Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:02
Bloomberg Audio Studios, Podcasts,
0:05
radio News. This
0:10
is Master's in Business with Barry
0:13
Ridholts on Bloomberg.
0:15
Radio this
0:17
week on the podcast, Boy
0:19
Do I have an extra special guest.
0:21
I know Jim O'Shaughnessy.
0:23
For I don't know, maybe twenty plus
0:25
years something like that. We actually first
0:27
met in the green room at CMBC
0:30
like early two thousands
0:33
and found we shared some similar
0:35
likes and philosophies. And I've
0:38
been a fan of his book What Works on Wall
0:40
Street pretty much from when it came out.
0:43
This is a fascinating conversation about a person
0:46
who has worked through multiple
0:49
locales and seats in
0:52
finance, not just running
0:54
a systematic investing at bear Stearn's,
0:57
but creating O'Shaughnessy asset
1:00
management, creating a unique
1:03
custom index product that
1:06
ended up attracting the attention of Franklin
1:08
Templeton, who paid some
1:10
undisclosed and ungodly amount of money
1:13
for the whole firm, and now
1:16
in a later phase of his career
1:18
doing O'shaughnessee ventures and the
1:20
O'shaughnessee Fellowship. I first
1:23
know him from really the first pant
1:25
book What Works on Wall Street? That
1:28
was a half a century of data analysis.
1:31
Really was never accessible to.
1:32
The public before.
1:33
I found the conversation to be fascinating and
1:36
I think you will also. And
1:38
at this point I am obligated to do
1:40
a disclosure. My firm Retults
1:42
Wealth Management has been working with O'Shaughnessey
1:45
on their direct Index platform. Really
1:48
we were one of the first beta testers. We now
1:50
have over a billion dollars on that platform,
1:53
maybe coming even closer to another
1:55
big round number with no further
1:58
ado. My discussion with
2:00
O'Shaughnessy ventures Jim
2:02
O'Shaughnessy.
2:03
It's great to see you, Berry and
2:06
congratulations. Wow, that's a.
2:07
Well congratulations to you. I'm
2:10
still My firm just had its tenth anniversary.
2:13
You guys, anytime I see
2:15
the phrase for an undisclosed
2:17
amount, my brain automatically
2:19
says.
2:19
Wow, that has to be a lot of money. If it's if.
2:21
They're not disclosing it, it's material,
2:24
but undisclosed, that's a lot of casts.
2:26
Or it could be like trading places
2:28
and the normal bet of a dollar.
2:32
The usual bet Mortimer one dollar.
2:35
So we know each other from
2:37
way back when you first came
2:39
into my orbit from the book What Works
2:41
on Wall Street. I read it from
2:44
cover to cover. I was on a trading desk when
2:46
that came out, and I'm like, huh,
2:48
so there's some science and math behind
2:50
this. It's not just rumors and whatever
2:53
happens to cross TV that
2:55
day.
2:56
I'm intrigued.
2:57
Before we get there, let's
2:59
talk a little bit about what you were
3:01
doing prior. Tell us about the
3:03
early Jim O'Shaughnessy.
3:05
Well, I was always fascinated
3:08
about the markets in general,
3:11
which stemmed from a very
3:14
angry conversation between
3:16
my uncle and father about
3:19
IBM. And I
3:21
had just been allowed to go to the adult
3:23
table, and I was sitting
3:25
next to my dad and he
3:27
and my uncle John were going hammer
3:29
and tong about whether IBM
3:32
was a good company or not. And I was listening, and
3:35
it was all about the chairman. It
3:37
was all about, you know, things
3:39
that I looked at as kind of soft intelligence,
3:42
squishy squishy, and
3:44
so I just thought I asked at
3:47
the dinner, I said, well, would
3:49
it make more sense to like look at how much
3:51
money they're making and what their earnings
3:54
are and how much you have to pay for that?
3:56
And they both just literally glared at
3:59
me. That's hilarious
4:01
kids. They don't know anything exactly
4:04
exactly. It's the chairman. How tall is
4:06
he? I like to cut his gips. It's
4:09
almost as if you were there that
4:11
bug got implanted, that mind
4:13
worm got implanted in my brain.
4:16
How old were you when that? I was seventeen.
4:18
Oh so you're just going into college?
4:21
Yeah?
4:21
Absolutely, And you were a
4:23
Minnesota kid, Is that right? I grew up in Saint
4:25
Paul, Minnesota and beautiful
4:28
country. Certainly in the summer anyway, gorgeous.
4:31
The winter's tough.
4:32
Yeah. Yeah. Well, if this were
4:34
the old USSR, that
4:36
is where all the political prisoners
4:38
would be of Minnesota.
4:42
But so I started doing research
4:44
on essentially the Dow thirty
4:47
because it was manageable, thirty
4:50
stocks I could list by hand
4:52
showing how old I am, because you literally
4:55
there were no computers that we could use
4:57
at the time. Simple
4:59
things like like what's the price, what's
5:01
the dividend, what's the price to earnings,
5:04
book value, etc. And I
5:06
found a definite trend,
5:09
right. I found that buying
5:11
the ten stocks and the diw with the
5:13
lowest pees from nineteen like
5:16
thirty five. I think I started through
5:19
when I was doing it, and this would have been about
5:21
nineteen eighty, absolutely
5:23
decimated the ten highest
5:25
PE stocks. So wow,
5:28
I love this. In the meantime, I had
5:30
computers, and the only reason
5:33
I actually got to write What Works on
5:35
Wall Street was because Ben
5:37
Graham didn't have computers.
5:40
If he had had them, I would have had
5:42
no chance because he would have done it. Basically,
5:45
what I wanted to see was,
5:47
is there any rhyme or reason to all
5:49
of these reasons people say they like
5:52
or hate a stock? Right? Where is the
5:54
proof? Where is the empirical
5:56
evidence that say, buying
5:58
the low PE stocks from the die works
6:00
very well over many market cycles.
6:03
So I wrote a first book called invest
6:06
Like the Best, in which I basically
6:08
showed you how you could clone your favorite
6:11
portfolio manager by taking
6:13
his or her stocks, putting
6:15
them on a big database like Compustat,
6:19
seeing how they differed from the overall
6:21
market, and then using those
6:23
as factor screens to get
6:25
down to a portfolio that looked,
6:27
acted, and most importantly, performed
6:30
like your favorite manager.
6:32
Now, the average investor typically
6:35
didn't have access to Compustat, to
6:37
big data, to big computers, and
6:39
so they relied on you who did.
6:42
And if I recall what Works on Wall Street.
6:45
You back tested like half a century
6:47
worth of data something like that, and it was
6:49
the full market, not just the
6:51
thirty dove stocks.
6:53
Yeah. Absolutely. And also
6:55
not just the full market, it was also
6:57
any company that had been but
7:00
went bankrupt or got taken over, the
7:03
very very needed research
7:06
database on compustat. So
7:09
no survivorship bias none, You
7:11
back that out, Yeah, great, Yeah, because
7:13
some of the early academic studies
7:16
were they
7:18
had a lot of survivorship
7:21
bias. They didn't properly
7:23
lag for when you actually
7:26
knew a number, so they just assumed,
7:28
right, well, there's the number on March
7:31
thirty first, I'm going to use that number.
7:33
Well, you didn't really know that for most
7:36
of history until maybe May or June.
7:39
Really interesting, so you run
7:41
these numbers. What sort of strategies
7:44
do you find perform best. Well,
7:46
we found that on the
7:49
value side, smaller value
7:51
stocks that had some catalysts
7:54
and had turned a corner and their
7:57
prices had started to go up.
8:00
A beautiful strategy.
8:02
Small cap value with a touch
8:04
of momentum.
8:05
Yes, okay. On the growth
8:07
side, we found momentum works
8:10
really really well. As
8:12
we continued the research, we found
8:15
okay, there's all sorts of caveats. So,
8:17
for example, we learned after
8:20
a severe bear market i e.
8:22
One in which the market had declined by
8:25
forty or more percent, not a
8:27
lot of those, not a lot, thank god, but
8:30
momentum inverted, and
8:32
the stocks with the worst
8:34
six or twelve month momentum actually
8:37
did vastly better than
8:39
the ones with the best. And if you think about
8:42
it even for a minute, it makes sense, right deepest
8:44
value. But what happened
8:46
was a lot of really great stocks during
8:49
the bear market got pushed
8:51
way low in price, and
8:54
so people when the market was
8:56
recovering jumped on those
8:58
stocks. They were like, I can't believe I'm getting
9:00
you know, these earnings at six times
9:02
earnings for an IBM or
9:05
a you know Qualcom, Right, that's the
9:07
baby with the bathwater strategy exactly.
9:10
And so but we
9:12
found, you know, that value actually
9:15
works now. It hasn't for a long time.
9:18
But we also found
9:20
that large stocks with high
9:22
shareholder yield i e. Dividend
9:25
yield plus buyback yield
9:28
was an excellent way to
9:31
identify big stocks that
9:34
are obviously much more conservative
9:36
than the smaller fry in
9:38
the small cap world.
9:40
Interesting, So let's talk a little
9:42
bit about your work at bear Stearns.
9:45
Really, where I first met you in the two thousands,
9:48
you were head of systematic Equity
9:50
at bear Stearn's asset management. I'm
9:53
assuming you were applying a lot of the lessons
9:55
you learned in what works on Wall Street to
9:58
the Bear institutional and tel
10:00
investing strategies.
10:01
Absolutely, and you
10:03
know, let me just say Bear was
10:06
really a great company, very unfortunate
10:09
what happened to it during the financial crisis.
10:12
But the reason I love Bear is,
10:14
you know, a lot of big banks talk about
10:17
being entrepreneurial. Bear Stearns
10:19
really was. And essentially,
10:22
if you were doing your thing and playing
10:24
by the rules and doing well, they
10:27
let you alone. Which
10:29
was pretty important for me because when
10:31
I got there, it was right after
10:33
the dot bomb, and
10:36
a lot of the brokers had done
10:38
pretty poorly because they were in a lot of those
10:40
names. And so I convinced
10:43
Steve Dantis, who was then head
10:45
of Private Client Services, that
10:48
wouldn't it be better if we did a
10:51
packaged portfolio, a
10:54
separately managed to count and we
10:56
offered at one time
10:58
I think we were all the way up to
11:01
the brokers so that they
11:04
could use a more systematic
11:07
time tested way of investing
11:10
for their clients.
11:11
Bringing a little discipline into
11:13
what had been, at least in the nineties
11:15
very much a cowboy type of environment.
11:18
And I'm not just for fron of Bear. The
11:21
entire retail stock brokerage
11:24
was wild totally.
11:25
He was very open to it.
11:28
We ended up putting together a separately
11:31
managed to count platform that they brokers
11:33
embraced. They loved it because literally
11:36
they did what they did well, which was
11:38
calm the client during bad times,
11:41
try to keep them from getting too excited
11:43
during great times. But they also loved
11:45
the idea that it had a very
11:48
explicit explanation for
11:50
why they were putting that client in
11:52
that portfolio. So that
11:55
was a lot of fun. By the time I left
11:58
Bear, my group controlled
12:00
about seventy percent of Bear
12:02
Stearn's asset management long only,
12:04
and that was a lot of money, wasn't it. It was
12:07
It was about fourteen billion dollars.
