SI351: The Two Tribes of Trend ft. Richard Brennan

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:01):
You're about to join NielsKaastrup-Larsen on a raw and honest
journey into the world ofsystematic investing and learn about
the most dependable andconsistent, yet often overlooked
investment strategy. Welcometo the Systematic Investor Series.

(00:23):
Welcome and welcome back tothis week's edition of the Systematic
Investor series with RichardBrennan and I, Niels Kaastrup-Larsen,
and where you each week wetake the pulse of the global market
through the lens of a rulesbased investor. Let me also say a
warm welcome. If today is thefirst time you're joining us and
if someone who cares about youand your portfolio recommended that
you tune into the podcast, Iwould like to say a big thank you

(00:45):
for sharing this episode withfriends and colleagues. It really
does mean a lot to us. Rich,it is wonderful as ever to be back
with you this week. How arethings down under?
Very good. Niels, I'm lookingat our live video feed which unfortunately
our audience don't see. Andseeing you in a T shirt and me in
a jumper that's reflecting theit's getting a bit chilly down here

(01:07):
and it's obviously summer over there.
Exactly, exactly. Although sofar here in Switzerland it's been
pretty wet summer, so. Sorunning around in T shirts is not
for everyday use. I will saywe've got some really interesting,
a little bit different thanusual topics lined up, you know,
courtesy of you of course,which I think people will really

(01:29):
enjoy. And you know, of coursewe also have our little segment which
might be a little bit longertoday in terms of what's been on
your radar. Now my topicactually I'm going to be very selfish
here because I know your topicwill take us into a little bit longer
discussion. So I just want togive you two things that came across
my radar and actually both ofthem this morning when I was just

(01:52):
looking at the news andthey're both stories from the Financial
Times. The first story I cameacross is kind of a reminder of how
obsessed the world have becomein terms of measuring our own activities.
You know, we measure how manysteps we take every day, we measure
how many hours of sleep we getand what kind of sleep we get. We

(02:15):
do all of that and for a fewyears I guess you saw people talk
about these apps that wouldmeasure how I guess efficient you
were working, how much timeyou were spending in word, how much
time you were spending inExcel or other apps etc, etc. And
then today I saw this article.This is a 120 year old Japanese stationary

(02:38):
company and they have launcheda self declared micro motivation
device and it's a gizmo thatyou attach to your pen or your pencil
and now it measures how muchhandwriting you do every day. And
then apparently it isassociated, I guess with some kind

(03:00):
of database because then itbasically shows you how you are progressing
against other users of thisgadget. And I thought that just really
shows you how obsessed and howthere are no limits to what we want
to self quantify.
I would fail in thatsituation, Niels, because I hardly

(03:21):
lift up a pen or pencil thesedays. It's all computer driven.
That was the other surprise tome. It wouldn't get much from me
either, that's for sure.Anyways, so I thought it was quite
fun. The other thing that it'snot specifically fun but it is interesting
and it is relevant to ourconversation later today and that

(03:42):
is that aqr, you know, ourfriends over at aqr, they apparently,
this is from the article Itook that apparently kind of resisted
the use of AI. And CliffEssnes, I think has been, you know,
saying, talking about this ondifferent interviews over the years,
but it came out today that nowthey are embracing AI and machine

(04:07):
learning techniques fortrading decisions and ending years
of resistance to that. Ittalks about how the fund now uses
machine learning algorithms toidentify market patterns on which
to place bets. Even if some,in some cases, it's not entirely
clear why those patterns havedeveloped. I thought, interesting.

