Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:02):
Bloomberg Audio Studios, podcasts, radio news. Wait, Gapy Palio logo or.
Speaker 2 (00:14):
I'm so excited for you to tackle that and not me.
Speaker 3 (00:16):
Good luck.
Speaker 2 (00:17):
I thought I had it down, but then I heard
you say.
Speaker 1 (00:19):
It, and I feel like when I first met you,
I think I asked you if you go by Gappy
because of your famous track record of taking gardening leave,
like having gaps in your career.
Speaker 3 (00:35):
Oh okay, I didn't you remember that. Yeah, that's a
great excuse for for a nickname. No, but the reason
is when I came to the States for grad school,
and this was a long time ago, in ninety five.
So the first thing that you did was set up
an email account. You still have the freedom to choose
(00:56):
an email account. Now, they just give you your initials with
the number, and so my initials are gap Gap, and
of course it was taken. So I said, okay, well Gappy,
and then everybody in grad school, and then my wife
was Italian, everybody started to call me Gappy and that's stuck.
(01:17):
And now at work they just have dispensed with my
real name, like on all systems, I'm just Gappy Paliologo.
So I expect that that will be you know, prosecuted
for tax evasion because on my tax forms there is
Gappy Palaiologo or something like that.
Speaker 1 (01:33):
Well, hello, and welcome to the Artistuff Podcast.
Speaker 2 (01:36):
I'm Matt Livian and I'm Katie Greifeld.
Speaker 1 (01:39):
And we have a guest today, Gabby pale Oligo, who
is now at pali Asne and has been at most
of the other couch funds and Hudson River Trading. I
do want to start by talking.
Speaker 3 (01:53):
About gardening me, okay, natural.
Speaker 1 (01:57):
I think that we counted for your link Your LinkedIn
is like famous for discussing your gardening live in some detail,
and I think we counted three years of gardening. Leave No,
I think it's a bit okay, it's not precise. Fifteen
months from Citadel one you're Hudson River Trading and four
months from millennium. Okay, so pretty close.
Speaker 3 (02:19):
Not terrible, though a bit less than two years.
Speaker 1 (02:22):
From my perspective, it seems very fun. Did you enjoy
your three years of gardening?
Speaker 3 (02:28):
I do so. I try to keep myself busy, so
I teach typically at some university. So the first time
during my Seitaadel two Millennium, Guardian leve I was teaching
at Cornell and in the HRT to Bam Guardian Leve,
I was at NYU and I love teaching, and then
(02:49):
what I do is it helps me focus on on stuff.
Usually what I do in you know, whenever I read
a book or paper that I like, I take notes.
I take notes in Lattech and then I really arrive
or think about things, and so that typically is the
(03:09):
basis for my course material, and then it becomes the
basis for my books. I've written a couple of books
during my noncompetes.
Speaker 2 (03:17):
Interesting because thinking about gardening leave. Matt and I talk
about it all the time because it's very alluring to me.
Gardening leave doesn't really exist in journalism. I love to
imagine what I would do. But one of the questions
I had for you was, you know, do you ever
have anxiety about losing your edge or falling behind? But
it sounds like teaching is one of the ways.
Speaker 3 (03:37):
That particularly worried with that. I think that there is
only a very specific subset of quantitative researchers who are
afraid of losing their edges. And yeah, that's not been
my case. I keep reading, I try to stay up
to date.
Speaker 1 (03:54):
Feedback into the work like, do you get ideas or
like deep in your understanding of techniques by teaching and
writing the books? Or are they just sort of like.
Speaker 3 (04:03):
Extracurricular No, no, No, it's definitely I learn a lot
from writing the books.
Speaker 1 (04:08):
How long do you I go to your next job
and generate more profits by of.
Speaker 3 (04:13):
Course plenty more profits. Tell that to my employers. No,
but I definitely I learn a lot from writing, from
the first drafts, and then I rewrite and rewrite, and
I learn a lot from discarding material too. It's very
useful to discard material. It makes you really focus on
what matters and what doesn't. So I try to give
(04:34):
a narrative, like a logical connection between various topics, and
that is something that is possible only when you write
a book. I really do not like writing. That Nobody
I think likes writing, maybe except for you.
Speaker 1 (04:48):
I understand that it's weird even among writers, but it is.
Speaker 3 (04:54):
Very I find it very painful. I find painful letting
go of material, Yes, but I also like it. You know,
it's some kind of strange delayed gratification.
Speaker 1 (05:05):
I guess one theory that I have written is that
Hedge fund and quantitative research gardening leave is like a
source of like human flourishing because you have all these
like highly trained people who haven't enforced a year of
And I've written that all the hedge fund researchers should
go work at LM companies or like analytics departments of
(05:27):
sports teams, and I'm like, partially kidding and partially not.
How true is it for you? Like how much of
like your quantitative skills at this point are really just
for investing, and how much of it is like if
you spent three months, you know, consulting for a soccer team,
(05:50):
you would be able to tell them how to find
better players.
Speaker 3 (05:54):
I'm not sure, so I'll say this right. I was
thinking a few days ago if there was a kind
of a common thread in my professional life, because it
seems kind of random, And actually I think that there is,
because I think that I was about fourteen when I
realized that I had an aptitude for applied math. I
(06:16):
discovered physics, and I liked math, and I also liked
literature very much, so I loved reading. I read a lot.
I was not a very social animal. And then basically
since then, I've been doing the same thing in various forms. Right,
I did physics, I did applied math. I didn't do
(06:37):
applied math in finance. I did applied math in weird
things like optimization and logistics. So I have been doing
kind of the same thing over and over, which has
been writing and applying math to something. So I think
that I could do it. I would like to do it,
but I also think that it's not that simple to
(06:58):
go to a new field and oh, after three months,
I know soccer. No, there is a lot of specificity.