12:09
Okay, so you mentioned you
12:11
left Bear. Let's put a little flesh
12:13
on those bones. Your
12:15
timing was perfect.
12:17
You exit Bear in two thousand and seven,
12:19
Is that right?
12:20
To set up O'Shaughnessy's asset
12:23
management was the thinking,
12:25
Hey, I want to do this out on my own shop,
12:28
or were you sniffing something
12:30
out in O seven that's like, hey,
12:33
maybe I don't want to be attached to
12:35
a giant ocean liner taking
12:38
on water.
12:39
You know, that's funny. I spent the next
12:41
two years after that trying
12:44
to convince reporters that I
12:46
really didn't know anything. Why
12:49
I left Bear was because I felt
12:52
that I really wanted to be on my own
12:54
again. I really wanted to
12:56
be able to just talk about
12:58
quantitative investing. Bear
13:01
was a boutique, so there were a lot of different
13:03
managers. Liked them all.
13:05
I thought they all were great, but I
13:08
really really wanted to focus just
13:10
exclusively on quant And
13:13
secondly, we had upgraded a lot
13:15
of our systems to the
13:17
idea that would become canvas,
13:20
right, because remember Netfolio was
13:22
our first try at that. That was nineties
13:25
or ninety nine. Yeah, really, well,
13:27
of course, you know the really funny
13:29
story here is in April
13:31
of nineteen ninety nine, I
13:34
wrote a piece called the Internet
13:36
Contrarian, and in that piece
13:38
I said, eighty five percent of the companies
13:41
currently extant in the Internet
13:43
space are going to be carried out of the market.
13:45
Feet First, I've never
13:48
seen a bubble like this in my history
13:51
of investing. And what did I do
13:53
next? Berry? I started an Internet
13:55
company.
13:57
Well just because the stocks or a bubble,
13:59
does I mean this internet thing isn't
14:01
going to catch on?
14:02
That's true, right, it's
14:05
there are you know?
14:05
It's funny we forget in the
14:08
thirties, forties, fifties there was only ma
14:10
bel Every company used telephones.
14:13
Yep, the way we describe internet
14:15
companies, if you use
14:17
the Internet as a core part of your
14:19
platform. There's difference
14:22
between the dot.
14:22
Coms and the nineties and people.
14:24
Who have just really integrated
14:27
the technology into their business. Right,
14:29
So I think Netfolio is not
14:32
a dot com but a comm
14:34
that used the net as a way to reach
14:37
more people and give them access to data.
14:39
Well, it's really funny because I made
14:41
a couple well I made more than a couple of mistakes,
14:44
but one of the big ones I made was
14:47
we designed Netfolio as
14:49
a B two C company, so
14:52
we called we were taking on at
14:54
the time mutual funds, which were dominant.
14:56
We didn't have ETFs while we
14:58
had them, but they were there very early day,
15:01
very very early days.
15:03
And so what did the spiders just turned
15:05
twenty five recently, Yeah, I think of something
15:07
like that. Yeah, So ninety nine is
15:10
like it was really the
15:12
beginning.
15:12
Oh totally. And basically
15:14
the idea was it was the first
15:17
online investment advisor.
15:19
And the reason that we
15:22
thought it would work so well was personalization,
15:25
tax management, all of
15:27
those things. So, for example, we
15:29
would they were all run by quant models
15:31
that we had developed, right, but
15:35
it gave the user the ability to say,
15:37
let's say they're anti smoking, right, and
15:39
Philip Morris is one of these selections
15:42
they could just check nope, don't want it.
15:45
Up comes the next stock that meets
15:47
the criteria, and so it had
15:49
a lot of really great features, but
15:51
the tech was not quite there yet.
15:54
You were twenty years ahead of
15:56
where you would end up in
15:59
the late tens, right.
16:01
I.
16:03
Was.
16:03
I really do have to give my son
16:06
Patrick the credit for resurrecting
16:09
the idea because when we were
16:11
at OSAM, I said,
16:13
listen, we left Bear
16:16
right into the Great Financial Crisis, and
16:18
I put the team together and I'm like, I don't
16:21
think that we're going to be able to sell many long
16:23
only portfolios after the market
16:25
has collapsed by nearly fifty percent,
16:28
So let's spend our time developing
16:32
internal technology that works
16:34
the way we work. The office
16:36
shelf stuff really wasn't cutting it, and
16:39
so the project to get there was
16:41
multi year and Patrick
16:43
oversaw that, and then he walked into my office
16:45
one day and he goes, you
16:48
know, Dad, we've been using the desk star
16:50
to kill a mouse. And I'm like, okay,
16:52
I like the metaphor, but what
16:54
do you mean? And he started talking about AWS,
16:58
talking about netfolio and he's like, we
17:00
have the perfect tech now
17:03
that our clients results being
17:05
one of them, could use and
17:07
I'm like, brilliant, let's go
17:09
with it.
17:10
So we're going to talk a little more about Canvas,
17:13
but I want to stay with the
17:16
launch of O SM and O seven. So
17:18
A, you don't need to disclose
17:21
this, but I'm going to assume you had a
17:23
lot of bear Stern stock options that you had
17:25
a vest on your exit, so
17:27
you probably had a pretty good sale,
17:29
pretty good print on those
17:33
when you first set up O'Shaughnessy, you're
17:36
running your traditional models, things
17:38
like Cornerstone value and Cornerstone
17:41
growth, and I'm a big fan
17:43
of your microcap
17:45
sleeve, which really operates
17:48
parallel to venture capital returns,
17:50
only using public stocks.
17:53
Am I getting that more or less right? Actually we use
17:56
that also. Yeah, we wrote a paper
17:58
saying that it was the poor man's
18:00
way to get exposure to private equity.
18:03
Private equity or venture capital are both both
18:05
really private equity closer because
18:08
the the microcap I love microcap
18:11
investing. The only real reason that
18:13
we offered that was because I loved it
18:16
so much. Well, and the data backs
18:18
it up, Oh, totally, totally it
18:21
is. Microcap is an
18:23
amazing place if
18:26
you've got the right tools to sort
18:28
through the thousands of names
18:30
in the microcap universe, because
18:33
you would not want to buy an
18:35
index of microcap stocks.
18:38
For the most part, there are microcaps
18:40
because they kind of suck. However,
18:44
there are so many diamonds
18:46
in the rough in microcap
18:48
that if you have a strategy, like
18:50
a quant strategy that can sort
18:52
through these thousands of names,
18:55
you can do extraordinarily well.
18:57
I love the strategy and I
18:59
know.
19:00
Oh the OSAM microcap
19:02
sleeve is what I call. It has
19:04
just really shot the lights out, especially
19:07
last year when the market was having.
19:09
A pretty good year. They did pretty
19:11
well, didn't they They did? They did. Now,
19:13
remember you introduced me as chairman of o
19:15
SAM, I'm no longer no longer. Yeah,
19:17
they let me retire and
19:21
actually Patrick is now chairman
19:23
emeritus over at OSAM.
19:26
Let's talk a little bit about Canvas, and again full
19:28
disclosure, we're a client.
19:30
We were a beta test. Do we love the product?
19:32
Which is kind of ironic because I
19:35
used to hate direct indexing every
19:37
time I would demo
19:40
or see a product. It was clunky,
19:43
it was cludgy. You would get these
19:45
statements that were like hundreds of pages
19:47
long.
19:48
You guys kind.
19:49
Of figured out the secret
19:52
sauce for how do we make this
19:54
clean, usable and easier
19:56
to understand. Tell us a little
19:58
bit about the genesis
20:01
of Canvas.
20:02
Well, first of all, we call it
20:04
custom indexing as opposed
20:06
to direct and the reason
20:08
I make that distinction is because,
20:11
as you point out, the direct
20:13
indexing products of that time
20:16
were clunky, they were difficult.
20:18
You got reams and reams of paper
20:21
reports, and they were really only
20:23
focusing on tax benefits.
20:26
Right. What we wanted to
20:28
do with Canvas, which
20:30
is custom indexing, is, as the
20:32
name implies, give you, as
20:35
the advisor, full control
20:38
over what your client portfolio
20:40
wanted to look like. You got the advantages
20:43
of tax harvesting. You got
20:45
the advantages of being able to mix
20:48
indexes in with active strategies,
20:51
but you could also do a social
20:54
investing fund if you want it. But the
20:56
way we did it was we didn't presume
21:00
what your client was going to think
21:02
of as good social investing.
21:04
So often when you see some
21:07
of the ESG portfolios,
21:09
they've been predetermined as to what
21:11
is going to be included. We give
21:14
you the tools to
21:16
turn a dial up or down on whatever
21:18
you want. I think last I looked, there were
21:20
over fifty eight separate
21:22
things that you could find tune around on
21:25
the idea of ESG. We wanted
21:27
to give the tools to you because
21:30
you knew your client vastly better
21:32
than we did, and we thought,
21:34
let's try. As you mentioned you were
21:36
one of the beta testers. That was actually
21:39
one of the smartest things we did, I think,
21:41
because we had really good
21:43
advice from a lot of people that we knew
21:46
in both venture and other places.
21:49
The first thing that many of them said to us
21:51
was do not try to go
21:53
big with this. Originally, find
21:56
advisors who you trust who
21:59
will give you you real feedback.
22:01
In other words, they won't shine you on if
22:03
they didn't like you. Guys were very good
22:06
at times.
22:06
And Michael batt nicking my office, one
22:08
of my partners, who was
22:10
over the moon when he first saw this. Every
22:13
time another product came in, it would
22:15
take me thirty seconds to poke holes
22:17
in it. And he came breathless
22:20
into my office, Dude, you got to see
22:22
this. And I'm like, yeah, yeah, okay, another
22:24
garbage right, tee it up. And
22:26
it took about thirty seconds to go, Oh my god,
22:29
how do we get a piece of this?
22:30
This is fantastic.
22:31
The interface, the design, all
22:34
of the bullet points that all the boxes
22:36
checked were great. Let's stick
22:39
with what we no longer call esg
22:42
and Meyer Statman famously
22:45
called values based investing. Some
22:48
people have called it woke investing, but that's
22:50
really the wrong phrase. I'm fascinated,
22:53
for example, by the Catholic
22:55
bishops whose endowment says,
22:58
look, we don't want any aboard of any
23:00
drugs that do that. We can't invest in those companies.
23:03
We can invest in hospital chains
23:05
that perform these sort of surgeries,
23:08
or insurers. You have the ability
23:10
to say, whatever your personal
23:12
preferences are, you could just tune
23:15
those out of pick an index,
23:17
the S and P five hundred the Vanguard Total
23:20
Market. You could say, I don't
23:22
want X or Y or Z, and
23:24
how it comes tell us a little bit
23:26
about that.
23:27
I felt that that was really really
23:29
important because everybody has
23:32
different ideas. As you point out, the Catholic
23:34
bishops wanted to exclude
23:36
certain things. Others might want to include
23:39
certain things. Actually felt it would
23:41
be very arrogant of us to
23:44
determine what good social investing
23:47
was because we had managed
23:49
money for a variety of religious
23:51
institutions, and guess
23:53
what, they all have different
23:55
takes on what they want to see.
