00:00:00
hello and welcome back to patrick boyle
00:00:02
on finance so this week in the news
00:00:05
legendary investor jim simons has
00:00:07
stepped down as the chairman of
00:00:09
renaissance technologies
00:00:11
which is of course the most successful
00:00:13
quant hedge fund in history
00:00:15
now simons hasn't been in charge of the
00:00:16
day-to-day running there
00:00:18
for probably over a decade but he has
00:00:21
stayed on as the chairman of the fund up
00:00:23
until this january
00:00:25
jim's retirement marks the end of an era
00:00:27
in finance simon's career and the fund
00:00:30
that he launched
00:00:31
proved that the finance textbooks which
00:00:33
claim that markets are perfectly
00:00:35
efficient were wrong
00:00:37
obviously the trading strategies at
00:00:39
renaissance are secret
00:00:40
but let's look at simon's career and see
00:00:43
what lessons we can learn
00:00:45
so simons didn't have a fancy upbringing
00:00:47
he was a middle class kid from
00:00:49
massachusetts
00:00:50
his father was the manager of a shoe
00:00:52
factory in brookline massachusetts just
00:00:55
outside of boston he excelled at school
00:00:57
graduating high school in three years
00:01:00
then skipping the first year of
00:01:01
mathematics while at mit
00:01:04
and eventually earning his phd from
00:01:06
berkeley at the young age of 23.
00:01:08
so simons was clearly a very smart guy
00:01:11
but he was also
00:01:12
focused and hard-working and he found
00:01:15
something he was interested in
00:01:16
mathematics and really pursued it
00:01:19
simon's then went on to teach
00:01:21
mathematics for a few years at harvard
00:01:23
university
00:01:24
before famously working at the nsa as a
00:01:27
codebreaker
00:01:28
while he was at the nsa simon's noticed
00:01:31
that his brilliant colleagues
00:01:33
weren't hired in for their experience
00:01:35
but instead just for their sheer brain
00:01:37
power and he realized that you can teach
00:01:39
a person how to do a job
00:01:41
but you can't necessarily teach someone
00:01:43
to be smart and this realization was one
00:01:46
of the keys to his later success
00:01:48
when you look at the reasons that
00:01:50
renaissance is so successful
00:01:52
it's not actually down to just one great
00:01:54
genius the success was driven by simon's
00:01:57
hiring a series of the best people that
00:01:59
he could find
00:02:00
it reminds me a little bit of the career
00:02:02
of david bowie he was a great musician
00:02:04
on his own
00:02:05
but he always picked the most capable
00:02:07
musicians to work with
00:02:09
in various styles of music and then he
00:02:11
stayed relevant by moving with the times
00:02:14
and we see this exact same
00:02:15
characteristic when we look
00:02:17
at jim simons while using computers to
00:02:20
break codes for the nsa
00:02:22
simons began thinking that you could
00:02:24
possibly use them
00:02:25
and use these mathematical approaches to
00:02:27
analyze and trade markets
00:02:30
he did some early investing and trading
00:02:32
from that point forward
00:02:34
but he really didn't hit his stride for
00:02:36
quite a while
00:02:37
after his time at the nsa simons was
00:02:40
hired to lead the math department at
00:02:42
suny stony brook i think he was around
00:02:44
30 years old
00:02:45
at the time his contributions to
00:02:48
geometry and topology while he was there
00:02:50
led to him winning the oswald veblen
00:02:53
prize in geometry in 1976.