And the beauty of I think being a good applied
mathematician is that they start with the problems and with
the domain first, and that they're sufficiently mature from a
mathematical standpoint that they are not making too much of
(07:20):
an effort in using math. So I think the good
art of being an applied mathematician is to study persistently
the application. So no, I don't think that after three
months it would be good enough. But after a year,
you know, about a year of being fully immersed in
an application, then you start getting a little bit better,
(07:41):
and then the math is not the problem, and then
you start doing some good work.
Speaker 1 (07:46):
You have a famous essay on like advice for quant
careers and you say that like the things that matter
the most creativity and genuine interest in the problems more
than you know, math, coorse power. Yeah, this is a
dumb question. But how does one develop how does one identify,
(08:07):
you know, creativity and interest in financial topics? And is
the obvious answer those are where the money is? Or
like like why why did you fall in love with
finance as a topic, And is the answer because that's what.
Speaker 3 (08:19):
The money is. So first of all, I think that
creativity is either a personality trait doesn't belong to You're
not creative in finance, you know, you're you're creative in
in cooking, you're creative in whatever. And it's a mix
I guess of extraversion opening and as to experience, and
I don't know what else. I'm not a psychologist, but
I do believe that people are genuinely creative. And in fact,
(08:43):
you see it right that sometimes you ask someone and
you find out that yes, they like writing, they play
some instrument if badly, and you know, and they paint
and they do whatever. And so I would say, if
you go to finance because where the money is, there's
nothing wrong with that. And in a way, that's my
(09:03):
story you know, I was, I was a researcher. I
wanted to have more money and whatnot. But eventually you
stay in finance, or at least in my you know,
little domain, because you're genuinely curious about finding out stuff, right, So.
Speaker 1 (09:20):
Like why are the problems like why do they arouse curios?
Like why are the problems of finance intrigue you after
years of doing it? Right? Like what's interesting about those
problems as opposed to other domains.
Speaker 3 (09:32):
It's really hard for me to say, Like I think
that I read once that a young songwriter asked Bob
Dylan how to become a good songwriter, and Bob Dylan
just answered, well, what's going on? I said, what do
you mean, what's going on? Yeah, what's going on? What's
going on in your life? Just you know, look around.
So sometimes I get these questions from investors, But you know,
(09:53):
how do you keep yourself interested? How do you find problems?
It's not a problem like the problems jump at you
like there are too many problems. There are too many
interesting problems. So if anything, the skill is in sorting
the problems in the right order, right. That is where
maybe having some maturity in doing research kicks in. But
(10:14):
there are lots of problems, infinite problems, weird problems.
Speaker 1 (10:18):
What's your favorite problem right now?
Speaker 3 (10:21):
I don't like right now? What are we working on?
I mean, we are trying to understand how earnings are monetized? Right?
How do you make money in earnings? It's such a
basic thing in fundamental equities.
Speaker 1 (10:32):
And you mean, if you're like correct about predicting.
Speaker 3 (10:35):
Earnings, yes, what are I mean without getting too much
into details, but you know there what are the relevant variables?
Imagine that you had an oracle who told you what
the variables are? What would you do with that? What
would you do if you'd had all the information in
the world right and everything in your world here in
existence would be like an approximation problem.
Speaker 1 (10:56):
There's an incredible STYLI story of like the guys into
I think, like one of the newswire services and got
earnings releases early like for hundreds of companies, and they
traded on this and they had like a seventy percent
success rate, which is great, but also like it means
that had a thirty percent, like they traded the wrong way,
knowing earnings perfectly in advance. It's like a good yeah, yes,
(11:19):
so I had the you know, it's still hard.
Speaker 3 (11:22):
Yes, it's still very hard. Actually, shout out to Victor Hagan,
who wrote the paper about ten years ago on this.
He made a organize a simple controlled experiment where he
gave basically a biased coin where you, I think had
a success rate of sixty percent forty percent failure, and
you had some capital and you could invest it over
time on these informed predictions, and a lot of subjects
(11:44):
went bankrupt. Okay, now I think we are better than that,
but still there are lots of problems related to trading
around an event.
Speaker 2 (11:53):
For example, before we get too far away, you mentioned
Bob Dylan. It actually reminded me of another Dylan quote
which I'm going to paraphrase poorly, but he basically said,
when asked about writing songs, do you think that you
could write whatever the work that was being referenced now,
And he said, I don't think so. It's like the
words were in the air and I just plucked them out.
(12:15):
They were just sort of hanging in the air and
they came to me. And it kind of also rang
true with what you were saying about you didn't go
looking for problems, They're just there. Necessarily. I actually want
to go back to applied math if it doesn't interrupt
the course of conversation too much. You tweeted on June
twenty fourth that there's no child prodigies when it comes
to poetry, when it comes to applied mathematics. And I'm
(12:38):
not saying that. You said that you were a prodigy,
but you were a child at fourteen. I mean, how
at fourteen do you realize that you have an aptitude
for something like applied mathematics?
Speaker 3 (12:50):
All right, I don't want to flex about this stuff, No,
you should. I think I'm honestly a little weird. I'm
just a little weird, I think, honestly, but I.
Speaker 2 (12:58):
Like prodigy weird or.
Speaker 3 (13:00):
I did have my share of yeah, adults telling me
that I was good at this or that or you know.
But yeah, I mean, okay, I'm just a little bit atypical. Also,
when I talk to investors, I think investors enjoy my
presence because I think I'm incredibly unfiltered for somebody who's
talking to them, so it's like fun for them. And
(13:23):
I was very unfiltered when I talk to my professors
in school. Sometimes I corrected them stuff like this, Yeah,
I don't know, honestly, I don't know.