23:57
We did one where, for example,
24:00
well you couldn't buy any company that
24:02
did anything with animals with eyes.
24:05
That was an interesting one. But then
24:07
on the other hand, we had a client who
24:09
wanted to see more female
24:12
board members and females in the
24:14
C suite, and you could you could screen for that.
24:16
You can screen and there's a bunch of research that shows
24:18
those companies. Now you don't know if it's posative
24:21
or just merely correlated, but those companies
24:24
tend to outperform. The request
24:27
we probably hear the most is
24:29
no gun stocks, no tobacco stocks.
24:31
Yeah, kind of interesting. Yeah, the tobacco
24:34
guns, those are pretty
24:37
large groups where majority
24:39
of investors want nothing
24:41
to do with them. But the other thing that's
24:43
cool about our dials on
24:46
canvas you Let's
24:48
say that Ritholtz has a wild
24:50
eyed libertarian walk in who
24:52
happens to have a billion dollars and
24:55
he says, you know what I want
24:57
the gun manufacturers. I want
24:59
I'm a big like an amendment guy, right,
25:01
Or I want the pharmaceuticals,
25:03
or I mean the sinstock, I mean gambling
25:05
and alcohol. Well, and you know the joke
25:08
there was that my first company, O'Shaughnessy
25:10
Capital Management, we used to keep
25:12
a joke portfolio which was called
25:14
the Eat, Drink and Be Merry for tomorrow
25:16
you die Berry. It
25:19
killed me sure.
25:20
So what ends up happening very often
25:23
is when there's a non financial
25:26
reason for kicking a stock
25:28
out out of a lot of portfolios.
25:30
Eventually a company with
25:33
still having decent financial prospects,
25:36
it becomes cheap.
25:37
Yep. Absolutely, But
25:39
the thing with the social style
25:42
investing, we wanted you
25:44
to be able to reflect your client's
25:47
unique needs. And
25:49
there really wasn't anything like that. I
25:51
don't know if there is now, but I haven't
25:54
seen anything like that.
25:55
Well, certainly not to this degree
25:57
of granularity. By the way, when
25:59
we first we're beta testing
26:02
Canvas. Internally, my
26:04
view was, Hey, people
26:06
are going to want to use this for value
26:08
based investing. Then they're going to want
26:10
to deconcentrate. If
26:12
I work for Google, do I really need all
26:15
this tech exposure My income is
26:17
coming from there, Let me diversify that way.
26:19
And then tax loss harvesting was going to
26:21
bring up the rear. I had it exactly
26:24
backwards, in large part
26:26
because I don't know, maybe a year into
26:28
it we had the COVID crash
26:31
market falls thirty four percent and
26:33
coincidentally bottoms just near
26:36
the end of the quarter. That
26:38
rebalance. You know, typical
26:40
tax loss harvesting own a
26:42
dozen mutual funds. You pick up
26:45
ten twenty basis points against
26:47
the portfolio of losses to offset
26:50
gains. The hope with this was
26:52
it would be fifty sixty. We
26:54
had clients getting two hundred, three hundred,
26:56
four hundred basis points. And I've talked
26:59
to some of your staff or
27:01
former staff, and they've told
27:03
us some unique use cases where
27:05
the numbers are bonkers. First off,
27:07
explain to the audience who
27:09
may not be familiar with this what is
27:12
tax loss harvesting.
27:13
So essentially what it does is
27:16
we had to build a non trivial
27:18
algorithm that could monitor
27:21
every portfolio we were managing
27:23
on behalf of clients and
27:26
as you know, they can go all
27:28
the way up get maximized
27:30
tax losses or all the way
27:32
down don't worry about them. So,
27:35
for example, you wouldn't care about it in an IRA
27:37
right right. But
27:40
the purpose was that we
27:42
found through our research that
27:45
a tremendous amount of alpha
27:48
was being left on the table, and that was
27:50
the alpha from tax lost harvesting.
27:53
When you're in a market like the market
27:55
we had when we went into COVID
27:57
and the bear market ensued, and
28:00
under other circumstances, well
28:02
kind of you're out of luck. But in
28:04
this particular case, that creates
28:07
the kick in for harvesting
28:09
the losses, reducing
28:12
the overall tax needs
28:14
for the portfolio, and you could
28:16
really look at that as that's money in your
28:18
pocket. By the way, we had
28:20
the benefits completely backward
28:23
too. A tax loss harvesting was
28:25
at the bottom of our list as well.
28:27
It's arcane and technical
28:29
and you don't really think about it, but
28:32
we have clients who were either you know
28:34
startup founders that cashed out, or
28:36
they inherited or or just
28:39
owned stock with a very low cost
28:41
basis. You know, it's always funny when you
28:43
see a five million dollar portfolio
28:46
and some stock has blown up
28:48
where it's eighty percent of the holdings.
28:51
Hey, if you have five million dollars
28:53
and four million of it is Apple
28:55
or Amazon or some combination
28:57
of big stocks, that's a lot of
29:00
angle stock risk. And to a
29:02
man, every person says, hey, you
29:05
should diversify. The answer as always, I'm
29:07
gonna get killed in capital gains taxes.
29:10
This worked out to be a really good way to say
29:13
we're gonna work out of your concentrated
29:15
position over three, four or five years,
29:18
and then twenty twenty comes along and what
29:20
should have been a five year process
29:23
took half as long because you had so
29:25
many losses. So for those people
29:27
who may not be familiar with this, let's
29:30
say you own ten mutual funds, right
29:32
and some are up, one or two are
29:34
down. You sell the ones that are
29:36
down, you replace it with something very
29:39
similar. Hey, now I got a little bit of
29:41
loss even and my portfolio looks the same, but
29:43
I have an actual realized loss that I
29:45
could use to offset my real
29:47
gains. But those losses are three
29:50
five ten percent.
29:51
They're nothing.
29:52
On the other hand, if you have a direct
29:55
index or a custom index that has
29:57
a couple of hundred stocks, well,
30:00
the worst stocks in those portfolios,
30:02
they're not down three four five percent, they're
30:05
down forty sixty seventy
30:07
five percent. You sell the ones
30:09
that are down, you replace them. And this is one
30:11
of the things I like about canvas. You
30:14
identify the replacement stocks
30:16
that are is it fair to say
30:18
mathematically similar?
30:19
They look, well, they come
30:22
from the same strategy, so yeah,
30:24
you could say they were mathematically similar.
30:27
So the overall portfolio more or less
30:29
retains the same characteristics.
30:32
You're just realizing losses deep
30:35
losses on some stocks and replacing
30:37
them with something relatively
30:39
similar exactly. And you know,
30:42
we're just basically making math
30:44
work for us. And because
30:46
the entire thing is operated
30:49
within the canvas architecture. After
30:52
getting the algorithm, which was non trivial
30:55
what do you mean by non trivial outbum, it took a
30:57
hell of a lot of work, okay to be able
30:59
to make that function properly, And
31:03
as we worked with firms like yours,
31:06
it became very very clear to us
31:09
that that was going to be a big deal in
31:11
canvas, So we wanted
31:13
that algorithm to work perfectly.
31:16
But as you also note, we wanted
31:18
the nearest neighbor, if you will, that
31:20
would replace that stock to
31:23
not affect the overall metrics
31:26
of your portfolio. So it's going
31:28
to look, act and perform very much
31:31
like the earlier portfolio, but
31:33
you've already taken that wonderful tex loss
31:36
so that you can offset the gains from
31:39
elsewhere. The other use
31:41
case that we thought would be number one
31:44
was, you know you have a concentrated
31:46
position. Let's say Google
31:49
right, don't give me any tech
31:51
exposure, or give
31:54
me tech exposure only in this
31:56
tech which is like hardware for example,
31:58
right that I can do and
32:01
that type of use case would
32:03
work hand in hand with the
32:05
tax loss, making it a much
32:08
much more efficient, more money
32:10
in the investors pocket. In
32:12
terms of final outcomes
32:14
with the portfolios.
32:15
What was the uptake on that approach? People
32:18
enthusiastic about.
32:19
It, they were, but they
32:21
were not nearly as enthusiastic
32:24
as we anticipated they would
32:26
be. There were a few advisors
32:29
that we were working with who worked
32:31
specifically with founders
32:34
and early employees who
32:36
had a lot of options in that particular
32:38
and usually tech, but
32:40
we also did work and do work
32:43
with a lot of people who just
32:45
amassed through employment a
32:47
huge position in their particular
32:50
company, and they wanted to
32:52
have the rest of the portfolio be built
32:55
to complement and offset,
32:57
if you will, any further invent's
33:00
over there. So it's worked actually quite nicely.
33:03
And then in twenty twenty one, Franklin
33:05
Templeton comes knocking at the door.
33:08
They're an investment giant with a
33:11
trillion plus dollars on their books
33:13
and they've been pretty acquisitive over
33:16
the past few years. Tell
33:18
us a little bit about how that transaction
33:21
began. If I recall correctly, you
33:23
guys weren't out shopping the firm to.
33:25
Be sold, were you not? At all? We
33:27
were. It's a funny story. We almost
33:30
got kind of a cold call from a
33:32
gentleman at Franklin Templeton. I
33:35
was sort of like, give it to Chris
33:37
Lovelace or you know who's the president of the
33:39
firm, And ultimately
33:42
Patrick spoke with him and came into my office
33:44
and he's like, hey, Franklin
33:46
Templeton is really interested in
33:48
canvas. I'm like, okay,
33:51
they want to use it. No,
33:53
no, they want to buy it. And
33:56
I'm like, okay, well, let's do
33:58
a due diligence on Franklin Templeton.
34:01
They're massive, as you know, I think trillion
34:03
and a half in assets under management,
34:06
and we were really having
34:08
great results, as you know, with Canvas
34:11
on our own. We thought about
34:13
it for a long time, and you
34:15
know, we really wanted custom
34:17
indexing to be a new category
34:20
of asset management, and we felt
34:22
really proud about that because it isn't
34:24
too often that you're able to invent
34:26
kind of a new category of
34:28
investing. And as
34:30
we chatted about it and talked it out,
34:33
we're like, you know, we're at an
34:35
inflection point here. We
34:37
are relatively small boutique,
34:40
even though this is working really really
34:42
well. If we want custom
34:45
indexing, custom portfolio creation
34:48
to really make the big time, it
34:50
probably makes sense for a
34:53
much larger asset manager
34:55
with all sorts of advantages
34:58
that we did not have to
35:00
take it and run with it. So
35:03
we let that be our guide, and
35:06
after doing quite a bit of due diligence on
35:08
the people at Franklin, we were like, okay,
35:10
let's negotiate about selling
35:12
the firm to them.
35:13
Talk about good timing. Morgan
35:15
Stanley bought one of your
35:18
competitors in that space. Vanguard
35:20
rolled out their own product,
35:23
which quickly amassed you
35:25
know, billions and billions of dollars on it. So
35:27
this has worked its way into
35:30
the mainstream, even though it's still
35:32
relatively I don't want
35:34
to call it a niche product because it's bigger
35:37
than that.
35:38
But it's not ETFs.