00:02:56
this is a prize that's only awarded
00:02:58
every three years
00:03:00
it's a very big deal in the world of
00:03:02
mathematics
00:03:03
while he was at stony brook his talent
00:03:05
for hiring
00:03:06
became clear he quickly built a
00:03:09
world-class mathematics department at
00:03:11
stony brook
00:03:12
including hiring james axe away from
00:03:15
cornell university and
00:03:16
ivy league university axe had won the
00:03:19
prestigious coal prize
00:03:21
in number theory at this point in
00:03:23
simon's career
00:03:24
he was 40 years old he'd been a star
00:03:27
cryptographer
00:03:28
scaled the heights of mathematics and
00:03:30
academia and he then quit
00:03:33
much to the surprise and disgust of his
00:03:35
colleagues
00:03:36
to start his fourth career opening an
00:03:39
investment firm which he named
00:03:40
money metrics a combination of money and
00:03:43
econometrics
00:03:44
we can see at this point in the story
00:03:47
that he wasn't someone to just rest on
00:03:49
his laurels he looked for new challenges
00:03:52
his mathematician friends thought that
00:03:53
this was just crazy but he did it anyway
00:03:56
he just
00:03:56
followed his vision he had an idea he
00:03:58
wanted to go with it
00:04:00
you'd probably guess that he had great
00:04:02
success right away right because
00:04:04
we wouldn't be telling his story if
00:04:06
things didn't work out
00:04:08
but that's not exactly how things went
00:04:10
he did all of the right things he hired
00:04:12
the smartest people initially partnering
00:04:14
with an old codebreaking friend of his
00:04:17
leonard baum
00:04:18
who had co-developed the bomb welsh
00:04:20
algorithm one of the most notable
00:04:22
advances
00:04:23
in machine learning they worked on
00:04:26
developing a probabilistic
00:04:28
approach to trading using hidden markov
00:04:31
models a year after that he recruited
00:04:34
james
00:04:34
axe the star professor that he had lured
00:04:37
to stony brook he heard him a way
00:04:39
to join them this was 1978.
00:04:43
they had very little luck early on
00:04:45
building quantitative trading models
00:04:47
and they drifted towards trading on news
00:04:50
and on their instincts about markets
00:04:53
james act stuck the most to the quant
00:04:55
approach
00:04:56
and developed some crude trend following
00:04:58
strategies
00:04:59
but they weren't necessarily what you
00:05:01
would have expected from a team of
00:05:03
brilliant mathematicians
00:05:05
they weren't exactly innovative either
00:05:07
even at the time
00:05:08
in 1982 simon's renamed the firm
00:05:12
renaissance technologies away from money
00:05:14
metrics
00:05:15
partially to reflect his interest in
00:05:17
making technology venture capital
00:05:20
investments
00:05:21
from 1978 through to 1984
00:05:24
they didn't really make an awful lot of
00:05:26
progress and they didn't they definitely
00:05:28
didn't earn the kind of returns that you
00:05:30
associate today with jim simons after a
00:05:33
40
00:05:34
loss in bonds in 1984 an
00:05:37
automatic clause in their agreement was
00:05:39
triggered where lenny bomb's positions
00:05:42
were liquidated
00:05:43
ending the partnership a year later in
00:05:46
1985
00:05:47
james axe moved to california forming a
00:05:50
new company called
00:05:51
axcom limited and simons would receive
00:05:54
a quarter of the profits for providing
00:05:56
trading help
00:05:57
and dealing with the firm's clients he
00:06:00
would continue on with his vc deals that
00:06:02
was kind of his main thing at this point
00:06:05
while the california team would focus on
00:06:07
quant trading
00:06:08
now some good things did happen over
00:06:10
this slow period
00:06:12
they did hire some good people sandor
00:06:15
strauss for example was a stony brook
00:06:17
professor
00:06:18
who was brought on as a computer
00:06:20
specialist in 1982.
