the other players.
link |
Was it Texas Holdem?
link |
Yeah, Texas Holdem tournament, yeah.
link |
Do you play poker yourself or was it?
link |
I mean, it was crazy.
link |
On my right was the CEO,
link |
who's the current CEO of Renaissance, Peter Brown.
link |
And Peter Muller, who's a hedge fund manager at PDT.
link |
And yeah, I mean, it was just like,
link |
and then, you know, just everyone.
link |
And then all these bracelet world series,
link |
like people that I know from like TV.
link |
And Robert Mercer, who's fucking crazy.
link |
He's the guy who donated the most money to Trump.
link |
And he's just like...
link |
It's a lot of personality.
link |
Character, yeah, geez, it's crazy.
link |
So it's quite cool how, yeah, like the, it was really fun.
link |
And then I managed to knock out Peter Muller.
link |
I have a, I got a little trophy for knocking him out
link |
because he was a previous champion.
link |
In fact, I think he's won the most.
link |
I think he's won three times.
link |
But I will say Jim outlasted me in the tournament.
link |
And they're all extremely good at poker,
link |
but they're also, so it was a $10,000 buy in.
link |
And I was like, this is kind of expensive,
link |
but it all goes to charity, Jim's math charity.
link |
But then the way they play, they have like rebis
link |
and like they all do a shit ton of rebis
link |
because it's for charity.
link |
So immediately they're like going all in
link |
and I'm like, man, like, so I ended up adding more as well.
link |
So it's like you couldn't play at all without doing that.
link |
Yeah, the stakes are high.
link |
But you're connected to a lot of these folks
link |
that are kind of titans of just of economics
link |
and tech in general.
link |
Do you feel a burden from this?
link |
You're a young guy.
link |
I did feel a bit out of place there.
link |
The company was quite new
link |
and they also don't speak about things, right?
link |
So it's not like going to meet a famous rocket engineer
link |
who will tell you how to make a rocket.
link |
They do not want to tell you anything
link |
about how to make a hedge fund.
link |
It's like all secretive and that part I didn't like.
link |
And they were also kind of making fun of me a little bit.
link |
Like they would say, like they'd call me like,
link |
I don't know, the Bitcoin kid or.
link |
And then they would say, even things like,
link |
member Peter, yeah, said to me something like,
link |
I don't think AI is gonna have a big role in finance.
link |
And I was like, hearing this from the CEO of Renaissance
link |
was like weird to hear because I was like,
link |
of course it will.
link |
And he's like, but he can see,
link |
I can see it having a really big impact
link |
on things like self driving cars.
link |
But finance, it's too noisy and whatever.
link |
And so I don't think it's like the perfect application.
link |
And I was like, that was interesting to hear
link |
because it's like, and I think it was that same day
link |
that Libra, I think it is, the poker playing AI
link |
started to beat like the human.
link |
So it was kind of funny hearing them like say,
link |
oh, I'm not sure AI could ever attack that problem.
link |
And then that very day it's attacking the problem
link |
of the game we're playing.
link |
Well, there's a kind of a magic to somebody
link |
who's exceptionally successful looking at you,
link |
giving you respect, but also saying that what you're doing
link |
is not going to succeed in a sense.
link |
Like they're not really saying it,
link |
but I tend to believe from my interactions with people
link |
that it's a kind of prod to say like, prove me wrong.
link |
That's ultimately, that's how those guys talk.
link |
They see good talent and they're like.
link |
And I think they're also saying
link |
it's not gonna succeed quickly in some way.
link |
They're like, this is gonna take a long time
link |
and maybe that's good to know.
link |
And certainly AI in trading,
link |
that's one of the most philosophically interesting questions
link |
about artificial intelligence and the nature of money.
link |
Because it's like, how much can you extract
link |
in terms of patterns from all of these millions
link |
of humans interacting using this methodology of money?
link |
It's like one of the open questions
link |
in the artificial intelligence.
link |
In that sense, you converting into a data set
link |
is one of like the biggest gifts to the research community,
link |
to the whole, anyone who loves data science and AI,
link |
this is kind of fascinating.
link |
I'd love to see where this goes actually.
link |
I think I say sometimes long before AGI destroys the world,
link |
a narrow intelligence will win all the money
link |
in the stock market.
link |
Way, like just a narrow AI.
link |
And I don't know if I'm gonna be the one who invents that.
link |
So I'm building Numerai to make sure
link |
that that narrow AI uses our data.
link |
So you're giving a platform
link |
to where millions of people can participate
link |
and do build that narrow AI themselves.
link |
People love it when I ask this kind of question
link |
about books, about ideas and philosophers and so on.
link |
I was wondering if you had books or ideas,
link |
philosophers, thinkers that had an influence
link |
on your life when you were growing up
link |
or just today that you would recommend
link |
that people check out blog posts, podcasts, videos,
link |
all that kind of stuff.