Speaker 1 (13:34):
When you talk to like fundamental equity portfolio managers, like
how much like matrix algebras they're in your conversations like
how quantity are the fundamental pms or whatever.
Speaker 3 (13:47):
I don't think they're quantity, but I think that they're
very analytical. So I don't think that they would make
great mathematicians, but I think they would make very very
decent applied mathematicians. Actually, they tend to be very analytical.
They tend to be very process oriented. And they have
also additional qualities that actually mentioned in that essay, like
(14:09):
they have very little disposition effect, so that's part of
being analytical. They have no sound cost fallacy in them.
So even though they don't do a lot of math,
but they do some math. Okay, So first of all,
they're fluent in a sense in basic literacy. But I
think it's more their process that is closer to if
not a mathematical one, but more of a scientific one.
Speaker 2 (14:32):
And when it comes to being a quant does it
basically boil down to being good at math and being
interested in math? Are things such as statistics and physics?
I mean, do you need to have any finance or
economics background at all.
Speaker 3 (14:49):
So I think that having an economics background is not
necessarily a benefit, might even be a disadvantage actually, But
just based on very few samples that I have a
lot of very good, outstanding quantitative researchers actually come from
physics and specifically from astrophysics. That's the experience that I've
(15:10):
had in a couple of places.
Speaker 2 (15:11):
In broad brushstrokes, could you talk about why economics in
the small sample size you have, how could that possibly
be a detriment good?
Speaker 3 (15:22):
So I can answer the second question more easily. I
think that astrophysicists deal with large amounts of data, and
they deal with observational data, so they don't get to
do a lot of experiments. And that's good for finance, right,
you deal with a lot of data. You need to
know how to have good agen for observational data, and
(15:43):
you need to have very good theory, like you need
to have very good instruments without being falling in love
with those instruments. Whereas I think economists, Okay, first of all,
my statement is purely empirical. Okay, So I'm just really
guessing on a economists and I'm going to be hated
by all economists or economists in finance, but I do
(16:07):
have my issues with their methods.
Speaker 1 (16:10):
Right.
Speaker 3 (16:10):
So, first of all, I think that there is an
original scene in economics, which is I think a lot
of economics is informed by a desire to be as
rigorous as mathematics. Right, And so a lot of theoreticians
in economics are very deductive in their approach. If you
think of you know, the unrealistic assumptions behind the welfare
(16:33):
theorems or arrows impossibility theorem or whatnot, or just pick
up you know Samuelson textbooks, and I think this is
just axiomatic rather very axiomatic, very deductive. Whereas physicists are
very happy to think in terms of small idealized models
(16:53):
that apply to a specific domain, and if the model
doesn't work out, they will discard and make another one.
The grand theory behind physical theories exists, like there are
people who do this for a living. But many, many
good theoretical economists physicists starting the small and then they
expanded domain of their models. So economists tend to maybe
(17:16):
in a sense, fall in love with methods too much,
with techniques too much.
Speaker 1 (17:32):
We had cliff Astness on the podcast a little while ago,
and my father, not a finance person, listened to the
episode and said, I still don't know what a quant is.
I just read skimmed your new book which is called
The Elements of Quantitative Investing, and as lays out the elements,
what is a quant like? What are the elements? Like?
(17:52):
What's the thing that makes someone a quant investor that,
like someone reading a slim book about the Elements of
quant investing needs to learn?
Speaker 3 (18:01):
Well, if I am being consistent with my book, investing
is really about problems and not about specific techniques or
anything like this, Right, So it's basically a way to
go through the whole investment process from let's say preparing
the ingredients to cooking to eating. That is very process driven. Ultimately,
(18:21):
you would imagine that one thing that you know, quant
investing has in common across multiple domains, you know, if
you do futures, stocks, event based and whatnot, is I
think the number of bets tends to be high in
systematic investing. Right, so you can be a very successful
(18:41):
microeconomic investor portfolio manager. And you you know, according to
even several statements by Buffett, you know, he made like
ten twelve very good bets. Okay, so that's great, and
that's not quant investing. You know, you could put enough
pms making you know, twenty bets in their lives, you
will get a few that have let's say twelve thirteen, right,
(19:03):
and they will be rich. We do not have that luxury, right.
We have to make millions of bets. You know, we
trade a portfolio with three thousand stocks sometimes in waves
of half an hour. You can't make a judgment on
all of these bets. So you need a method that
reduces the dimension of your problem to something that can
(19:23):
be treated in a systematic manner. I don't know if
that answers for you. You know that, but you know, basically,
basically the idea is, think about if you make a
lot of bets, you cannot bet individually. You have to
have some kind of juristic or some kind of method
around that.
Speaker 1 (19:39):
Right, and like to me, like the book sort of
you know, the standard method I guess I'm quite investing
is you build a factor model of what drives your
universal investments. You're shutting your.
Speaker 3 (19:50):
Head, Yeah, I yes, and no, I think yes because
the book, you know, has maybe one hundred and fifty
pages on factor models, but also no, because maybe in
one hundred years from now, I suspect there will be
still something left. But you know, we might have better
techniques and not necessary factor models any longer.
Speaker 1 (20:09):
I don't know, we don't want to go. Two attractions
of that, one is like, are the better techniques something
more neural netty unstructured?
Speaker 3 (20:17):
Who knows? Yeah, something like that. I mean there is,
there is a revolution every five years.
Speaker 1 (20:22):
So my other question is like I've never fully understood
like a factor model is like the here are some
factors that drive the returns of stocks, and then there's
like some residual idiots and credit return. There are clearly
people whose business is to identify factors and then invest
(20:46):
in factors. My impression is that at like the places
that you work, the business is the opposite of that
is to hedge out your factor risk as much as
possible and to get as much idiosyncratic risk as possible.