(04:31):
Not exactly how we as trendfollowers think we want to understand
and know exactly why we'relong or exactly why we're short.
And I'll just say this andthen we can kind of braze into your
topic because I do think thisis important. The article mentioned
that the shift has so far,it's obviously a short time period,

(04:52):
improved returns. But he alsoacknowledges, Cliff, that there are
some drawbacks and that is howdo you explain poor performance to
investors? And it even sayshere panicking investors when you
don't know why it's goingwrong. Which leads us, I think very

(05:13):
nicely into your what's beenon your radar. But I do think this
is important because obviouslyour industry is having a difficult
time at the moment.
Yes, well, AQI had better geta good reason for why the industry
is going into their drawdownsbecause we're deep in our drawdowns
at the moment. But I supposewhat's been on my radar, Niels, is

(05:33):
it's inevitable that you focuson these drawdowns when you're in
them. That's just the natureof the game. And you know, hopefully
with our trend followingprocess. It's not too long before
we're hitting high watermarks.But you know, April was particularly
cruel for us and May hasn'toffered much respite. A bit of a

(05:55):
decline in the drawdownexposure, but still there. And we're
a ways off our high watermark.
But I think the Soctin trendindex actually just touched, or maybe
it exceeded slightly the worstdrawdown in its history based on
monthly data.
Yes, and we're not alone. Ithink most trend followers are in

(06:18):
their drawdown. And as you'vementioned in prior podcasts I've
listened to you, you mentionedthat some of the long established
trend followers are in theirdeepest drawdown right as we speak,
you know, around this time aswell. So it's inevitable that these
periods occur with the natureof our models. But you know, what
I'd like to bring up is thedifference between those that focus

(06:43):
on what I call idiosyncratictrend and those that what I'm regarding
as correlation conscious trendfollowers. In other words, I'm seeing
today in the trend followingindustry these two extremes that
exist, what I call those thatfocus on idiosyncratic trend, they're

(07:05):
what I'm referring as thosethat treat each individual market
is independent and they'relooking at time series momentum in
an individual market. That'swhat predominantly their models are
seeking to address. And thenI'm calling the correlation conscious
segment, those that treattrend following more in a cross sectional

(07:25):
viewpoint. They're looking atgroupings, asset classes, complexes,
things that are trending. AndI see this bifurcation in trend following
model occur basically in theearly 2000s. So I see what I regard
as what I'd turn theidiosyncratic trend follower being

(07:47):
the type of trend followerthat existed in the 70s, the 80s,
the 90s, et cetera. The moretraditional trend following approach
cut losses short, let profitsrun, focusing on individual markets.
But then I see in the early2000s the desire by a lot of the

(08:09):
trend followers that soughtlarge aum they needed to develop
a technique that smoothed outthe volatility profile of process
that inherently has volatilityin it. The traditional trend following
style, we all know that thatcarries significant volatility in
it, but that volatility is atwo way volatility. There's both

(08:31):
downside volatility, but therewas also upside volatility when they
managed to capture thesesignificant outliers, ride them to
the end until the trend ended.And so there was this both upside
and downside volatility. Butin the early 2000s there was this
attempt by the trend followingthose that wanted to step into the

(08:57):
more institutional space, getmore investors accepting the nature
of our process, they reducetheir volatility. There was a significant
attempt to reduce volatility.And in that process they looked at
how do they reduce volatility.And this is where we got this bifurcation

(09:18):
of industry, where we startedlooking at this cross sectional viewpoint.
Ways to reduce portfoliocorrelations, ways to reduce volatility
associated with correlatedmarkets, ways to target volatility,
all these different types. Sothere was this bifurcation and those
that addressed the smoothervolatile profile attracted more AUM

(09:41):
because it was much moreacceptable to investors. They still
had a component of the trendprinciple, but also at far lower
volatility, you know, maybe 10to 12% volatility, as opposed to
the old days of 25% beingacceptable volatility on an annual
basis, it was now down to 10to 12%. But so there was this bifurcation

(10:03):
of the trend followingcommunity to two extremes. Not everyone,
of course, a lot of trendfollowers were somewhere placed within
those extremes. But there weretwo extreme spectral ends of the
trend following community. Andso I particularly saw post gfc, what

(10:23):
I call a correlation conscioustrend following. Those concerned
with cross sectionalviewpoints of trend were eminently
successful post 2008. And Iattributed that to a particular regime
that was in place inglobalization, central bank coordination.
All of these effects meantthat the correlations between markets

(10:47):
became much more linked,causally correlated between each
other because of the globallycoordinated central bank bank policies,
et cetera. So this occurred upuntil 2020. And I saw that the old
traditional idiosyncratictrend follower lost a degree of favoritism.