00:06:22
it turned out that he had a passion for
00:06:24
data and he began
00:06:26
collecting market data wherever he could
00:06:28
get it he would buy pricing data from
00:06:30
exchanges from the federal reserve
00:06:32
he would extract it from old newspapers
00:06:35
and he would organize and clean the data
00:06:38
he even started collecting his own tick
00:06:40
data from a market feed back then
00:06:43
something that really no one else was
00:06:45
doing at the time
00:06:46
henry laufer was another stony brook
00:06:48
mathematician who came on board back
00:06:50
then as well
00:06:51
and was developing computer simulations
00:06:54
to test
00:06:55
strategies simon's once again at the
00:06:57
good sense to hire these bright
00:06:59
and driven people and to let them do the
00:07:02
things they did best he didn't
00:07:04
hire average people and micromanage them
00:07:07
he hired the best people that he could
00:07:08
find
00:07:09
and then let them do their thing within
00:07:11
reason
00:07:12
by 1986 xcom was trading 21 different
00:07:16
futures contracts using a mix of
00:07:18
quantitative strategies
00:07:20
and judgment calls they had mixed
00:07:22
results and they weren't really doing
00:07:24
very much
00:07:25
new research to improve their trading
00:07:27
strategies
00:07:28
the team didn't yet fully believe in
00:07:30
their big idea the idea of quant trading
00:07:33
and they were often overriding the
00:07:36
system or trading just based on instinct
00:07:39
in 1987 they brought on a guy named
00:07:41
renny carmona
00:07:43
who was able to work with the data
00:07:45
strauss had collected
00:07:46
and he started building a model that
00:07:48
looked for similar market environments
00:07:50
in the past
00:07:51
and then built forecasts for the future
00:07:54
based upon that
00:07:56
this was an early machine learning
00:07:58
approach and the team were initially
00:08:00
very uneasy with this
00:08:02
uh to begin with there were lots of
00:08:04
squabbles within the group back then
00:08:07
and in truth squabbling between team
00:08:09
members has been an
00:08:10
issue up until the present day at rentec
00:08:14
in truth it's just never going to be
00:08:16
easy to manage the type of smart hard
00:08:18
charging and
00:08:19
frankly prickly people that you find in
00:08:22
organizations like this
00:08:24
in 1988 elwyn berlecamp joined the team
00:08:28
another brilliant mind who had worked
00:08:30
with claude shannon at
00:08:31
mit and with john kelly at bell labs
00:08:35
the team that had developed the kelly
00:08:37
criterion influencing ed torp the father
00:08:40
of quantitative trading
00:08:41
i made a video about that topic a few
00:08:44
months ago
00:08:45
interestingly ed torp happened to be
00:08:48
winding up his fund right around the
00:08:50
time that these guys were getting going
00:08:53
they weren't the first obviously to try
00:08:55
a quantitative approach
00:08:56
they just went on to become the very
00:08:59
best
00:08:59
added so up until now returns had not
00:09:02
been amazing and
00:09:03
investors in the fund had grown
00:09:05
frustrated with jim's venture capital
00:09:07
investments
00:09:08
so he sold all of those off and launched
00:09:11
a new fund
00:09:12
and this fund was named medallion in
00:09:14
honor of the mathematical medals that
00:09:16
both simon's and axe had won
00:09:18
they had around 20 million dollars in
00:09:21
assets under management at the time
00:09:23
and they had the right team in place it
00:09:26
was 1988 they'd been working on this for
00:09:28
around 10 years at this point
00:09:31
the first few months for medallion fund
00:09:33
did not go well
00:09:34
they had early losses of around 30 and
00:09:37
simon's had to halt trading
00:09:39
he and acts were at each other's throats
00:09:42
they hired lawyers and hurled lawsuits
00:09:44
at each other
00:09:45
and so early on it just looked like this
00:09:47
wasn't going to work everything would
00:09:48
have to wind up
00:09:50
burl camp in order to keep things going
00:09:53
bought most of axe's shares
00:09:55
in the partnership i think x had hung on
00:09:58
to 10 percent
00:10:00
berle camp now owned 40 and simon's own
00:10:03
25
00:10:04
and the rest were split up amongst the
00:10:06
team they shut down
00:10:07
ax's trend following system now focusing
00:10:11
on burl camp
00:10:12
and carmona's black box approach it's
00:10:14
worth noting that simons was nervous
00:10:16
about this approach at
00:10:18
the time as the signals didn't seem to
00:10:19
make much sense to him they weren't sort
00:10:21
of linear signals that could be