Is that right? And like, like, how do you discriminate
between like a factory return and idiosymcratic return, Like what
makes the thing a factor as opposed to another ring.
Speaker 3 (21:08):
So that's a good question. So first, a lot of
systematic investing is still about factors, just not the factors
that get published in the literature, you know, not the
factors that Cliff maybe was talking about. And yet a
lot of successful systematic investing is really factor driven.
Speaker 1 (21:25):
It in the sense that you have a model that
has like twenty factors and like ten are like value,
and you neutralize those and you try the other time.
Speaker 3 (21:34):
You do, and you do the rest. You have other
terms that matter. So that's one thing, But there are
two other things. There are sometimes sources of returns that
are factor like but not quite like factors. So you
may have a theme. For example, you may identify a
theme in the market that is not pervasive enough or
(21:57):
is alive only for a few months, but is there
and it's not only affecting let's say two stocks. Right,
So these brought thems can be invested on, but cannot
really model in the traditional way as a traditional factor model. Also,
there is a lot of good modeling in factors as
opposed to bad modeling, So it's it seems easy, but
(22:18):
it's not that easy. So there is a little bit
of craftsmanship in making these models, okay, And then the
third thing is that there are also returns that have
nothing to do with factors, or almost nothing to do
with factors. So if you really know how a company works,
and you have a little bit of an edge in
predicting its future performance, you can bet on it, and
(22:39):
you make enough bets and again you will make some
money if you repeat, and you know recycle. So even
discretionary investing in this sense has inherited a little bit
of the spirit of systematic investing.
Speaker 1 (22:53):
I think of that as like that a pod job,
but like aval like you have discretionary investors who know
a lot about a company make bets on the company,
and then someone like you tells them, these are your
factory exposers. You have to get those down to zero.
That you're making pure bets on your idios and creditnoledge
of the company. Is that like kind of right?
Speaker 3 (23:14):
Kind of right?
Speaker 1 (23:14):
Yeah?
Speaker 3 (23:15):
I think that at this point it is very interesting
how the mind of professional portfolio managers has been remolded
in a factor based world, so that a modern portfolio
manager discretion The portfolio manager thinks in factors, you know,
so I don't even need to tell them, hey, this
(23:36):
is your exposure. They see their exposure, they have the
tools to see it, and they control it in real
time with minimal intervention from me. So what we do
is we have you know, a good team that models
factors in a way that is suitable for the investment
universe and style in which they operate. That's a very
very sophisticated and difficult and portfolio managers use that and
(24:00):
then neutralize It's become like second.
Speaker 1 (24:02):
Nature, and they've internalized that their goal is to create
idiosyncratic alpha rather than factors. That's right. I feel like
a criticism that people sometimes have of like the pod
shop model is that, like there's some universe of factors
that exist in commercial models and the like are known
in the literature, and then portfolio managers have a set
of exposures to factors that are sort of in code
(24:25):
or unknown, but like ultimately, when you become really, really smart,
you'll know that, like, actually the bet they were making
was some you know particular knowing the company really well
means like they had exposure to like some you know,
personality factor in the CEO or something that like eventually
someone will be able to write that down and it'll
come out of like being idiosyncratic and become a factor.
(24:46):
And then I don't know, what happens.
Speaker 3 (24:48):
I think that there is some truth to that. There
is definitely some some truth to that, in the sense
that sometimes for folio managers, especially in specific sectors, will
use some heuristics that you could call characteristics in a
factor model, but they are not in a factor model,
and then they trade that. However, it's also true that
(25:10):
the decision that enters a particular investment is usually not
that simple as taking a ration spreadsheet, so it's a
bit more complicated than that. You could still argue that
there is a factor, right, And what's the factor is
ultimately the set of feces that are highly correlated or
relatively highly correlated across portfolio managers across firms, because if
(25:34):
there is an expected return, and if you have skill,
and you have sufficient skill to be close to the
best possible portfolio, you have to be also relatively close
to other people approximating that best possible portfolio. Right, So
then it becomes a truism, Right, there is a factor,
and that's the factor of investor, of informed investors. So
(25:55):
it's true.
Speaker 1 (25:56):
I think if it is like there's like a scientific
process that ever pursuing I hear are the best people
and they like do the best work to pursue that
scientific process, and so they'll eventually converge on something that
is like truth. But that means buying all the same stocks.
Speaker 3 (26:12):
Yes, it's very difficult to get to that truth.
Speaker 1 (26:14):
Sure it is. Yeah it's not.
Speaker 3 (26:20):
Let's let's tire about it.
Speaker 1 (26:21):
Weird if they weren't hurting among.
Speaker 3 (26:22):
The best, yes, yes, but there is, there is And
by the way, and this brings to one of the
limitations of factor models, right, which is effectively a factor
model is a form of glorified regression over time. Right,
And behind a regression there is a bit of an
assumption to some extent, of independent observations over time. And
(26:45):
the market and hedge funds are not in dependent random variables.
They are super dependent random variables, and they are in
a sort of continuous in direct conversation through their portfolios.
And sometimes the conversation gets really nasty when one hedge
fun is in state of distress and all of a sudden,
or not even a hatch fund, it could be also
(27:05):
an institution investor and decide to liquidate part of their portfolio.
And then it becomes a process where you have a
lot of reflexivity and positive feedback and everybody suffers. And
in this case, factor models don't really You can still identify,
like if the system is running a temperature with some characteristics,
(27:25):
but they are not factors in the traditional sense. I
do want to.
Speaker 2 (27:29):
Talk about before we move too far away, I do
want to talk a little bit about how and if
factors can die, because we've talked a bit about identifying factors.
But when do you decide that this doesn't work anymore?