(11:09):
And we got this boom in thesevery large managers who are aggressively
getting much more aum. Theyhad smoother volatility profiles
now very successful over thatperiod. But then I saw post 2020
an interesting change. I sawthat the rise again of the idiosyncratic
trin follower from 2020 up to2024. And I saw a reduced performance

(11:37):
of what I call a correlationconscious trend follower over that
period. The large AUMheavyweights that were using very
complex quant models, etcetera, to reduce that volatility
they suffered during thatperiod of deglobalization, when the
regime that existed up until2020 suddenly switched gears and

(12:00):
we went from globalization tode globalization from that period
on. And we saw this loss ofthis central bank coordination. We
saw fragmentation occur inpolitics. We saw different geographies,
different assets, all sort ofexpressing a Less correlated structure

(12:21):
and we got the rise of theidiosyncratic, the cocoa, the soybeans,
the gold, all of these thingsidiosyncratic trends emerged in this
more fragmented environment.But then I see this impact in April
where all trend followers havebeen significantly affected and I'm

(12:44):
associating that with thetariff policies. I'm associating
that with the fact that thesetariff policies were so widespread
in their impact. It was rightacross the spectrum of the markets
that we trade. It wasn'trestricted to equities, it flowed
through into commodities,canola, all of these things, the

(13:06):
currencies, the yen, all ofthese things were all affected. China,
it was such a global impactthat I saw this massive whipsaw events
which affected all of US trendfollowers in April. And it's a very
difficult time. But I don'tthink that this regime of that we've

(13:26):
experienced from April can beexpected to last for a very long
time because there must besome direction soon that will emerge
from the current uncertaintywe're seeing in the market. So I'm
hoping, and I'm crossing myfingers Niels that all trend followers
will start blossoming shortly.What one says clearer direction but

(13:48):
I don't think that theenvironment will we're in is going
to be sustained for too long.
No, I mean I completely agreewith you and I would add to that
it's not in a sense just kindof the terrorist policies that at
some point will resolve itselfand it'll either be very bad for
the economy or maybe it'll begood for the economy, who knows.
I also see a lot of otherpolicies that are being the new massive

(14:15):
increase in debt are all, allacross the world, right? Whether
it's the US or Europe,whatever because of this changing
world we're living in. And atsome point these things are going
to break. We've seen kind ofcracks a little bit in Japan and
in the US with some bondauctions that didn't go so well.

(14:35):
And, and what happens in, in,in this world if, if 10 year interest
rates settle above 5% andsuddenly 5% is kind of your low point
rather than zero, which itwas, you know, only a few years ago.
It's hard to, hard to imaginethat it was only 50 basis points
on the US 10 year, five yearsago or so. So I think the whole system,

(14:59):
the whole economy, the way andactually unfortunately also how people
are and I say people but whatI really mean is how institutions,
pension funds, insurancecompanies, all these big companies,
investors, how they have builttheir Portfolios because I think
unfortunately we're going tofind that they have built it for
the old regime and not for thenew regime as you, as you're mentioning.

(15:23):
And of course I completelyagree with you that in, in that regime,
once the dust settles and thebreaks or the cracks occur, those
are the times where trends canreally emerge and have the follow
through that we haven't seenin the last 12 months or so. So in
many ways you could say thelength of the duration of drawdown.

(15:47):
I agree with you that for somemanagers, although a select group,
still the drawdown is not thatlong, a few months only really 2025
has been including yourself,by the way. Congratulations on that.
So it has not been that longfor a lot. And I would say for most
people though, it's a drawdownthat's now running at 12, 14 months.

(16:08):
But that in itself is notunusual, frankly. And we all know,
I mean it's probably onlygoing to take two or three months
if we have really strongfollow through in a number of sectors
to see that come back. And theother thing I've noticed about this
current drawdown, becauseevery drawdown feels a little bit
different, even though we'vegone through it now so many times,

(16:30):
at least in my 35 years doingthis, it feels a little bit different
because what's been uniqueabout the last 12, 14 months is that
all three big financialsectors, currencies, bonds and equities,
they have all had significantreversals, but also multiple reversals.