35:40
It's not giant yet, but
35:42
it's still growing at a pretty rapid clip, isn't
35:45
it Totally? And I think that
35:47
ultimately we might
35:50
look back ten years from now and
35:53
have the thought, can you imagine that people
35:55
just bought packaged products? I
35:58
mean, like, my god, no
36:00
tax advantage, none of the customization,
36:04
none of the immunization for concentrated
36:06
positions that I have. And
36:09
so we definitely think that
36:11
this is a way of investing
36:14
that well, you know, once a
36:17
client sees their portfolio
36:20
under canvas and with the
36:22
customization, it's
36:24
really really hard to go back
36:27
to thinking, ah, you know what, I think I'll
36:29
just go with five mutual funds or five
36:31
ETFs. I don't really care about
36:33
much of the other I think that
36:35
you know, these things take time. But I
36:38
mean again, your firm is a classic
36:41
example here. You were able
36:43
to use custom in a way
36:46
that was good for your firm,
36:48
good for your clients. And
36:50
you know, the clients that we speak
36:52
with love it. You know, they
36:55
all love it.
36:55
That's been our experience. It's really
36:58
Mark Andriesen's software is the
37:00
world writ large. Because
37:03
there are two aspects to this, and I'm going to circle
37:05
back to the database part of it
37:07
in a bit. But the front
37:10
end, the user interface and the software
37:13
that allows a very simple
37:16
set of choices and that you could
37:19
go increasingly down the rabbit hole and
37:21
find more and more and more issues certainly
37:23
is a big factor a lot
37:26
of what is done. The
37:28
technology just wasn't quite
37:31
mature enough fifteen twenty
37:34
years beforehand. And when you
37:36
look at it, it's just well, this is just software.
37:38
It's just a user interface and a
37:40
way of organizing it. But now
37:43
let's circle back to the database,
37:45
which I recall you saying was
37:47
the secret sauce. Tell us a little
37:50
bit about the database that you've been working
37:52
on for a quarter century. That
37:55
drives canvas.
37:57
So we use the copystat
38:00
universe. They cover virtually every
38:02
company that trades both here
38:05
on American exchanges and elsewhere,
38:08
and it is kind of
38:10
the gold standard really in terms of
38:13
databases.
38:14
How does it compare to something like CRISPER
38:17
or some of the other.
38:18
Well, so, CRISP. It comes to us from
38:20
the University of Chicago Center for Research
38:23
and Security Pricing. The downside
38:25
of CRISP is it's, first
38:27
off, I love chris We used it in
38:29
the most recent edition of What Works,
38:32
but it doesn't have enough
38:34
of the fundamental factors
38:37
attached to it. In other words, it's mostly
38:39
price history rice history. And
38:42
it also tries and
38:44
generally succeeds to
38:47
include all of the names
38:49
that might have been around trading
38:51
on the AMEX or the New York Stock Exchange
38:53
or NASDAC. But the challenge
38:55
is a guy by the name of Macquarie wrote
38:58
a really compelling paper talking about
39:00
how a lot of the historical data
39:03
not compustat, but further back
39:05
right in the twenties and thirties, come
39:07
from the papers. Yeah, and also
39:11
wasn't nearly as thorough as
39:14
say the compustat is. In
39:16
fact, one of the things that we were doing
39:19
before Franklin Templeton approached us
39:21
is we were literally digitizing
39:25
old Moody's manuals. They go
39:27
back to nineteen hundred, and what
39:29
we wanted to do was marry
39:32
into the CRISP data all
39:34
of the fundamental factors that
39:36
would have given us the ability to run
39:38
a nineteen hundred through nineteen
39:41
fifty five when Compustat begins
39:44
test We ran some test runs.
39:46
We did price to book, and we did a couple
39:48
others. And what we were finding and
39:51
won't surprise you, generally
39:53
speaking, same kind of results,
39:55
right with the exceptional price to book, we
39:57
actually took price to book of
40:00
our composites. You know how we have
40:02
the composits for value and momentum
40:04
and all of those things, and we took
40:06
price to book out because of the research
40:09
that we did that covered the great depression
40:11
from the thirties. You know, and I
40:13
know if you've taken any finance courses,
40:16
price to book previously had been used
40:18
as a proxy for likelihood
40:20
of bankruptcy. Right, Well,
40:23
guess what during the thirties,
40:25
a lot of those low price to book companies
40:27
went bankrupt. Well, when your book value
40:30
collapses, exactly, it's the book
40:32
isn't much value exactly exactly.
40:34
So we did find some learnings
40:37
where we jiggered with the
40:40
composites that we use. That's another thing
40:42
we do. We don't use a
40:44
single factor. In my first version of What
40:46
Works on Wall Street, we would sort down
40:49
for the final portfolio on a single
40:51
factor, and we found that that wasn't
40:53
nearly as effective as
40:56
a composite of factors. Again,
40:58
a lot of people the old joke about quants, right,
41:01
what do you guys do golf all day? You
41:03
know you're just running your models. Well,
41:05
we don't golf all day. But what we do
41:08
do all day is research
41:10
the underlying models. What
41:12
we're always trying to do is
41:14
improve them. But it's
41:17
evolutionary, not revolutionary.
41:20
Listen, the foundations are very, very
41:22
similar. By the way. They make a lot of
41:24
sense too. I say, if
41:26
we changed it and walked out
41:29
onto Lexington Avenue here and we
41:31
found a food truck, right, and we
41:33
went up and long line. Everything
41:35
looks good. And we talked to the owner
41:37
and we said, how about you clearing
41:39
a year And he says, well, I'm clearing
41:41
one hundred thousand. And we're like, well,
41:44
would you take a buy offer from us? And
41:46
he goes, yeah, you can buy it for ten
41:48
million. You and I are going to go get
41:50
out of here. There's no way we're going to buy this
41:53
right, Well, change it to a stock
41:55
ticker. There's a lot of stocks trading
41:58
at that kind of multiple, and so
42:00
when you look at the underlying strategies,
42:02
they make intuitive economic
42:05
sense, and so the
42:07
data set that you're using becomes of
42:10
paramount importance. The other
42:12
thing I found was that, and
42:14
this one disturbed me a little. I haven't looked
42:16
at this recently, but when I was doing
42:18
it several years ago, you could
42:20
get really different numbers if
42:23
you went to Bloomberg, or if you
42:25
went to Reuter's, or if you went to Dow
42:27
Jones or any other innumerable
42:30
providers of data. And so that
42:32
was another huge project for us,
42:35
and also part of the data set
42:37
that we're talking about. One of
42:39
the other things that I was widely hated
42:41
for by my research team was
42:45
we went on a multi year data
42:47
cleansing exercise because
42:50
we found that a lot of it had a
42:52
lot of hair on it. And
42:54
so I'd made no friends
42:56
on the research desk when I said,
42:59
listen, we've got to get this
43:01
pristine. And so our
43:04
data cleansing of the universe
43:06
also is another real important
43:08
distinction between just
43:11
generally available data and
43:13
that which we are using.
43:14
Huh, really really interesting. Let's
43:16
stay with price to book because
43:19
I want to ask your opinion on something,
43:21
and you're the perfect quant to bring this up
43:23
to. Which is all right? So
43:25
we're talking about price to book. Back
43:28
in the day when manufacturing
43:31
required a lot of men and material
43:34
and capital, and you had big factories
43:36
and railroads were laying thousands
43:39
of miles of steel, and you
43:41
know, you were building these forges and foundries
43:43
to make cars. The modern
43:46
era, especially with technology,
43:49
there are a lot of intangibles that don't
43:52
seem to find their way to book
43:54
value. Things like patents and copyrights
43:56
and algorithms and processes
43:59
that are prepared rietary that really
44:02
are the whole value of the company,
44:04
but somehow never show
44:07
up in metrics like price to book,
44:09
which has led to some people
44:12
and I'm not positive who
44:14
to name. I don't want to mischaracterize anybody,
44:17
but some folks have said we're
44:19
mispricing companies
44:22
that operate in the tech space because
44:24
we're not giving them the appropriate credit
44:26
for all of this intellectual property. Is
44:29
that an overstatement or is there
44:31
some truth there?
44:32
I think there's more than some truth
44:34
to that. We published a papers
44:37
called the Veiled Value, and it looked
44:39
at the idea that brand
44:41
value, that all of the items
44:44
that you just delineated, we're
44:46
not being captured in Trademark's
44:50
research and development straight
44:52
across the board. When we took
44:54
a look at that, we found that
44:57
you could figure out a way
44:59
to price that into the model. So
45:02
you are absolutely right. This
45:04
is one of my bugaboos, things like GDP,
45:07
all of the metrics that we continue
45:09
to report and get obsessed about.
45:12
Basically they've lost a lot of their
45:14
meaning because they were designed
45:16
for the world you just articulated.
45:19
They were designed for manufacturing.
45:21
They were designed for physical things,
45:24
and we moved off that for many
45:26
many decades. Atoms to bitts was a
45:28
big transition, huge transition, and
45:31
so we think that we another
45:34
aspect of research right when
45:36
when we got the idea, you know, we think
45:38
we're missing something here. That's
45:41
what resulted in the paper about
45:43
brand value and goodwill
45:46
and all those things not being taken into
45:48
account by investors at all.
45:50
And so we found ways
45:53
we could do that with factors and
45:56
improved the efficacy of the underlying
45:59
modelsificantly.
46:00
I think one of the greatest quotes ever
46:03
issued by a statistics professor
46:05
is George Bachs All models
46:08
are wrong, but some.
46:09
Are useful, exactly. I quote him
46:11
all the time because he's
46:13
absolutely right. The idea
46:16
that you're going to get anything to
46:18
perfection is a fool's errand right.
46:21
I have a writer that we're working with
46:23
under O'Shaughnessy Adventures, one of our
46:25
new verticals, which is Infinite Books,
46:28
and he's got a great quote, which is perfection
46:31
is a one hundred percent tax.
46:34
Really interesting. Let's talk a little
46:36
about O'Shaughnessy Adventures starting
46:39
with your mission statement. Osv's
46:41
mission is to fuel creators in the
46:43
worlds of art, science
46:46
and technology with the advice,
46:49
data and resources they need to
46:51
stay focused and get great
46:53
ideas out of their heads,
46:56
off of their whiteboards, and
46:58
out into the world discuss
47:01
I had.
47:01
A thesis that started to develop
47:04
around twenty seventeen twenty
47:06
eighteen as I watched old
47:09
playbooks that used to work beautifully
47:12
stop working, and so
47:14
I came up with this idea that we
47:16
were in a great reshuffle
47:19
where all of the old models
47:21
were collapsing and people
47:24
were kind of freaked out. They were like,
47:26
this has worked for decades, Why
47:29
doesn't it work anymore? And I think
47:31
that one of the reasons it didn't work
47:33
anymore was because the tools,
47:36
the tech tools, and the platforms
47:39
and the Internet and all of
47:41
that put together allowed
47:44
for much more innovative
47:47
business models in a variety
47:49
of industries. Right, So if
47:52
you look at the verticals of O'shaughness
47:54
adventures, you'll see what we think. Right,
47:57
So we have what we call infinite
48:00
adventures. That's venture
48:02
capital. But I love in the old days
48:04
they used to call venture capital adventure
48:06
capital. And the one I really loved
48:09
liberation capital. Uh,
48:12
well, to find that what is what is liberation? And
48:14
I've heard the phrase, yeah in the old days,
48:16
the so called hateful eight that wanted
48:18
to leave shockly.