Necessarily that the market has fundamentally changed and this worked
maybe ten years ago, maybe fifteen years ago, but maybe
(27:52):
now it's devolved.
Speaker 3 (27:56):
Well, there is the good old reason, which is people
make mistakes in the sense that we think that there
is a factor and then we look back and there
is no factor. Right, So there are so many factors
that some of them have got to be a little
bit redundant. So that that's one reason, right, So just
pure in a sense research revisions. And then there is
(28:21):
also the fact that there are two other things that
can happen. One is the moment that you tell people
that there is a factor, the factor comes into being
to some extent, right, So it's never black and white
that the factor did not exist. Maybe the factor did
exist and then the moment you identify it, it becomes
more existent, like you know, speak yeah, yeah. So esg
(28:46):
is is one case where the focal point that it
became makes into an investible theme.
Speaker 2 (28:52):
I thought that was just black rock pumping.
Speaker 3 (28:55):
As possible, but everybody had to incorporate it in some sense, right,
so it became a major source of revenue for the vendors. Right.
So that's that's one thing. And then there is the
adaptive nature of the market. So things that before generated
a priced return. So you run some risk, you made
some money, and then it becomes table stakes, it becomes
(29:19):
incorporated into factor models, it becomes it becomes a smart beet,
it becomes a smart patent, and then it becomes so
I think, you know, you could say definitely that medium
tonmamentum worked much better. You could say that even you know,
short term reversal worked better. There were years when short
interest was great, and there are factors or data sources
(29:42):
that work well now and then maybe in five years
will become known and become part of the I mean
credit card data. Right for consumer that was like there
were people who were making a lot of money in
two thousand and eleven through sixteen seventeen, and then it's
become it's very hard to make money in that.
Speaker 1 (30:02):
You said the market is a conversation among catch funds.
One thing that I think might be true that I'm
not entirely sure of, is like, to what extent the
market is a conversation among four patch funds? Now? Like,
to what extent is like the marginal price or of
every stock a portfolio manager at you know, one of
the places you've worked.
Speaker 3 (30:22):
It's a very good question. I don't really have the
answers to this. I'm not sure.
Speaker 1 (30:26):
It's it's like, what is the intuition at places like that, Like,
is it like the market price is determined by like
the collective thought of like the top people at the
top hedge funds, Or is it like we are a
little bump on the market and we're trading against the
whole random universe.
Speaker 3 (30:42):
I mean, you'd like to think that the prices are
determined by the marginal informed investor, Right, so by people
like us at the time horizon where we predict right,
which is not the same as at the time of
Rizon of half a day. Right, that's a different player.
Speaker 1 (30:57):
What is your time horizon, like I think of it as.
Speaker 3 (30:59):
Well, it depends well, yes, it depends. Within a hedge fund,
you have a variety of even within long shore equities,
you know, you have you know, portfolio managers who are
very tactical, and so they think in terms of they
have strong daily or intra day alpha, even though they're
fully discretionary up to pms that think easily in terms
of months. Also depends on the sector. So you know,
(31:22):
financials typically probably monetizes a little bit less on earnings
and tends to have a longer horizon. Banks are basically
modeling giant balance sheets, right, and then in a hedge
fund you also have systematic, but even in systematic there
are all sorts of time scales, and this cacophony makes
the prices. I really don't know, Like I said, another
(31:43):
question is basically, are how inefficient is the market? How
incorrect are the prices are within a factor of two?
Like Black used to say, or I don't know, Like
I don't think that the market is becoming so super efficient,
but it's getting it seems to be more efficient.
Speaker 1 (31:58):
I do feel, like, you know, the big stories is
the rise of like these big multi strategy hedge funds,
like you would hope. Maybe you wouldn't hope because it's
sort of the economic and just, but like one might
hope that like the rise of these big multi strategy
hedge funds and a lot of capital being allocated to
them would observably make the market more efficient.
Speaker 3 (32:20):
Yeah, I don't know if observably holds. I don't. It's
really hard to like, can you can you tell when
a bubble is forming?
Speaker 2 (32:31):
A lot of people would say that they can.
Speaker 3 (32:34):
Yeah, I can point you to a few papers, yeah
that you know made all the wrong calls. Okay, I
don't want to shame academics in public.
Speaker 2 (32:45):
I do like the idea that the market is a
conversation between four hedge funds because I live in the
ETF world, and you know, the big thing is passive
is just distorting the market and there's no price discovery anymore.
And it sounds like that's on the opposite end of
that spectrum.
Speaker 3 (33:02):
I didn't say, I think exactly that it's a conversation
between It's a beautiful thing to say, though. It sounds
really cool. It sounds good podcasts. Yeah, that's great. Yeah,
But I think your question is whether the rise of
passive has made markets less efficient.
Speaker 2 (33:21):
More of a statement. I don't think I was a
bad podcaster and didn't actually.
Speaker 3 (33:24):
Ask a question, But okay, how do you know?
Speaker 2 (33:28):
How do I know that passive is the story in
the market? People on Twitter tell me?
Speaker 3 (33:32):
So, oh, okay, don't trust people on Twitter.
Speaker 2 (33:37):
That's true. Number one, real number one.
Speaker 3 (33:39):
Now I don't know. I mean, the rise of passive
has made index rebalancing a weirder strategy, right, so where
the margins have compressed, but the size has become so
big that you can still make money in it and periodic.
It's a very you know, cyclical strategy. So I don't know.
Speaker 1 (33:58):
So you're an indexy balancing PM do take like eight
months of vacation a year and like all day rebalance.
Speaker 3 (34:09):
Not the ones I know who probably listen to this podcast, Okay,
they work very hard.
Speaker 1 (34:14):
Sure indexes aren't paying rebalanced all the time, planning more
than you would think.