link |
Is there something that just kind of had an impact on you
link |
that you couldn't recommend?
link |
A super kind of obvious one that I really,
link |
I was reading Zero to One while coming up with Numerai.
link |
It was like halfway through the book.
link |
And I really do like a lot of the ideas there.
link |
And it's also about kind of thinking big
link |
and also it's like peculiar little book.
link |
It's like why, like there's a little picture
link |
of the hipster versus Unabomber.
link |
And it's a weird little book.
link |
So I like, there's kind of like some depth there.
link |
In terms of a book on a, if you're thinking
link |
of doing a startup, that's a good book.
link |
A book I like a lot is maybe my favorite book
link |
is David Deutsch's Beginning of Infinity.
link |
I just found that so optimistic.
link |
It puts you, everything you read in science,
link |
it like makes the world feel like kind of colder
link |
because like it's like we're just coming from evolution
link |
and coming from nothing should be this way or whatever.
link |
And humans are not very powerful.
link |
We're just like scum on the earth.
link |
And the way David Deutsch sees things
link |
and argues, he argues them with the same rigor
link |
that the cynics often use
link |
and then has a much better conclusion.
link |
That's some of the statements of things like,
link |
anything that doesn't violate the laws
link |
of physics can be solved.
link |
So ultimately arriving at a hopeful,
link |
like a hopeful path forward.
link |
Yeah, without being like a hippie.
link |
You've mentioned kind of advice for startups.
link |
Is there, in general, whether you do a startup or not,
link |
do you have advice for young people today?
link |
You're like an example of somebody
link |
who's paved their own path
link |
and were, I would say exceptionally successful.
link |
Is there advice, somebody who's like 20 today, 18,
link |
undergrad or thinking about going to college,
link |
or in college and so on that you would give them?
link |
I think I often tell young people don't start companies.
link |
Is it not, don't start a company
link |
unless you're prepared to make it your life's work.
link |
Like that's a really good way of putting it.
link |
And a lot of people think, well, this semester
link |
I'm gonna take a semester off.
link |
And in that one semester,
link |
I'm gonna start a company and sell it or whatever.
link |
And it's just like, what are you talking about?
link |
It doesn't really work that way.
link |
You should be like super into the idea,
link |
so into it that you wanna spend a really long time on it.
link |
Is that more about psychology or actual time allocation?
link |
Like, is it literally the fact that you need to give 100%
link |
for potentially years for it to succeed?
link |
Or is it more about just the mindset that's required?
link |
Yeah, I mean, I think, well, any, I think, yeah,
link |
you don't wanna have,
link |
certainly don't wanna have a plan to sell the company
link |
like quickly or something,
link |
or it's like a company that has a very,
link |
it's like a big fashion component.
link |
Like it'll only work now.
link |
It's like an app or something.
link |
So yeah, that's a big one.
link |
And then I also think something I've thought about recently
link |
is I had a job as a quant at a fund
link |
for about two and a half years.
link |
And part of me thinks if I had spent
link |
another two years there,
link |
I would have learned a lot more
link |
and had even more knowledge to be where,
link |
to basically accelerate how long Numerai took.
link |
So the idea that you can sit in an air conditioned room
link |
and get free food,
link |
or even sit at home now in your underwear
link |
and make a huge amount of money and learn whatever you want
link |
and get, it's just crazy.
link |
It's such a good deal.
link |
Yeah, oh, that's interesting.
link |
That's the case for, I was terrified of that.
link |
Like at Google, I thought I would become really comfortable
link |
in that air conditioned room.
link |
And that, I was afraid the quant situation is,
link |
I mean, what you present is really brilliant
link |
that it's exceptionally valuable, the lessons you learn,
link |
because you get to get paid while you learn from others.
link |
If you see that, if you see jobs
link |
in the space of your passion that way,
link |
that it's just an education.
link |
It's like the best kind of education.
link |
But of course you have, from my perspective,
link |
you have to be really careful on that to get comfortable.
link |
Again, in a relationship, then you buy a house
link |
or whatever the hell it is, and then you get,
link |
and then you convince yourself like,
link |
well, I have to pay these fees for the car,
link |
for the house, blah, blah, blah.
link |
And then there's momentum and all of a sudden
link |
you're on your death bed and there's grandchildren
link |
and you're drinking whiskey
link |
and complaining about kids these days.
link |
So I'm afraid of that momentum, but you're right.
link |
Like there's something special about the education
link |
you get working at these companies.
link |
Yeah, and I remember on my desk,
link |
I had like a bunch of papers on quant finance,
link |
a bunch of papers on optimization,
link |
and then the paper on Ethereum, just on my desk as well,
link |
and the white paper, and it's like,
link |
it's been amazing how kind of, and you can learn about,
link |
so that, I also thought, I think this idea
link |
of like learning about intersections of things,
link |
I don't think there are too many people that know
link |
like as much about crypto and quant finance
link |
and machine learning as I do.