48:20
Right, the early days of semiconductors
48:22
and the pre in Fairchild
48:24
semi conductor.
48:25
Exactly exactly right, good
48:27
call. And back
48:30
then, the idea that a group of
48:32
engineers or even you
48:34
know, regular business people would leave
48:37
a big company that was well
48:39
funded by a bank or a series
48:42
of other investors was almost unthinkable.
48:45
And so what came to be known
48:47
as the hateful eight who created Fairchild
48:50
got pitched by a variety of
48:53
investors, external investors
48:55
saying why don't you guys to start your own
48:57
company. He finally talked them
48:59
into it, and that's when we
49:01
use the term this is your liberation
49:05
capital, where you can focus
49:07
on just what you want to focus
49:09
on, making better semiconductors.
49:12
You don't have to play any of the politics
49:14
of the big company, you don't have to answer
49:16
to people who don't really understand
49:18
what you're doing. Right, the people
49:20
in New York that might have owned it or financed
49:23
it had very little understanding
49:25
of what semiconductors were all
49:27
about in the fifties and sixties, and
49:30
so I like that part very
49:32
very much.
49:33
That's the genesis of Intel right,
49:35
as well as the whole run of other
49:37
semiconductors, can trace its roots
49:40
back to fair Child right
49:42
exactly.
49:43
And so there we're looking for
49:46
companies that we think will
49:49
expand the opportunity set
49:52
for very clever entrepreneurs
49:54
and creators. Another vertical
49:57
is infinite films. Why that, well,
50:00
I think we're approaching a period where
50:03
you can make films, documentaries,
50:06
You can use AI to augment your
50:08
filmmaking in such a way
50:10
that the people who couldn't make movies
50:13
in the past are going to be able to make
50:15
them in the future. You could legitimately
50:17
make.
50:17
A film with an iPhone. Now that you
50:20
can you couldn't do even five
50:22
years ago. Is kind of on the border.
50:24
Barry. Some of the things that I've seen as
50:28
submissions to infinite films, Oh
50:31
my god, Like literally,
50:34
I'm sixty three. If I had
50:36
seen that as a trailer for
50:38
a movie at a movie theater like
50:40
ten years ago, I would have thought, Wow,
50:43
this is amazing, this is cool. And
50:45
then the guy at the bottom says, by
50:47
the way I made this on my iPhone, that's crazy.
50:49
That really is great, And so that Unlock's
50:52
tremendous talent that never
50:55
had access to the Hollywood
50:57
infrastructure. So
51:00
our thesis is there are tons of
51:02
really creative people out there
51:04
who now have the tools to
51:07
make great movies. Another thing
51:09
I wanted to do was where are
51:11
the Rudies of movies today?
51:13
Now? Rudy's, of course, is about the
51:16
kid who goes to Notre Dame and he's five
51:18
foot nothing and weighs a buck nothing
51:20
and he gets on the team, the
51:22
Notre Dame team. Why was
51:25
that such a great movie? Because it's incredibly
51:28
inspirational, It gives
51:30
the viewer like, you know what, I can take
51:32
a shot at it, I can do it. Hollywood
51:35
seems to have completely forgotten about making
51:37
these types of movies.
51:39
And just for people who might not
51:41
remember the movie Rudy, it's
51:43
the story that drives the whole thing
51:45
and the characters. There's
51:47
not a whole lot of expensive special
51:49
effects or you know, they're
51:52
not flying out to Nepal. It's all done
51:55
pretty much on the cheap. And
51:57
that's the area of film you're looking
51:59
to explore, or narrative driven
52:02
accessible story.
52:04
Narrative driven accessible
52:06
stories that we're also
52:08
changing the underlying economics on.
52:11
So here's how we're gonna do that. Everyone
52:14
who comes and works on one of our films
52:17
is going to own a piece of
52:19
that film and back
52:21
end points, back end points. But for
52:23
everybody, we're not going to use Hollywood
52:26
accounting. Our accounting is very, very
52:28
straightforward. Here's what it
52:30
cost us to make it. What happens
52:32
after we recover those costs, You
52:35
own x percent. If we manage
52:37
to sell it or generate revenue from
52:40
it through the multiple platforms you
52:42
can put it out on, you're going
52:44
to benefit from that. The other thing
52:46
that we're going to do is we're going to give young
52:48
people a shot right
52:51
now, if you want to try to beat
52:53
Let's say you graduate from NYU Film
52:55
School and you decide you're gonna go out
52:57
to Hollywood and you're going to pitch all of these
52:59
student video its, say you want to
53:01
get good luck, because it ain't
53:03
gonna happen. Right There is almost
53:06
a guild like system out
53:08
in Hollywood where you know, it's
53:11
kind of the idea that, yeah,
53:13
I want to get in the Screen Actors Guild. How
53:15
do I do that? Well, to get in
53:17
the Screen Actors Guild, you have to be in three
53:19
movies. Well, wait a minute, how do I
53:21
get in the movie if I'm not in the Screen Actors
53:24
Guild. So there are a lot of
53:26
really old fashioned rules. And it's
53:28
not just Hollywood, by the way, it's
53:31
much of media. It's much of
53:33
all of the things that we consume every
53:36
day. And so basically
53:38
what I did was say, what
53:41
industries that I find fascinating
53:43
that I'm interested in have the
53:45
greatest arbitrage ability.
53:47
Huh.
53:49
I love that concept. And you know
53:51
it's funny you mentioned films because
53:54
that dynamic tension of
53:56
indie films. Look at how great A
53:59
twenty four has been doing amazing as
54:01
a as an independent studio.
54:04
The timing is really good, and the
54:06
technology tools, the ability
54:09
to film on a phone, edit on
54:11
your laptop and then distribute
54:13
it by uploading to YouTube
54:15
or wherever.
54:17
Barry, that's the key. There's
54:19
always cultural lag, right, you
54:21
know the S curve tech
54:23
adoption, right, it's real,
54:26
And let's change industries
54:28
and let's look at publishing. Right,
54:31
So we are launching Infinite
54:33
Books. Why well, because
54:35
the current publishing industry is
54:38
still playing under nineteen
54:40
twenty rules, not twenty twenty
54:43
rules. We no longer
54:45
have to have minuscule amounts
54:47
going to the author. We can,
54:50
because of the tech, because of our ability
54:52
to produce that book, give the
54:54
author much more of
54:56
the upside. So, for example, we're
54:59
going to give any where between depending
55:01
on what the author wants us to do for
55:03
them. It's going to always be
55:05
above fifty percent. Mostly it's going
55:08
to be seventy percent. But that's
55:10
just the start. Imagine Berry, you
55:12
write a book, you bring it to Infinite
55:14
Books, and I say, hey, Barry, what
55:16
other languages do you want this published
55:19
in? And You're like, I don't know, maybe
55:21
Spanish, maybe French. Maybe
55:24
done. Because of AI, we
55:26
can translate the entire book and
55:29
have it available for the French or Spanish
55:31
speaking markets. Even better,
55:33
let's say you want to do an audiobook and you
55:36
want to read it because you've got a great voice. I
55:38
say, Berry, do a minute on
55:40
this for me, say Express
55:43
Surprise or Anger or whatever.
55:46
It will model your voice
55:48
and you can read your book
55:50
on all the audiobooks. But what's really
55:52
cool is we can translate
55:55
your voice into French,
55:58
into Spanish, in to Russian,
56:01
into anything. Wow. And so
56:03
all of these tech advantages
56:06
are being left just lying around
56:08
on the floor, and we think that's crazy.
56:11
We're still early days of the transition,
56:14
oh very early, to technology,
56:16
to AI, to all these
56:18
changes in platforms. It's
56:20
amazing how slowly it
56:23
takes place. I think our mutual
56:26
friend Morgan Housel described
56:29
how long it took from
56:32
the Wright Brothers doing
56:34
the first test flight in Kitty Hawk before
56:36
it even made its way into newspapers.
56:38
Exactly takes forever, and it does.
56:41
And this lag, even in
56:43
our twenty four to seven always online
56:46
environment remains right.
56:48
It's like, if you think about it, it makes tons
56:50
of sense. People are habitual,
56:53
right, they get into habits, they
56:55
do all of these things. Now, I
56:57
think that the pandemic really
57:01
sped up a lot of these trends,
57:03
things like work from Anywhere. O'shaughness
57:06
Adventures is a work from anywhere enterprise.
57:09
We have people in Singapore, India,
57:13
UK, all over the world because
57:16
we can, and the
57:18
idea that we have to have
57:21
a traditional office, the
57:23
idea that we have to do any of those
57:25
traditional things goes right out the
57:27
window. It becomes a much
57:29
less costly enterprise when
57:32
you can do it this way. But we
57:35
back to infinite books like we
57:37
also are going to at the
57:39
author's decision. Right, We're not going
57:41
to force anything on our authors. But
57:44
if the author wants an
57:46
AI agent to Let's
57:48
say, for example, your new book, Let's
57:51
say if it were an Infinite Books publication
57:54
and you noted that it
57:57
quadrupled sales in Omaha, Nebraska,
58:00
how about having an AI agent find
58:02
out what podcasts in Omaha
58:05
are interested in the subject Berry's written
58:07
about. How about sending them a query
58:09
letter. How about setting them a clip from the book
58:12
and saying you really ought to have him on
58:14
your show or podcast or
58:16
write about them. In your substack. All
58:19
of the tools that are available to
58:21
us work today, and
58:25
people aren't using them, and so we
58:27
suspect that this is going to really
58:30
I hate the word revolutionize because
58:32
that's, you know, come on, but
58:34
it's certainly going to accelerate. That's
58:38
a better ride.
58:39
So I want to talk about another aspect
58:41
of O'shaughnessee ventures, which
58:44
is the fellowship program, which
58:46
I find to be absolutely
58:48
fascinating. How does this work
58:51
tell us a little bit about the O'shaughnessee Fellowship.
58:53
For most of history, a genius
58:55
could be born, live and die without
58:58
even knowing they were a genius, far
59:00
less other people knowing it. Right, We
59:03
were really bound by our geography
59:05
and by our networks, and those
59:08
networks were pretty small. Like who'd
59:10
you grow up with, who'd you go to school
59:12
with, who'd you mary? Where are your kids
59:14
going to school? What church do you go to? That kind
59:17
of stuff pretty random. You're just random
59:19
where you were born. I was just dumb. Luck was
59:21
kind of dumb. Luck. You could move, of course,
59:23
but changing your digital zip
59:26
code is a hell of a lot easier than changing
59:28
your physical zip code. But
59:30
more importantly, we now are
59:32
interconnected. I can
59:34
find somebody who's a genius who
59:36
happens to live in Bangladesh. I would
59:38
have never under the old system ever
59:41
known about that person. Now
59:44
I have the ability to know about
59:46
that person and find and
59:48
fund them. The whole idea
59:50
behind the fellowships was we
59:53
wanted to come up with something that highlighted
59:55
the fact that there are tons
59:58
millions of brilliant
1:00:00
people who in the past
1:00:03
just didn't have the right connections, didn't
1:00:06
have the right credentials, you'd name
1:00:08
it, to get into a place
1:00:10
where they could get funding, they could
1:00:12
make their idea come to life. And
1:00:14
so the idea is quite simple. We're
1:00:17
gonna find and fund them and
1:00:19
see what comes from that. I
1:00:21
think that it allows for
1:00:24
so many things, Like it allows we
1:00:26
have a guy who got one of our grants,
1:00:28
which is the smaller amount. It's ten
1:00:30
thousand. The fellowships are one hundred thousand
1:00:33
over a year, no strings.