Speaker 3 (34:23):
Index rebalancing is another you know, poster child for a
strategy that seems so simple that everybody can talk about it,
and then it's full of nuances and it requires a
lot of skill to trade effectively.
Speaker 1 (34:37):
I believe that just because like I thought a little
bit about like just like the sort of like accounting
of like you basically know how many index mounds there are,
let's say, can predict what will come in and out
of the index, and like what the so like there's
like some mechanics around, like you know, figuring out the market,
calves that will come in and whatever, but then it
(34:57):
feels like the unknown is like who else is doing
the rebalance strategy? Is that? Right?
Speaker 3 (35:01):
I think you're mostly right because I don't want to say,
because you know, out of respect for for the CMS,
did I know?
Speaker 1 (35:10):
I love? Yes? Like so we had Cliff Askiness on
a few weeks ago, and like, to me, Cliff Asness
is like a quantitative investor, like a systematic investor, but
like what he's doing is sort of recognizably what a
sort of traditional asset manager. He's like trying to find
companies that are undervalued, right, he talked about it's like
(35:31):
being a Grammar dot investor. You know, you want like
valuation plus a catalyst. And he's like, oh, or you
know trading, you know, value and momentum and like you
look at what eight or two is maybe a little different,
but there's like you know, the hypercacy trading firms, Like
you can model those as like those are quantitative versions
of like a voice market maker fifty years ago, where
they're like trying to keep inventory flat and like trying
(35:52):
to you know, make the bid asks bread. So like
those are like very traditional economic functions that have been
quantify like turned into systematic what's the intuition for like
what a bally Asni or a star doll or a
millennium does? Like what business are you in? Do you think?
(36:12):
Like as a philosophical matter, like one thing I think,
like I think about like.
Speaker 3 (36:18):
You're asking from a social kind of point or.
Speaker 1 (36:21):
Well, I think like the index rebalancing, Like to me,
it feels like the sort of trade and I think
to something that was the sort of trade that like
an investment bank would have done twenty years ago, thirty
years ago, and like some of that function I think
has moved to like the big multimanagers. But like I
wonder like from where you say, like how you see
the like role in the financial markets of those firms.
Speaker 3 (36:42):
So at a very high level, we don't do anything
different than everybody else in the sense that what we
provide is always this, right, is we provide shifting time preferences,
which means we provide liquidity. We house you know, risk
for people who don't want to hold it right now.
And that's what you do when you do indext rebalancing, right,
(37:03):
that's what you do when you do merger ARB and
when you do the various subtypes of basis traits.
Speaker 1 (37:09):
Right.
Speaker 3 (37:09):
So we do provide liquidity, which is very important. And
then the second thing we again very high level, we
provide price discovery, right, So we study the firms and
we think, okay, this is at the margin mispriced and
we're going to short it or we're going to invest
in it, and that's a beautiful thing. So we do
it at a different time scale, right. So you always
(37:29):
want to do things at the margin where you don't
have a lot of other participants, and at the margin
of the let's say month to three month investment horizon,
there are not that many participants. So in the words
of another hatch fund manager I cannot name, but it said,
once you know, we don't invest in securities, we dated them,
(37:51):
and so we are in the dating service. Not that
many people are doing it, and so we do it.
But I would say also this right, not at the
social level. I just want to answer the like my
personal level. What we do. We are a massive filter
of talent, and the talent that we hire is a
massive filter of information. So it's like information squared.
Speaker 1 (38:12):
Maybe this is like a bad question, but like, do
you think that like long only asset managers are worse
than they were thirty years ago because that filter has
been so successful? In other words, like there are lots
of jobs you could have gotten in finance in nineteen ninety,
but like, yeah, there's like a clear hierarchy now.
Speaker 3 (38:31):
I think that the market and the set of investors
has learned right, and I think the distinction between VITA
and HALFA has been useful for investors, and so active
investors who are mostly long only, I think have suffered
from this distinction because the vast majority of them underperforms
(38:53):
their benchmarks and so there is no reason for them
to exist. And then what we do is we provide
really uncorrelated returns to the benchmarks to most factors, and
investors want that, right, So there is a future where
active investors, long on investors asset managers will become even
(39:16):
less influential, smaller, and also.
Speaker 1 (39:20):
I think of that as like a customer demand side,
but also like a talent filter side.
Speaker 3 (39:23):
Right, yes, Yeah, And then the interesting thing is and
then there is also a process where the multimanager platforms
are able to make the business model of a single
portfolio manager that is not sustainable in isolation working in
this kind of federated system. So why would you or
how could you survive as a single portfolio manager hedge
(39:47):
fund nowadays? It's really really difficult, but you can do
it in a multimanager platform provided that you have you know,
sufficient talent, sufficient edge.
Speaker 2 (39:56):
That's also where you can blame the passive influence on Twitter.
If you're a long entry manager that you know it's
impossible to be the market now because you just have
this money constantly pouring in.
Speaker 3 (40:05):
Yeah, I don't disagree. Yeah.
Speaker 1 (40:08):
One more question, like social roles is just like you've
worked at most of the big pod jobs, but you
also worked at HRT, Like what's the difference in roles
and like what they do all day? Because HRT, I
think of is a classic like high frequency treating firm
where I don't know they're exactly a market maker, but
they're certainly on the higher frequency side, and then like
the pod jobs have a lower frequency and a you know,
(40:30):
they're not prop they're running hedgephones. Like what's the cultural
and role and differences?
Speaker 3 (40:37):
Yeah, okay, So I briefly mentioned the HRT in a
in an interview with The Financial Times, and my manager
told me that, you know, people at HRT were both
annoyed and delighted by what I've had said about about HRT.