00:10:23
clearly understood as to why the system
00:10:25
was doing what it was doing
00:10:27
there were other difficulties in the
00:10:28
early days too
00:10:30
like when they discovered that floor
00:10:32
traders were front running their trades
00:10:34
in the pits
00:10:34
which was cutting into their ability to
00:10:37
generate profits
00:10:38
1989 was unfortunately a down year for
00:10:41
medallion investors in fact it was the
00:10:43
only down year that they had but i'm
00:10:45
getting ahead of myself
00:10:47
in 1990 they returned 55
00:10:50
and that's after fees of 5 and 20. the
00:10:53
return before fees was
00:10:54
i think around 78 at this point they
00:10:58
were managing 45 million dollars
00:11:00
they traded commodities and currencies
00:11:03
and had an
00:11:04
average holding period of around a day
00:11:06
and a half
00:11:07
and they'd continue to trade this
00:11:08
product mix with roughly this holding
00:11:11
period for another decade
00:11:13
simon's after the big up year set out to
00:11:15
raise more investor capital in the early
00:11:18
1990s
00:11:19
and he didn't really have much luck at
00:11:21
all investors felt that his fees were
00:11:23
much too high
00:11:24
that he didn't have a long enough track
00:11:26
record and they were outraged that
00:11:28
simon's would not explain how the models
00:11:30
worked and that's often a difficulty for
00:11:33
quant traders throughout the 1990s they
00:11:36
continued to hire well
00:11:38
and they consistently improved their
00:11:40
approach the system at this point was a
00:11:43
living thing it was constantly being
00:11:45
improved
00:11:46
they analyzed things like slippage they
00:11:48
improved analysis and execution
00:11:51
the team had lots of discussions trying
00:11:53
to understand
00:11:54
if their strategy was winning who out
00:11:57
there was losing
00:11:58
it didn't appear to be floor traders or
00:12:00
hedge funds most of these guys seem to
00:12:02
be making money at the time
00:12:04
and so they came to the conclusion that
00:12:07
they were picking up the pennies that
00:12:09
other investors were dropping
00:12:11
through being too cocky and through
00:12:12
making behavioral mistakes
00:12:15
they decided it was most likely dentists
00:12:17
who were losing the money that they were
00:12:19
making
00:12:20
most investors approached the market
00:12:21
filled with their own cognitive biases
00:12:24
they let their emotions get the better
00:12:26
of them the systematic approach avoided
00:12:29
emotion
00:12:30
the computer never had too much to drink
00:12:32
the night before trading and it
00:12:34
never traded badly because it had an
00:12:36
argument with its girlfriend
00:12:38
they found that they did best in very
00:12:40
turbulent markets
00:12:41
as in times of stress human behavior
00:12:44
became even more predictable
00:12:46
as the fund grew they stopped taking on
00:12:48
new investors
00:12:50
they increased the fees for existing
00:12:52
investors and they put a lot of work
00:12:54
into modeling slippage
00:12:56
when you're trading in very large size
00:12:58
your trades start to move the market
00:13:01
and they aim to be invisible in markets
00:13:04
they broke all of their trades up into
00:13:06
smaller trades
00:13:07
and they aimed to trade just the right
00:13:09
amount that would erase the
00:13:11
inefficiencies that they found in
00:13:12
markets
00:13:13
but have no additional impact on the
00:13:16
markets
00:13:17
their competitors would analyze the data
00:13:19
and never even see the trades they had
00:13:21
done
00:13:22
the idea was that they could step in and
00:13:23
out of markets with such
00:13:25
care that a competitor would not see
00:13:27
that an
00:13:28
inefficiency had existed and been armed
00:13:31
away
00:13:32
does their competitors when doing
00:13:34
analysis wouldn't even know that an
00:13:36
opportunity had existed and thus they'd
00:13:38
never compete with them for phil's the
00:13:40
next time it came
00:13:41
up now simon's had seeded a guy named
00:13:44
robert frye
00:13:45
a former morgan stanley pairs trader and
00:13:48
had set him up with a fund called kepler
00:13:50
financial management
00:13:52
fry's approach was to deconstruct the
00:13:55
movement of stocks
00:13:57
identifying the factors responsible for
00:13:59
their moves it was
00:14:00
kind of a more sophisticated approach uh
00:14:03
to
00:14:03
traditional pairs trading which was done
00:14:05
back then at morgan stanley in fact i
00:14:07
think
00:14:08
the morgan stanley process uh what they
00:14:11
call process driven trading team