1:00:35
No strings attached. He has a check for one hundred
1:00:37
k. Go do something interesting.
1:00:39
We don't care what it is exactly.
1:00:41
And we wanted to do no strings
1:00:44
because like we don't want Gotcha's
1:00:46
we don't want. But you've got to do
1:00:49
You've got to give us right of first refusal.
1:00:52
The way I look at it is if if we got
1:00:54
somebody so wrong that they're
1:00:56
going to take one hundred thousand fellowship
1:00:58
from us, develop something really
1:01:00
cool, decide to start a company
1:01:02
around it, and then take it to a different person
1:01:05
for funding, well, we made the mistake
1:01:07
right, because generally speaking, what
1:01:09
we're finding is they love being part
1:01:12
of the community. Because I'm
1:01:14
also a huge believer
1:01:16
in cognitive diversity, right,
1:01:19
there's a great quote that is like, no
1:01:21
matter how smart somebody is, no matter
1:01:23
how insightful, no matter
1:01:26
how brilliant, you
1:01:28
can't ask them to make a list
1:01:30
of things that would never occur to them.
1:01:33
And so essentially what happens when you
1:01:35
get all of these really bright people
1:01:38
in our fellowship and grant community
1:01:40
communicating with each other. Wow,
1:01:43
the ideas that come out of those cross
1:01:46
pollenization of ideas are
1:01:48
really extraordinary.
1:01:50
But this sounds like this is really an
1:01:52
incubator of sorts.
1:01:53
It can be, but it needn't
1:01:56
be. Here's a great example. One
1:01:58
of the guys that we gave a great aunt too, his
1:02:00
name's just that's
1:02:03
his stage name, was an
1:02:05
accountant in India who decided
1:02:07
he really had music in him
1:02:10
and he really wanted to do a
1:02:12
musical video using traditional
1:02:15
Indian songs and singing in
1:02:17
Hindi and other Indian dialects.
1:02:20
He went super viral,
1:02:23
tens of millions of downloads
1:02:25
of his song. He's being put on
1:02:27
all of their Good Morning India.
1:02:30
You know, we have Good Morning America being
1:02:32
written about in all of their newspapers.
1:02:35
And essentially that was because
1:02:38
we thought, Wow, this guy's got talent. Let's
1:02:40
see what happens. We're not incubating
1:02:43
him for anything. Right, if he goes off
1:02:45
and signs a deal with a music company,
1:02:47
we don't do music, so God
1:02:49
bless.
1:02:50
This sounds a little bit like the MacArthur Genius
1:02:53
Awards, where here's a chunk
1:02:55
of money, go be a genius.
1:02:57
There's just so much potential
1:03:00
around the world, Barry, that
1:03:03
I feel compelled to
1:03:06
amplify. Everybody loves to bag
1:03:08
on the generation before or after
1:03:10
them, Right, Listen, the
1:03:13
kids today, young people today
1:03:15
are digital natives. They
1:03:17
know how to use these tools in
1:03:20
ways that we boomers probably are
1:03:22
never going to get to. And I say,
1:03:24
let's empower them. Let's demonstrate
1:03:27
to the world that this makes
1:03:30
real practical sense. Right,
1:03:32
Now let's take somebody else who
1:03:35
is turning his grant
1:03:38
into a company. It's a guy in
1:03:40
Africa who faced
1:03:43
a problem I knew nothing about, which
1:03:45
was the cost of sanitary
1:03:47
napkins for women who are
1:03:50
menstruating is out of reach.
1:03:52
They are all imported from the West
1:03:55
and they can't buy them because they don't have enough
1:03:57
money. Well, he came up with an
1:03:59
idea where his mostly
1:04:01
female staff and researchers
1:04:04
use banana leaves and other biodegradable
1:04:07
products that they can make on
1:04:09
the ground in Africa sell
1:04:12
for a fraction of the cost that
1:04:15
the important ones
1:04:17
work just as well. Now
1:04:19
I believe he is turning that
1:04:21
into an enterprise. He's founding a company.
1:04:24
Will take a look at investing
1:04:26
in it, because of course he's asked us
1:04:28
to. It can be on the business
1:04:31
side, definitely an incubator, but
1:04:33
on the social side, on the music
1:04:35
side, on the art side. So for example, this year,
1:04:38
I really want to have a fine artist
1:04:41
get one of these grants, because
1:04:43
again I want really people
1:04:46
to be able to see there is so
1:04:48
much talent in the world, and
1:04:51
I always try to look for things to root
1:04:53
for as opposed to against. They're
1:04:57
so easy to root
1:04:59
against something thing, right, you don't have to be
1:05:01
terribly bright to say that
1:05:03
sucks. That sucks. Here's why. How
1:05:06
about doing things the other way around?
1:05:08
How about finding things you can root for?
1:05:11
And then the results have been
1:05:13
kind of like the coolest things we've ever
1:05:16
seen, like the guy going viral
1:05:18
in India, like we have. We
1:05:20
funded a guy trying to advance open
1:05:22
source quantum computing. He
1:05:24
now is a big deal in quantum
1:05:27
computing. It's a great
1:05:29
thing to do in general.
1:05:30
Tell us about some of the first few
1:05:33
you tried. Who were the
1:05:35
people that were the first couple of recipients
1:05:38
of.
1:05:38
The guy I just mentioned with
1:05:41
the quantum computing. He had me at hello
1:05:43
because I love that stuff.
1:05:44
What about people who are looking at
1:05:47
markets in the economy, I know that that's
1:05:50
a pieve of yours. Oh.
1:05:51
Absolutely, the thing there
1:05:54
is we wanted it to be significantly
1:05:57
different than our
1:06:00
traditional quant One of the reasons
1:06:02
I became so interested in machine learning
1:06:04
and AI was I viewed that as
1:06:07
the next frontier for quant The
1:06:09
dirty little secret of week quants
1:06:11
is if you really press us
1:06:14
and ask us to really explain
1:06:16
your model like you would to a five year old
1:06:19
we're using pretty much the same stuff,
1:06:22
right, So what we wanted
1:06:24
to do there was push
1:06:27
the needle as far as we possibly
1:06:29
could. But then one of the
1:06:31
first people to get one of the fellowships
1:06:34
was a married couple, Matt and Martha
1:06:36
Sharp, and what they wanted
1:06:38
to do was make a documentary about
1:06:42
non traditional schools for their
1:06:44
kids. They have a bunch of young kids below
1:06:47
the age of seven, and they
1:06:50
put out a great documentary about
1:06:52
a particular school, which was really novel.
1:06:55
And so we really are all
1:06:57
over the map in the type of per
1:07:00
groups that were willing to
1:07:02
consider. Yet another was a
1:07:05
refugee in Ireland who
1:07:07
found that she couldn't figure out
1:07:09
a way in her native language
1:07:12
to work her way through the halls
1:07:15
of the bureaucracy, to
1:07:17
figure out how do I get a place to live?
1:07:19
How do I do all of these things? So
1:07:22
we funded her to make an app. And then
1:07:24
finally another one that I just
1:07:26
love is we have a doctor who
1:07:29
came to us and said what he wanted to
1:07:31
do was make an app for an iPhone
1:07:34
or an Android where you could
1:07:37
completely non invasively, I
1:07:39
could point the phone at you, get
1:07:41
your vitals on the phone, just
1:07:43
by the camera on the phone. Really
1:07:46
yeah, And what was cool
1:07:48
for us was we really pushed
1:07:50
him. We're like, why, why, why why?
1:07:53
And finally at the end of
1:07:55
our interview with him, he
1:07:57
was near tears and he went the
1:08:00
real reason for this is my dad
1:08:02
died of a stroke and I was in medical
1:08:04
school and I didn't save him.
1:08:07
I didn't even know that he
1:08:09
had a problem. And so this
1:08:11
is why I'm so passionate about
1:08:13
this. To get a life saving
1:08:16
thing in the hands of an
1:08:19
on something that we all carry with
1:08:21
us, these smartphones, is
1:08:23
what motivated him. And on top
1:08:25
of that, looks like it could also be a great
1:08:27
business.
1:08:28
Well that's really interesting. Let's stay
1:08:31
with AI and talk about
1:08:34
medicine in particular. I'm
1:08:36
fascinated by the concept of AI
1:08:39
running through billions or even
1:08:41
trillions of molecular
1:08:44
combinations to identify
1:08:47
promising drugs, some
1:08:49
of which are already out there, some of which haven't
1:08:51
been created. But it really gives
1:08:53
us the ability to take millennial
1:08:56
worth of experimentation and do
1:08:58
it in a really very short period of time.
1:09:00
It's a world changer. The ability
1:09:03
to, as you mentioned,
1:09:05
take different molecules where there isn't
1:09:08
a drug addressing a certain problem
1:09:10
and or taking existing
1:09:12
research from drugs and repurposing
1:09:15
it. AI can go
1:09:17
into all of those spaces that
1:09:19
we humans simply can't do and
1:09:22
find the connections on an existing
1:09:24
drug. You know what, this drug
1:09:26
was originally done for malaria,
1:09:30
Well it doesn't work for malaria, but
1:09:32
it works really well for this
1:09:34
disease over here, and then
1:09:37
new drugs that the discovery
1:09:40
is going to be amazing. And you've got to
1:09:42
remember a lot of this stuff can be done
1:09:44
what they call in silico. You
1:09:46
don't have to test it on humans
1:09:49
or animals. You can test it on
1:09:52
the clone of we humans
1:09:54
that you set up in the computer.
1:09:57
And so these types
1:09:59
of thing things like I honestly don't
1:10:01
think it's an overstatement to say,
1:10:04
like this, this AI and
1:10:07
it's many use cases belong
1:10:09
up there with the wheel and fire
1:10:12
and the printing press, because
1:10:15
it is a multi use
1:10:17
technology that's going to affect everything
1:10:20
from drug discovery to
1:10:23
financial analysis. What
1:10:25
about we had trained an AI to
1:10:29
generate nothing but null sets? Right,
1:10:31
Like, if you're a medical researcher
1:10:34
and you're trying to get funding, what
1:10:36
do you want to do? You want to prove something new?