I think HRT is a really special place, even in
the in the context of proper training firms. So I'm
(40:59):
a little bit hesitant and to just being them in
as a representative, right, So they're not representative because there
is something in the culture of HRT that is special. Okay,
it's collaborative, it's truly kind. Yeah. So I think it's
(41:19):
a great place to work, and it is fundamentally monolithic,
so you have, you know, sharing of ideas and you
can work at the intersection of these ideas. It's also
a place that is very tech oriented, so it's a
bit of a technology firm operating in the financial space.
(41:40):
And because of that, it also attracts i think the
best technical talent that I've ever worked with. It's just
a pleasure to work with great technologists, people who are
very competent in that respect. So nothing against the hedge funds.
I love edge funds for different reasons. You know. I
love BAM, which is also very collaborative and it's an
investor meant company. But HRT has as a technical side
(42:04):
to it and also gain a cultural side to it.
It's great.
Speaker 2 (42:23):
We didn't talk about AI AI.
Speaker 3 (42:26):
Yeah, of course you have to talk about it.
Speaker 1 (42:29):
Like I have like three models of how investment works
systematic about Like one is like you have like some
economic intuition and you build a model of like the
stock market that predicts prices. And in other way is a
sort of like neural netty ai E way, where like
you throw a lot of data at a neural net
and it build its own model of how to predict
stock prices. And then the third model is like you
(42:51):
get really good at prompt engineering and you get a
chat GPT and you say what stocks will go up?
But you ask it in the right way, and then
chat JPT tells me what stocks will go up? How
good is it? I see? The third model no one uses,
but like someone uses.
Speaker 2 (43:06):
I think a lot of people use that.
Speaker 3 (43:08):
All right, So first thing like, Okay, nobody knows anything,
and anybody saying the opposite, you know, should be heavily discounted. Okay,
so we agree on this, and so let's forget for
a second all the technical details of AI just from
a pure industrial organization standpoint, Right, So what's going to happen?
(43:33):
Consider AI just like another technology like Internet and whatnot. Right,
So you know, first of all, we're going to observe
economies of scale, So there's going to be concentration, and
there was going to be some kind of monopolistic competition.
I was thinking about Bloomberg specifically, which could be I
hope for you people to be among the winners, because
(43:56):
you have a good starting point, right, You have lots
of data, you have a customer base, and maybe in
the future we'll finally not see the good old Bloomberg terminal,
which has been kind of unchanged since I remember it,
and instead people will just prompt Bloomberg to conduct very
complex actions where it will act on a sequence of
(44:18):
keywords and connect them and give you, like a much
more valuable product for which Bloomberg will charge twice as
much as they do already. So this is going to
happen in one form or another. If it's not Bloomber,
somebody else will do it. Okay, But the same thing
applies to other areas of finance. So maybe once upon
a time, you know, a big sufficiently big fund could
(44:42):
build their own client for email.
Speaker 1 (44:45):
Right.
Speaker 3 (44:45):
Of course, nobody builds a client for email anymore. Right,
so a lot of this stuff gets outsourced. We will
outsource at some point some of the functions that we
conduct internally using AI to other AI agents. It's perfectly fine.
So this will become a utility to some extent. Yes,
(45:05):
functions include well not stock picking. Not stock picking. I
think that the functions that we will see available are
essentially like another self, like another Mathlevin. Who can you
be a good baseline for you?
Speaker 1 (45:22):
Okay?
Speaker 3 (45:22):
You could feed a post train and AI system with
all your gazillions of words, right, and that agent will
reproduce your sense of humor, your investigative style and everything. Okay,
it's a good approximation. It's not going to be perfect,
but why not?
Speaker 1 (45:39):
Right?
Speaker 3 (45:40):
So I would be very happy to have a replica
of myself that can answer most simple questions. Now, I
think that the decision to invest in a particular stock
is a very demanding cognitive function, and I don't see
that really being replicated very well. But I think that
this will be baselined to some extent.
Speaker 1 (45:59):
Is it many kind of function because because it exists
in a competitive market, So like the sort of like
whatever the kind of function is, is going to get
like the baseline is always going to get higher because
like someone else will will have the same information as
you do or the same.
Speaker 3 (46:14):
Well, this is getting really in the highly speculative side
of you know things. I think that in order for
an AI agent to be good at this, they have
to be able to experience the world the same way
that an investor experiences it. And our inputs are much
more complex than just a string of text or YouTube videos.
(46:36):
Right we have a model of the world which comes
from visually experiencing the world, talking to humans, consuming the goods,
right anything, It's vastly more complex than the way an
AI system right now experiences the world and also influences
the world. So an investor has a fundamentally different experience
(46:57):
of a company than an LM that has an experience
thats is mediated by multiple layers of processing. You know,
they learn about a company through text that is written
by somebody. So I don't think that that's in danger
for the time being. But maybe, you know, again, in
five years, maybe we will have our glasses feeding our
(47:18):
experiences to AI agents. Who knows, right, But I don't
think that it's that close, and I don't think AI
is that's smart also, so I think that having a
baseline system would be already pretty good.
Speaker 2 (47:30):
That's somewhat comforting that our experiences count for something, our
physical experience of the world.
Speaker 1 (47:36):
It's interesting because I always think of like the comparison
as like investing in self driving cars, or like investors
do a lot of things, but like one thing they
do a lot is sit at a desk and read
computers and like look at numbers, right, and like those
things seem like things that a computer can do well,
whereas like you know, drivers like have physical reflexes and
like have a you know, complicated field division. I always thought,
(47:59):
like investing should be easier than self driving cars for
computer and a master, but you, I think you're learning this.
Think of like investing as like the great liberal art,
where it's like you incorporate all of human experience and
so the AI can't relate.