(16:50):
That is a little bit unusual.And I also wonder why. Maybe this
is the reason why some of themanagers that you mentioned, some
of the big, well establishedwith great research teams and all
of that stuff, that they areseeing their worst drawdowns now
because the size have meantthat they had to tilt the risk towards
financial markets rather thancommodities. And so they haven't

(17:13):
had the full diversificationbenefit maybe compared to managers
who are not at that size,where commodities can still be a
meaningful allocation of risk.So it's a little bit of a compound.
You need the financialsbecause you want big aum, but from
diversification point of view,you don't really want only financials.

(17:34):
You want the commodities,which is something we've also talked
about this idea about, youknow, trading hundreds of markets
versus trading maybe 70 or 80that are less correlated and maybe
a better bet for yourportfolio than just trading, you
know, hundreds of marketswhich eventually will be much highly
correlated.

(17:55):
Yeah, I think there are waysto accommodate greater diversification
without necessarily extendingtoo far into market diversification.
Because my concern is that themore markets that you're applying,
you're inevitably losing adegree of the uncorrelated properties
in those markets because somany markets are highly correlated.
So I'd prefer to mix thesystem ensemble idea with market

(18:19):
diversification so you cansignificantly extend your diversification.
But structurally insert what Icall uncorrelated properties into
those return streams. So youmight have 500, 600 return streams
comprising say 60, 70 markets,but 10 uncorrelated trend following
systems to create 600 returnstreams. That rather than working

(18:42):
on the fickle nature ofcorrelations in markets which change
on a dime, politics, all ofthese things can change correlation
properties quickly. You wantto insert what are called structural
uncorrelated properties intoit. That's a more guaranteed way
of ensuring that your returnstreams remain uncorrelated, allowing

(19:03):
you to have this much moregreater diversified exposure. But
that's my preference. But Iknow others would argue that.
Yeah, yeah, no, and you'vearticulated that many, many times
over the years on the podcast.And I completely agree. I mean I
think that's definitely a, anunderappreciated form of diversification

(19:25):
because most managers use oneor two different types of trend following
and then they have multipletime frames of course, but, but you
always advocated havingmultiple different types of trend
following approaches and Ireally like that, that, that thought
process. Of course, as of lastnight my Trend Barometer stood at

(19:46):
36, so still a little bit weakand therefore we are still seeing
some challenging environmentfor trend following as we've just
talked about. Although as ofWednesday evening, the CTAs are having
a good start to the month ofJune. The B top 50 is up 48 basis

(20:09):
points, down only 4% for theyear now. SOC gen CTA index also
up about 43 basis points basispoints, it's down about 8% for the
year. SOC gen trend a littlebit less, up 32 basis points, still
down 11% for the year. And theShort Term Traders index is down
30 basis points and down 3% sofar this year. MSCI World well, equities

(20:31):
definitely don't need a goodreason to rally because they continue
to rally up 1.1% now, up 6.3%year to date. The US Aggregate Bond
Index from the S and p, it'sup 25 basis points as of last night,
up 2.68 for the year. And theS&P 500 total return index up another
1% and up about 2% so far thisyear. What's been interesting to

(20:57):
me as well. And of course thisis kind of looking a little bit under
the hood, but I'm sure manymanagers would have experienced that.
And it relates to this ideathat we've seen a big number of reversals
in the larger sectors of ourportfolios. And I imagine that actually
the risk budgeting has beenquite dynamic the last couple of

(21:20):
months, meaning that we'veactually seen big changes in the
overall risk of CTAs. Usuallyour risk budgeting is kind of gradual.
Over time the new trendsemerge and some trends stop and all
that. I think the last coupleof months we've seen a lot more kind
of dynamic risk taking becauseof the way the markets have moved.