00:14:13
uh might have even invented the idea of
00:14:15
pairs trading
00:14:16
so fry's idea was we'll take an example
00:14:19
like exxon
00:14:20
he he would have worked out that exxon
00:14:22
was possibly driven by
00:14:24
oil prices by interest rates and by
00:14:27
growth in gdp
00:14:29
and then he would look and see if those
00:14:31
factors moved
00:14:32
but exxon hasn't moved as much as you'd
00:14:35
expect it to have
00:14:36
he would then be able to trade based
00:14:38
upon this
00:14:39
and his approach worked but it never
00:14:41
seemed to work on
00:14:42
size a medallion up until around 2000
00:14:46
was a futures trading fund
00:14:48
but it had just reached a point where
00:14:50
they could no longer bring in additional
00:14:52
capital they'd reached capacity
00:14:54
and simon's wanted to grow the business
00:14:56
and felt that he could put a lot of
00:14:58
capital to work in equities
00:15:00
but he just couldn't find a good trading
00:15:02
strategy he hired in the
00:15:04
brilliant team of robert mercer peter
00:15:06
brown and david maegerman
00:15:08
and they had been building innovative
00:15:10
voice recognition software at ibm
00:15:13
using probabilistic models in fact
00:15:15
models developed by
00:15:17
simon's old partner lenny baum
00:15:20
the bom welch algorithm i think i
00:15:22
mentioned it
00:15:23
earlier they also had been involved in
00:15:26
the
00:15:27
deep blue team at ibm these guys were
00:15:30
different to the existing team
00:15:32
at renaissance in that they knew how to
00:15:35
build this kind of
00:15:36
big business software system that would
00:15:38
be used at a firm like ibm
00:15:40
they were able to take all of fry's
00:15:42
individual studies and signals
00:15:45
and build them into a much more
00:15:46
efficient piece of software that could
00:15:49
trade really well the software was able
00:15:52
to find
00:15:52
signals take execution factors like
00:15:55
liquidity or shorting restrictions into
00:15:57
account
00:15:58
apply risk management and correctly size
00:16:01
the trades
00:16:02
and this was really a new error for
00:16:04
medallion this was when medallion took
00:16:07
a further leap forward by 2003
00:16:11
stock trading was responsible for
00:16:13
two-thirds of medallions profits
00:16:15
up from zero the year before the big
00:16:18
lesson that we can learn here
00:16:20
is we can see how willing they were to
00:16:22
change and the importance of teamwork
00:16:24
within the organization
00:16:26
we can equally see that simon's worked
00:16:29
away at getting this to work like
00:16:30
initially he couldn't get more equities
00:16:32
to work for years
00:16:34
and then finally he just worked away it
00:16:36
got the right people
00:16:37
in eventually he got it to work so
00:16:40
shortly after that renaissance began
00:16:43
trading international stocks too
00:16:45
and this allowed them to put more
00:16:47
capital to work but it also added
00:16:49
greater diversification to the portfolio
00:16:51
meaning it reduced the volatility
00:16:54
giving them a sharp ratio of six in 2003
00:16:57
which is just an amazing risk return
00:16:59
ratio
00:17:00
in particular when you consider that
00:17:02
they were managing 5 billion
00:17:04
at the time a big lesson that we can
00:17:06
take away from this
00:17:08
is just that adding international stocks
00:17:10
for diversification purposes
00:17:12
is a good idea for all investors now
00:17:15
renaissance had many other tricks up
00:17:17
their slaves too
00:17:18
in the early 2000s they negotiated with
00:17:21
barclays bank and with deutsche bank
00:17:24
to trade basket options instead of
00:17:26
actually buying and selling the stocks
00:17:28
needed in their portfolio what they did
00:17:30
was they bought basket options which are
00:17:33
options on a basket of stocks
00:17:35
from the banks that represented their
00:17:37
portfolio
00:17:39
the banks allowed them to constantly
00:17:41
change the constituents in the basket so
00:17:43
they were essentially able to
00:17:45
actively trade within the structure of
00:17:47
this options contract
00:17:49
now this had a number of interesting
00:17:51
effects for one it gave them more
00:17:53
leverage gave them way more leverage
00:17:55
than was available to them before
00:17:57
in fact i think up to 20 times leverage
00:17:59
on their stock portfolio at times now
00:18:02
i should note that they were not usually
00:18:04
that levered
00:18:05
in addition it pushed a lot of the
00:18:07
portfolio's risk to the banks
00:18:10
as the most that they could lose was the