1:10:38
Right, you don't. You're not going to get funded to prove
1:10:41
you know that aspirin works, but
1:10:43
you want to find something new and you also
1:10:45
want it to be a positive
1:10:48
finding. So what happens is
1:10:51
the incentives preclude
1:10:53
a lot of brilliant scientists from looking
1:10:55
for things that don't work. And
1:10:58
yet, like the dog that didn't
1:11:00
bark in Sherlock Holmes, there's
1:11:02
a lot of really cool
1:11:05
information, useful information
1:11:08
via negativia. And so
1:11:10
one of the things that we want to do is
1:11:13
just have a large language model,
1:11:15
churnout hypothesis
1:11:17
after hypothesis that is
1:11:20
going to generate a null set, publish
1:11:22
them to a database that all scientists
1:11:24
can have access to. Because there's
1:11:26
a wealth of information in
1:11:28
the stuff that doesn't work. Here
1:11:31
are things you don't want to waste your time exactly.
1:11:33
Let's talk a bit about stability
1:11:35
AI. You're on the board of directors,
1:11:38
you're the executive chair, and you started
1:11:40
back in September twenty twenty two. Pretty
1:11:43
good timing. Tell us a little bit about
1:11:45
what stability AI does and
1:11:47
how does this relate to the rest of O'Shaughnessy
1:11:50
ventures. So stability AI
1:11:53
builds foundational open
1:11:55
source models. I had
1:11:57
a very pointed
1:12:00
point of view that with
1:12:02
a technology this powerful, I
1:12:05
did not want it controlled by
1:12:07
a panopticon controlled by
1:12:09
a few, and
1:12:11
I saw that with that kind
1:12:14
of power could come some pretty
1:12:16
negative externalities. And
1:12:19
so Stability AI was
1:12:22
the one that really caught
1:12:24
my eye because they really were the ones
1:12:26
who shot the gun. Back in the
1:12:28
summer of August of twenty
1:12:31
two, they released a stable
1:12:34
diffusion model which generates images,
1:12:36
right, but they did something
1:12:39
that no one had done before. They released
1:12:41
that model with all of its weights.
1:12:44
Now, not to get too geeky
1:12:46
here, but the only way people
1:12:48
can build on that type of model is
1:12:51
to know what the weights are. And
1:12:53
so what they did was
1:12:56
show it all. They released the whole thing,
1:12:58
full open source, driven source,
1:13:01
fully transparent, and bury
1:13:04
the Cambrian like explosion
1:13:07
of creativity that happened
1:13:09
almost immediately really
1:13:12
proved to me. Yeah, back
1:13:14
to cognitive diversity, right, when
1:13:16
you allow all of these clever people
1:13:19
the ability to play with it, to
1:13:21
tinker with it, you get
1:13:23
a much better model. For example, that's
1:13:26
why Linux runs the web.
1:13:28
Linux is open source, right, and
1:13:31
it does so because a bunch of
1:13:33
different people work on different problems.
1:13:36
And so my point of view
1:13:38
was I'm all for the open I use open
1:13:40
AI. I use all of the commercial
1:13:43
Uh.
1:13:44
What are some of the commercial apps?
1:13:46
So perplexity, I love perplexity.
1:13:49
It's on my phone's open
1:13:52
AI. I'm looking at
1:13:54
Claude, the new Claude that you
1:13:56
know.
1:13:56
Perplexity can be driven by either
1:13:59
Claude or there's like
1:14:01
four different engines.
1:14:02
That which really interesting. One of the things
1:14:04
I love about perplexity.
1:14:06
It's just so great and it's cheap
1:14:08
and it's so useful. Exactly every
1:14:10
interview I do, I don't
1:14:13
start with perplexity, I finished
1:14:15
with perplexity and what did I
1:14:17
miss?
1:14:18
What did I get wrong?
1:14:19
Although you still have to be careful because every
1:14:21
now and then, like O'Shaughnessy
1:14:24
is not the rarest of names, you
1:14:26
know. I had Bill Dudley, former
1:14:29
New York Fed chair, and I learned
1:14:32
that he was a running back in the NFL
1:14:35
in the forties, which is kind of interesting
1:14:37
because he wasn't born till the fifties. But
1:14:40
every now and then something will pop up that
1:14:44
is a little off. I love the phrase hallucination
1:14:46
for that. What else do you use besides
1:14:48
perplexity and chatching.
1:14:50
The stability
1:14:52
AIS for his models. Are
1:14:54
they available? Are
1:14:57
they accessible to the lay person? Like, that's
1:14:59
the beauty of they are,
1:15:01
but through different APIs. We
1:15:03
really wanted to focus on being
1:15:05
the builder, right, so we
1:15:08
did not want to try to compete in
1:15:11
the direct to consumer space.
1:15:14
And so what we're focusing on is
1:15:16
multimodals, including generative
1:15:19
models, including specific
1:15:22
models for medical research, obviously,
1:15:25
generative art models, movie
1:15:28
models, et cetera. The
1:15:30
thing I wanted to mention when
1:15:32
you were talking about perplexity in it coming up
1:15:34
with I also passionately
1:15:36
believe that the
1:15:39
models that are going to win, or
1:15:41
not the models, the approach
1:15:44
that's going to win is human
1:15:46
plus machine, so
1:15:49
called Senator model. I
1:15:51
think that you're going to see, you know, we're going
1:15:53
to see a deluge of AI
1:15:56
only generated stuff, content,
1:15:58
movies, et cetera. And to be honest, most
1:16:00
of it's gonna suck. Right. The
1:16:03
magic comes when you add
1:16:05
a human in the loop. The magic
1:16:08
comes by being able
1:16:10
to partner with that and co create
1:16:13
and sometimes iterate on your own stuff.
1:16:16
Like you said, the ideas
1:16:18
that you can generate through putting
1:16:21
your own stuff into the
1:16:23
various models is really cool. We
1:16:25
invest in a
1:16:28
startup called wand and what
1:16:30
they do is it's for graphic artists
1:16:32
and it's an AI, but it has an actual
1:16:35
tool, thus the name wand And.
1:16:38
What the artist is able to do is feed
1:16:40
their own work into
1:16:42
the model and then ask it hey,
1:16:45
spin out variations on it, and
1:16:47
then the artists will look at it and say, wow,
1:16:49
I never thought about it that way. That's really cool.
1:16:52
And then he or she will iterate,
1:16:54
iterate, send it back and this is an
1:16:56
iterative process. But what's
1:16:59
really cool is they end
1:17:01
up in places. We had one artists say
1:17:03
to me, I would never have
1:17:06
thought to do it this way, but
1:17:08
I absolutely love it. It's
1:17:10
his work. He's iterating on his own
1:17:12
work, but he's using a tool,
1:17:15
the WAND, that makes it infinitely
1:17:18
easier for him to get these great
1:17:20
ideas. Huh.
1:17:21
Really interesting. Last
1:17:24
question before we jump to our
1:17:26
favorites, we ask all our guests, which
1:17:28
is I want to bring this back to
1:17:31
stocks. I know thanks
1:17:33
to perplexity as an example, but
1:17:36
there are lots of other tools.
1:17:38
I find myself going to Google
1:17:40
a whole lot less than I used to, and
1:17:43
in fact, the Google
1:17:45
search results like, suddenly
1:17:48
you realize these are crude.
1:17:50
They're much less useful
1:17:52
than they used to be. They're festooned
1:17:55
with a lot of advertising and
1:17:57
a lot of Google in internal
1:18:00
products dominate that first
1:18:02
page.
1:18:04
What else is AI? What other
1:18:07
companies?
1:18:07
What other sectors might
1:18:09
AI affect, either positively or
1:18:11
negatively?
1:18:12
Well, honestly, how much time
1:18:15
do you have it? I think that AI is
1:18:17
going to transform virtually every
1:18:19
industry. And one
1:18:21
of the things that people they
1:18:23
get afraid when they hear that, and my
1:18:26
view is quite different. It's going
1:18:29
to transform for a lot of industries,
1:18:31
the pure drudge work, the
1:18:34
pure copy and paste stuff.
1:18:36
What do you want? Do you like copying and
1:18:38
pasting? I hate it? And so
1:18:41
it also is going to be able to
1:18:43
create jobs that we can't even
1:18:45
conceive of right now. Right
1:18:48
like two years ago, would you have known
1:18:50
what a prompt engineer was? No,
1:18:52
I certainly wouldn't have, right, and
1:18:54
yet there's a lot of people doing really
1:18:57
well pursuing that is a
1:18:59
career. And so I
1:19:01
think that entertainment is going
1:19:03
to be materially affected
1:19:05
media, materially affected search,
1:19:08
as you well point out, like you
1:19:11
can do a customized search just
1:19:13
for Barry and it, you know, depending
1:19:15
on how much information you want to give that
1:19:17
AI about yourself. You're going
1:19:20
to be at a place where you're going to be able to say, hey,
1:19:22
what was that place that I had lunch with Jim
1:19:24
last time? We both really really liked it. I'd
1:19:27
like to go there again and guess what, it's
1:19:29
going to give you the name and address of
1:19:31
that restaurant. Because
1:19:34
it has access to your calendar,
1:19:36
It has access to all of that type of
1:19:38
stuff.
1:19:39
It feels like I'll
1:19:41
never forget. I tweeted out this
1:19:43
really interesting Roman
1:19:46
Pizza place, and Roman Pizza is
1:19:48
a different type of and
1:19:50
I just you know, I used Sarah
1:19:53
to speak into the iPhone. Hey
1:19:55
we had a fend. This is really different
1:19:57
than your usual pizza. And
1:19:59
some how it showed
1:20:01
up on Twitter as woman Pizza
1:20:04
and like, wait, I'm standing right in
1:20:06
front of the place. Any correlation
1:20:09
between my geotag
1:20:11
and business I'm in front of it just
1:20:13
felt like technology should
1:20:16
have figured that out. Yeah, what you're saying
1:20:18
is that sort of access
1:20:21
to your contacts, access to your where
1:20:24
you are, access to your calendar, once
1:20:27
there's an intelligent agent running all
1:20:29
of that. A lot of these sort
1:20:31
of silly why can't Siri
1:20:33
talk with this person, Why can't
1:20:35
Alexa? It just seems like the
1:20:39
pre AI era was
1:20:41
filled with a lot.
1:20:42
Of pretty dombai. It's
1:20:44
starting to get smarter. Yeah,
1:20:46
And that's the thing going back to your Right
1:20:49
Brothers example. You know when
1:20:52
the Right Brothers did that very
1:20:54
brief flight, it was only a
1:20:56
matter of eight yeah, I think it was twelve
1:20:58
seconds, and I think they went like one hundred
1:21:00
and odd feet. Like
1:21:03
you could see why a lot of people would going, Eh,
1:21:06
they didn't accomplish much, but I
1:21:08
like the person who was watching and
1:21:10
said, this changes everything,
1:21:13
And so that's kind
1:21:16
of how I see AI. Of course,
1:21:18
we're in the early innings of this, and
1:21:20
of course it's going to this is
1:21:22
the worst you're ever going to see it. It's going to
1:21:24
improve, improve, improve. But
1:21:26
the other thing I want to really underline here
1:21:29
is it's the quality of the data
1:21:31
that you train your AI on that
1:21:34
determines its value to you. And
1:21:37
one of the big reasons I'm a huge believer
1:21:39
in private AIS is
1:21:41
that you will feel if you
1:21:43
know that no one else can have access to
1:21:46
that right, you're going to give it a
1:21:48
lot more access to things
1:21:50
than you might otherwise. That's
1:21:52
happening right now. Wow. And
1:21:55
so one of the things,
1:21:57
you know, a lot of people see this as, you know,
1:21:59
like the great model
1:22:01
that will figure everything out. I
1:22:04
don't see it that way at all. I see it
1:22:06
as a lot of smaller but
1:22:08
incredibly useful AI
1:22:11
agents doing specific
1:22:13
things for each of us. Again, Canvas
1:22:16
fits in beautifully. Here. We are now
1:22:18
in an era of mass customization.