Speaker 3 (48:12):
Okay, let's let's take the metaphor to you know, extreme consequences.
Imagine that you had a system that is the equivalent
of a perfect self driving car in investing. So now
I'm giving you a machine, a box that is telling
you the long term value, if not the returns, right,
because the moment that the value is known, you immediately
(48:33):
equilibrate to that level.
Speaker 1 (48:35):
Right.
Speaker 3 (48:35):
So imagine that you know the true value of everything
because a box tells you so, and it's invaluable. It's
an oracle. Okay. Would you think that finance stops existing.
I wouldn't say so, right, So I think that a
lot of arbitrush trades, you know, would maybe change significantly,
but every risk, right, every return would be correctly priced
(48:58):
by the risk of the agent's training it. So there
still would be trading because we still have different preferences,
but basically every risk would be priced. There would be
in a sense, less alpha, but finance will still exist.
Speaker 1 (49:09):
It's a lot of like service provision, like liquid.
Speaker 3 (49:12):
Liquidity provision and yeah, and so the liquidity provision would
still exist. The informational services maybe will stop existing in
the current form, but that's okay. I think that we'll
all still be employed.
Speaker 2 (49:24):
Mhmm.
Speaker 1 (49:26):
It's interesting I think about it because I do think,
like we talked about, like, one thing that the big
hedgehunds do is things that have the flavor of liquidity
provisions basis trades and merger urb and whatever. Things that
like I think of as like something that a bank
would have done thirty years ago, and then now a
big hedge fund does. And then another thing they do
has the flavor of information provision, where it's getting prices right.
(49:49):
Like to me, those things seem quite intellectually separate, but
I guess they feed each other in the sense that
the better you are prices, the better you can be
a liquidity provision.
Speaker 3 (50:04):
The value yeah, I mean a short, short horizon. Liquidity
provision and information tend to be very closely rated. Like
you know, a limit if you're good at if you're
either good at crossing even good at crossing, you should
be pretty good at adding okay, adding liquidity, so you know.
But I mean like you could make you know a
(50:24):
profit by posting a lot of limit orders and providing
liquidity to the market or crossing the spread and making
money with predicting the future prices. If you're good at one,
you're good at the other, most likely, right at that
time scale, I think that this though might I'm not
sure because I haven't thought about this very very carefully,
but I think this might decoupled at longer time scale,
(50:46):
so you know you're when you're out. I'm not sure.
And in any case, at that time scale is really difficult
for an AI or for a human being anyone, like,
there are not that many hard data, even the unstructured
data are not that any So it's a very difficult problem.
It's the coupled it's it's complicated. So yeah, but I
(51:07):
tend to believe at longer time scales you have more
or less liquid provisioning and you know, violations of law
of one price on one side and predicting on the
other side.
Speaker 1 (51:20):
But you combine both.
Speaker 3 (51:22):
But you can combine both, and it's a very potent mix.
Speaker 1 (51:25):
Right. There is normally different people, it is, right.
Speaker 3 (51:28):
Very different people for sure, different parts, very different very
different people, very different cultures.
Speaker 1 (51:33):
Yeah, can you summarize the difference in cultures between like
I have a guess.
Speaker 3 (51:38):
But well, as you said, people who typically trade in arbitrades,
if not historically but also historically come from banks.
Speaker 1 (51:47):
Yeah.
Speaker 3 (51:48):
Right, Whereas you still can see long only portfolio managers
being recycled and reformatted into long short portfolio managers, you
can have an excellent short specialist becoming a long short
portfolio manager like it happened, I.
Speaker 1 (52:05):
Mean makes sense. Is that like the people on the
information version long short side are more academic and research orientered,
and the people on the ARB side are.
Speaker 3 (52:14):
More Yeah, I think you can actually have a very
good long short portfolio managers who were journalists in their
past lives.
Speaker 1 (52:25):
I've heard of some of these thought about this about
it no, just like.
Speaker 2 (52:31):
No not breaking news on your podcast.
Speaker 1 (52:35):
I've noticed that's that's better than podcasting. Not thought about
it in the sense that I'd be good at it,
just in the sense that the money is good.
Speaker 2 (52:46):
You could be bad at it and paid really well
for a short amount of time.
Speaker 1 (52:50):
I don't know that that's true. Actually, they're they're an
excellent talent filter or so I hear.
Speaker 3 (52:57):
Yes, I think that you could interest a few huge funds.
They might be listening.
Speaker 1 (53:07):
On a note, Kathy, thanks for coming on the.
Speaker 3 (53:14):
Thanks for having me, And that.
Speaker 1 (53:21):
Was the money Stuff podcast.
Speaker 2 (53:23):
I'm Matt Levian and I'm Katie Greifeld.
Speaker 1 (53:25):
You can find my work by subscribing to The Money
Stuff newsletter on Bloomberg dot com.
Speaker 2 (53:29):
And you can find me on Bloomberg TV every day
on Open Interest between nine to eleven am Eastern.
Speaker 1 (53:35):
We'd love to hear from you. You can send an
email to Moneypot at Bloomberg dot net, ask us a
question and we might answer it on air.
Speaker 2 (53:42):
You can also subscribe to our show wherever you're listening
right now and leave us a review. It helps more
people find the show.
Speaker 1 (53:48):
The Money Stuff Podcast is produced by Anna Masarakus and
Moses on Them.
Speaker 2 (53:52):
Our theme music was composed by Blake Maples.
Speaker 1 (53:54):
Brandon Francis Nunim is our executive.
Speaker 2 (53:56):
Producer, and Stage Bollman is Bloomberg's head of Podcasts.
Speaker 1 (53:59):
Thanks for listening to The Money Stuff Podcast. We'll be
back next week with more stuff.
Speaker 2 (54:08):
Mm hmm