00:18:12
premium that they had paid for these
00:18:13
basket options
00:18:15
most importantly though these options
00:18:18
had long enough expirations that they
00:18:20
converted the short-term capital gains
00:18:22
of all the trading
00:18:24
into long-term capital gains as the
00:18:26
options lasted longer than a year
00:18:28
i'm actually a little bit surprised that
00:18:30
the banks even agreed to do this deal
00:18:33
so shortly after losing a ton of money
00:18:35
with long-term capital management in
00:18:38
1998
00:18:39
but they did this structure added tax
00:18:42
efficiency saving them over 6.8 billion
00:18:44
dollars
00:18:45
in taxes now all investors obviously do
00:18:48
need to pay attention to the tax
00:18:50
efficiency of their investments
00:18:53
and you do need to look at the
00:18:54
investment returns after fees and after
00:18:57
taxes in order to compare them
00:18:59
the track record of medallion fund is
00:19:01
just phenomenal it's it's unbelievable
00:19:04
they started out with high fees
00:19:06
the fees only got higher over time after
00:19:08
2002
00:19:10
the fees were 5 and 44. now traditional
00:19:13
high hedge fund fees are 2 and 20. they
00:19:15
were 5
00:19:16
and 44. but the after fee returns were
00:19:20
still
00:19:20
industry beating the after fee returns
00:19:23
for medallion funds since it launched in
00:19:25
1988
00:19:27
were 39.1 per year
00:19:30
with only one down year and two years
00:19:32
where the returns were below
00:19:34
10 and one of those was the down year
00:19:37
there were many difficulties of course
00:19:39
along the way that the great returns
00:19:41
might appear to cover up
00:19:43
at one point trading strategies and code
00:19:45
were stolen by
00:19:46
ex-employees in addition there were
00:19:49
severe losses during the dot-com crash
00:19:51
and the quant quake of 2007.
00:19:54
renaissance was likely a very difficult
00:19:57
place to work with many disagreements
00:19:59
between staff members at the firm
00:20:01
the enduring lesson for me is that hard
00:20:04
work pays off
00:20:05
some of the most successful employees at
00:20:07
the firm had murphy beds in their
00:20:09
offices
00:20:10
where they could take naps while working
00:20:12
through the night on improving
00:20:14
strategies
00:20:15
and these people did that for year after
00:20:17
year you know it wasn't a short-term
00:20:19
thing one of the big differences it's
00:20:22
worth noting between
00:20:23
renaissance and an awful lot of other
00:20:25
quant firms and in fact
00:20:26
even between renaissance and a lot of
00:20:28
other multi-strategy firms
00:20:30
is that at renaissance all employees had
00:20:33
access to all
00:20:34
of the trading code while they were
00:20:36
secretive
00:20:37
to the outside world there were no
00:20:39
secret trading strategies internally
00:20:42
huge gains came from this collaborative
00:20:44
environment the way they were able to
00:20:46
work together and improve each other's
00:20:47
work
00:20:48
but it also meant that you couldn't have
00:20:50
a higher and fire culture like you
00:20:52
sometimes see at other firms
00:20:54
or you'd lose your competitive edge they
00:20:56
had to hire very carefully
00:20:59
and to keep their team happy so that
00:21:01
these guys wouldn't leave
00:21:03
if you want to learn more of this story
00:21:05
i've only really scratched the surface
00:21:06
of it in this video
00:21:08
you should read the man who solved the
00:21:10
market by
00:21:11
greg zuckerman in fact this is why i
00:21:14
can't work in advertising on
00:21:17
the just dust cover so you should read
00:21:19
the man who solved the market
00:21:21
uh by gregory zuckerman it's a really
00:21:24
excellent book
00:21:25
on this topic i think it's a bestseller
00:21:28
right now
00:21:29
um it's worth noting that you know it's
00:21:31
not all smooth sailing like it's not
00:21:33
just a hero story if a guy comes up with
00:21:35
a great idea and grows really rich
00:21:38
when you when you read the story you see
00:21:40
that simon's had
00:21:41
many difficulties in his life in fact
00:21:43
there's one um
00:21:45
there's one line in there where he says
00:21:48
to a friend after a personal tragedy
00:21:50
says
00:21:50
for my life is all either aces or deuces
00:21:53
you know things either go really well or
00:21:56
really horribly for me anyhow i strongly
00:21:59
recommend the book i'll put a link to it
00:22:02
in the video description
00:22:03
below don't forget to like and subscribe
00:22:06
and i'll see you guys again next week
00:22:10
bye
00:22:16
[Music]
00:22:18
you