1:22:22
We are in an era where it's going
1:22:24
to be able to design it just
1:22:27
for you and your likes and dislikes.
1:22:30
That's really profound when
1:22:32
you think about.
1:22:33
It, really fascinating. So let's
1:22:35
jump to our speed round, our favorite
1:22:37
questions. We ask all of our guests, starting
1:22:40
with what has been keeping you entertained
1:22:42
these days? What are you either watching or
1:22:45
listening to?
1:22:46
So we rewatched
1:22:48
True Detective, my wife
1:22:50
and I I would highly recommend
1:22:52
rewatching the first season
1:22:55
of that. It was brilliant. It
1:22:58
led us into a reward of the
1:23:00
entire series. And now
1:23:02
we're on number three, the second one.
1:23:05
Here's one of the funny things, like in memory,
1:23:07
I kind of my wife and I were both kind of like, yeah,
1:23:09
that second one wasn't very good. It was
1:23:12
good, and so we're
1:23:14
doing that Masters of
1:23:16
the air that's on. Yeah,
1:23:19
great, really loving that. I loved
1:23:21
Band of Brothers, so we're
1:23:23
both really really liking that.
1:23:26
And then we are also
1:23:29
watching a series,
1:23:32
or I guess I should say rewatching
1:23:34
a series which kind
1:23:37
of kicked off the idea of
1:23:39
the Golden Age of television. It was one of the
1:23:41
earlier ones. I'm not the Sopranos, but
1:23:44
The Wire.
1:23:45
Now, I recall The Wire being
1:23:47
very brutal and difficult.
1:23:49
It watch it is, But
1:23:52
what's so cool if you choose to watch
1:23:54
it again, you see that
1:23:57
the reason it kicked off that kind
1:23:59
of tea was because it
1:24:01
was brutally honest about
1:24:04
things. It wasn't trying to lie to you about
1:24:06
anything, and the characters
1:24:09
are incredibly complex,
1:24:11
even though even the evil
1:24:13
guys are incredibly complex,
1:24:17
And so watching it now
1:24:19
from the vantage point of like twenty
1:24:22
years or more, it's
1:24:25
really amazing.
1:24:27
Really interesting. Tell us about your
1:24:29
mentors who helped to shape
1:24:31
your career.
1:24:32
Primarily, I
1:24:34
would list my grandfather. I
1:24:36
was lucky enough, he was very successful
1:24:40
in the oil industry, and I
1:24:42
am the youngest of the third generation at
1:24:44
least the males. I have one younger
1:24:47
female cousin and she's just a few
1:24:49
months behind me. But I
1:24:51
lived in the same town my grandfather did,
1:24:54
and after my grandmother died, he
1:24:56
would come to our house twice a week for
1:24:58
dinner and literally
1:25:00
I would literally sit at his knee.
1:25:03
And he was a wonderful storyteller.
1:25:06
He was a wonderful teacher,
1:25:10
and he taught me this idea of
1:25:12
premeditating that I have
1:25:15
written a lot about and use all
1:25:17
the time. Another was
1:25:19
a wonderful man, not related to
1:25:21
me at all, by the name of Jim Myers.
1:25:24
Any entrepreneur, you hit some rough spots,
1:25:26
sure, and I had hit a really rough
1:25:28
spot and was basically broke and
1:25:31
trying to pay for a house
1:25:34
because we'd moved to Greenwich and
1:25:37
keep my business afloat and all of that,
1:25:40
and the banks are like, dude, like, you're
1:25:42
an entrepreneur. This is back in the nineties.
1:25:45
Yeah, sorry, we're not going to give you a
1:25:47
mortgage. He stepped in and he's
1:25:49
like, Jim, I believe you're going to be tremendously
1:25:52
successful and gave
1:25:55
me one on handshake, which I
1:25:57
was able to repay rapidly.
1:25:59
But more than that, just being
1:26:02
a super high quality man, he
1:26:05
taught me more about real business
1:26:08
than any textbook because
1:26:11
I was young, right, and I
1:26:13
started with him when I was in
1:26:15
my early twenties and
1:26:18
just an amazing man.
1:26:20
And then finally the other
1:26:23
mentors that I would say are like
1:26:26
the greatest minds of history. I love to read.
1:26:28
I particularly like to read
1:26:31
biographies about people I admire.
1:26:33
And you know what, Barry life was
1:26:35
not easy. We remember them now, right,
1:26:38
like, oh, they were this huge success.
1:26:41
When you read their biographies, you see
1:26:43
they went through a lot of muck
1:26:46
to get where they got and so kind
1:26:48
of universal lessons.
1:26:49
So perfect segue. Let's talk
1:26:51
about some of your favorite books and
1:26:54
what are you reading right now?
1:26:56
So right now I am reading
1:26:58
about four different books
1:27:01
and I which.
1:27:03
By the way, is an occupational hazard
1:27:05
for folks like us, because there's
1:27:07
always a book I'm prepping for a podcast,
1:27:09
there's a book I'm reading for work,
1:27:12
and then there's a book I'm just like, I'm going to relax
1:27:14
and read this.
1:27:15
Yeah, so for fun. Right
1:27:17
now I'm reading Burned book by
1:27:20
Karra What's her last Swisher?
1:27:23
Which I find very interesting. She's
1:27:25
always fascinating, yeah, kind of an inside
1:27:28
look. My only comment
1:27:30
there was she might be a little
1:27:32
guilty of the things that she accuses
1:27:34
the people she doesn't like are. But
1:27:37
other than that, it's a fun and kind of a rollicking
1:27:40
read. I am reading
1:27:43
or rereading several of the
1:27:45
books from Will Durant's Story
1:27:47
of Civilization, which
1:27:49
I read as a kid or a young man,
1:27:52
loved and thought,
1:27:54
you know what, we moved recently,
1:27:56
and so I was going through all my books and
1:27:58
I found that I'm like, I should reread
1:28:01
some of these just to see if it still stands
1:28:03
up. Barry, It's still stuff,
1:28:06
really really stands up. And
1:28:08
then just finished an
1:28:11
additional biography about
1:28:13
Teddy Roosevelt Teddy Rex
1:28:16
And then finally I'm reading
1:28:18
a lot about AI
1:28:20
and scientific development. The
1:28:22
book I'd recommend there is written
1:28:25
by a pair of authors, one
1:28:27
an AI expert, the other a great
1:28:29
storyteller, and it's called AI
1:28:32
twenty forty one, Ten
1:28:35
Visions of Our AI Future.
1:28:37
Huh, highly recommend.
1:28:38
I'm going to check that out. We've been talking
1:28:40
about the Ripe Brothers. Did you ever read the David
1:28:43
McCullough biography of the Ripe Brothers?
1:28:44
I did.
1:28:45
Fascinating, right, really really really fascinating.
1:28:48
And our final two questions, what
1:28:50
sort of advice would you give to a recent
1:28:53
college graduate interested
1:28:55
in a career in either quantitative
1:28:57
analysis, finance as
1:29:00
management.
1:29:00
What's your advice for them? My advice
1:29:03
is to focus
1:29:05
on the parts of learning
1:29:08
that might not be
1:29:11
included in a business or finance
1:29:13
degree. My line is that
1:29:15
markets change second by second,
1:29:18
but human nature barely
1:29:20
budgees millennia by millennia.
1:29:23
Arbitraging. Human nature is the
1:29:25
last sustainable edge
1:29:27
in investing. And so if
1:29:29
you read about evolutionary psychology
1:29:32
and biology, regular psychology
1:29:34
and biology and history, what
1:29:36
you're going to see is no, history
1:29:39
doesn't repeat, but it rhymes,
1:29:41
and you can see in you know, all you
1:29:44
got to do is go read a book about the south
1:29:46
Sea Scandal where Isaac Newton,
1:29:48
one of the most brilliant guys of his era, lost
1:29:50
a fortune, causing him to lament
1:29:53
that he could measure the motion of heavenly bodies
1:29:55
but not the madness of men. And guess
1:29:58
what we're not changing. So
1:30:00
you can read it in a market related
1:30:03
way, or just understand
1:30:05
human nature better, you're
1:30:07
going to be miles ahead of
1:30:09
the people who are just studying
1:30:11
math or finance or economics.
1:30:15
Really interesting, and our final question,
1:30:18
what do you know about the world of investing today
1:30:21
you wish you knew forty or so years
1:30:23
ago when you were first getting started.
1:30:26
I think maybe just the advice that I just gave,
1:30:28
I wish that I would have known forty
1:30:30
years ago that markets
1:30:33
are market prices are
1:30:35
determined by human beings,
1:30:39
and if you are ignorant
1:30:41
of all of the ways that
1:30:43
we let things affect us,
1:30:46
from whether we're hungry or not, or
1:30:48
whether we're angry, or whether we're
1:30:50
calm, I would have understood
1:30:54
that it was not just numbers on a
1:30:56
page, that markets are
1:30:59
full blooded, almost
1:31:01
human like things because they're
1:31:03
driven and created by humans.
1:31:06
If I could have told Jim
1:31:08
of age twenty three that
1:31:11
it would have hastened but also improved
1:31:15
the pretty circuitous path that
1:31:17
I took to becoming a quat
1:31:20
really interesting.
1:31:22
Thank you, Jim for being so generous with
1:31:24
your time. We have been speaking with
1:31:26
Jim O'Shaughnessy, founder
1:31:28
of OSAM Asset Management
1:31:30
and currently CEO and founder of
1:31:33
O'Shaughnessy Ventures and host
1:31:35
of the Infinite Loops podcast. If
1:31:38
you enjoy this conversation, well.
1:31:41
Be sure and check out any of.
1:31:42
The five hundred previous discussions
1:31:45
we've had over the past ten years.
1:31:47
You can find those at.
1:31:49
iTunes, Spotify, YouTube, wherever
1:31:52
you find your favorite podcast.
1:31:55
Be sure and sign up for my new podcast
1:31:57
At the Money, where we speak with
1:31:59
an X and give you information
1:32:02
on a topic relative to your money
1:32:05
in short eight to twelve minute
1:32:07
batches. You can find those in
1:32:10
the Masters in Business podcast
1:32:12
feed, or wherever you get your
1:32:14
favorite podcasts. I
1:32:16
would be remiss if I did not thank the Cracked
1:32:18
team that helps us put these conversations
1:32:20
together each week. My audio
1:32:23
engineer is Sebastian Escobar, My
1:32:25
producer is Anna Luke. Sean Russo
1:32:28
is my head of research. Attika Valbrun
1:32:31
is my project manager. Sage
1:32:33
Bauman is the head of podcasts.
1:32:36
I'm Barry Riddolts. You've been listening
1:32:38
to Masters in Business on Bloomberg
1:32:40
Radio.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More