link |
And that's a really nice set of three things
link |
to know stuff about.
link |
And that was because I had like free time in my job.
link |
Okay, let me ask the perfectly impractical,
link |
but the most important question.
link |
What's the meaning of all the things you're trying to do
link |
so many amazing things, why?
link |
What's the meaning of this life of yours or ours?
link |
Yeah, so I have yet had some people say,
link |
asking what meaning of life is,
link |
is like asking the wrong question or something.
link |
The question is wrong.
link |
No, usually people get too nervous to be able to say that
link |
because it's like your question sucks.
link |
I don't think there's an answer.
link |
It's like the searching for it.
link |
It's like sometimes asking it.
link |
It's like sometimes sitting back and looking up at the stars
link |
and being like, huh, I wonder if there's aliens up there.
link |
There's a useful like a palette cleanser aspect to it
link |
because it kind of wakes you up to like all the little busy,
link |
hurried day to day activities, all the meetings,
link |
all the things you'd like a part of.
link |
We're just like ants, a part of a system,
link |
a part of another system.
link |
And then asking this bigger question
link |
allows you to kind of zoom out and think about it.
link |
But there's ultimately,
link |
I think it's an impossible thing for a limited capacity,
link |
like cognitive capacity to capture.
link |
But it's fun to listen to somebody
link |
who's exceptionally successful, exceptionally busy now,
link |
who's also young like you,
link |
to ask these kinds of questions about like death.
link |
You know, do you consider your own mortality kind of thing
link |
and life, whether that enters your mind?
link |
Because it often doesn't,
link |
it kind of almost gets in the way.
link |
It's amazing how many things you can like that are trivial
link |
that could occupy a lot of your mind
link |
until something bad happens or something flips you.
link |
And then you start thinking about the people you love
link |
that are in your life.
link |
Then you started thinking about like,
link |
holy shit, this ride ends.
link |
Yeah, I just had COVID and I had it quite bad.
link |
What wasn't really bad is just like,
link |
I also got a simultaneous like lung infection.
link |
So I had like almost like bronchitis or whatever.
link |
I don't even, I don't understand that stuff,
link |
but I started, and then you're forced to be isolated.
link |
And so it's actually kind of nice
link |
because it's very depressing.
link |
And then I've heard stories of, I think it's Sean Parker.
link |
He had like all these diseases as a child
link |
and he had to like just stay in bed for years.
link |
And then he like made Napster.
link |
It's like pretty cool.
link |
So yeah, I had about 15 days of this recently,
link |
And it feels like it did shock me
link |
into a new kind of energy and ambition.
link |
Were there moments when you were just like terrified
link |
at the combination of loneliness?
link |
And like, you know, the thing about COVID is like,
link |
there's some degree of uncertainty.
link |
Like it feels like it's a new thing, a new monster
link |
that's arrived on this earth.
link |
And so, you know, dealing with it alone,
link |
a lot of people are dying.
link |
It's like wondering like.
link |
Yeah, you do wonder, I mean, for sure.
link |
And then there are the even new strains in South Africa,
link |
which is where I was.
link |
And maybe the new strain had some interaction
link |
with my genes and I'm just gonna die.
link |
But ultimately it was liberating somehow.
link |
Oh, I loved that I got out of it.
link |
Because it also affects your mind.
link |
You get confused, you get confusion
link |
and kind of a lot of fatigue
link |
and you can't do your usual tricks
link |
of psyching yourself out of it.
link |
So, you know, sometimes it's like, oh man, I feel tired.
link |
Okay, I'm just gonna go have coffee and then I'll be fine.
link |
It's like, now it's like, I feel tired.
link |
I don't even wanna get out of bed to get coffee
link |
because I feel so tired.
link |
And then you have to confront,
link |
there's no like quick fix cure and you're trapped at home.
link |
But that, so now you have this little thing
link |
that happened to you that was a reminder
link |
that you're mortal and you get to carry that flag
link |
in trying to create something special in this world.
link |
Listen, this was like one of my favorite conversation
link |
because the way you think about this world of money
link |
and just this world in general is so clear
link |
and you're able to explain it so eloquently.
link |
Richard, that was really fun.
link |
Really appreciate you talking to me.
link |
Thanks for listening to this conversation with Richard Crave
link |
and thank you to our sponsors,
link |
Audible Audio Books, Trial Labs,
link |
Machine Learning Company, Blinkist app
link |
that summarizes books and Athletic Greens
link |
all in one nutrition drink.
link |
Click the sponsor links to get a discount
link |
and to support this podcast.
link |
And now let me leave you with some words from Warren Buffett.
link |
Games are won by players who focus on the playing field,
link |
not by those whose eyes are glued to the scoreboard.
link |
Thank you for listening and hope to see you next time.