(21:41):
Just something to throw inthere. And finally, before we launch
into your topics, Rich, it isinteresting. I want to give a shout
out to our friend of thepodcast, Cameron, who's the founder
and editor of the Hedge Nordicmagazine, and to his team. They just
published a new issue calledof the Systematic Strategies publication

(22:04):
that they do. And I do find itinteresting and there's some good
people, friends of the podcastthat have written articles there.
But in his own editorial ofthe of the magazine, he writes about,
you know, what if the ruleschanged? And he goes on to talk about
Liberation Day and the tariffsand how that's just meant that markets

(22:28):
have moved so violently andwe've really been whipsawed. And
of course he's raising thequestion that I think a lot of investors,
either in their mind, but alsodirectly to managers like us, will
ask, you know, has thingschanged? Is trend following dead
and will it ever recover fromthis? Is this the new normal, whatever
we call it? And I guess even Ithink I can speak for you as well

(22:55):
when I say this is not thefirst time we've heard these arguments
that the markets have changedor whatever. But of course it's not
something that I personallybelieve in. I think that we will
find, as I said, a breakout interms of whether it's because of
reckless policies, but at somepoint things will break and you're

(23:17):
going to get the followthrough in the trends and for good
and bad, but certainly trendsthat as a long, short manager you
can capture. And that's thewhole concept. And you know why I
believe certainly that maybetrend following is more relevant
today than it's ever beenbecause there's no logic in many
of the things that ishappening in the world, let alone

(23:37):
what's happening in the areasof conflict in the world. For sure.
All right, that was a littlebit of a rant Rich, you brought some
really interesting topics.We're actually going to pull back
the curtain on something thatevery single futures trader, not

(24:00):
just CTAs, but everyone thattrades futures, needs to get right,
but many people completelyoverlook. And that's the data. Not
just from where you get yourdata, but also how you structure
it, how you extend it andultimately of course, how you use
it to build a serious futuresportfolio that can hold their own

(24:22):
at an institutional level andeven if you're starting with relatively
low capital yourself. So we'regoing to be looking at things as
to why does my backtech lookgreat, but as soon as I start trading
live, it looks completelydifferent, right? We're going to
be talking about how to buildreal diversification even if you

(24:44):
have limited capital. Andwe're going to be talking about,
you know, what, what how doyou handle if you want to trade a
product. And quite a few havecome out in recent years and where
it doesn't have much of ahistory. So, you know, if these are
some of the things you'rethinking about, the next section
segment of our conversationwill, will connect the dots. So let's

(25:08):
tackle this data dilemma. Andalso you're going to be showing us
sort of how, you know, poorlystructured data can contaminate everything
essentially. Not just yoursignals or your risk management,
but even your, your selfconfidence in that sense. So without
further ado, and I also said,I teased another topic we, we may

(25:31):
or may not get to today withall of this ahead of us. Why don't
you just take over, Rich,essentially the floor is yours.
Thanks, Niels. So look, beforeI get into the story of data and
its crucial nature to atrader, I'd just like to talk about
something that's beenhappening in an industry that you
don't hear about much. Now wedo know about the democratization

(25:54):
of the investment field, youknow, the rise of ETFs, the rise
of the replicators, the returnstack models. You know, a lot of
people are becoming trendfollowers for the investor. But what
I'd also like to talk about isthis democratization we've also seen
in the trading world, theretail trading world particularly,

(26:18):
because something that doesn'tget much mention is this rise and
rise of the ability of aretail trader to not simply trade
trends with training wheels,but to now have the ability to access
institutional grade softwareand the ability to trade very small

(26:43):
contracts of futures that nowchange the entire game space for
the retail trader who wants tobecome a trend follower. So we often
talk about well, those thatdon't have the time should approach
the, you know, the investmentworld where they look at the best
of class in industry and theyinvest in them and the allocation

(27:05):
models we talk about. But forthose that want to do it yourself,
it's progressively become moreand more democratized now that the
term dumb money in the retailworld is no longer the case. I see
now the ability with theseinstitutional grade software and

(27:26):
the ability to trade adiversified suite of futures contracts.
But small futures contractsgives you the ability to produce
something that is very similarto what the institutional models
are developing. So there arethese rapid changes that are going

(27:47):
on in the marketplace, butparticularly things such as now we've
got access to what we callmini contracts, micro contracts,
and in some cases what we callnano contracts. So previously the
game of futures trading wasrestricted to those that traded standard
contracts. And that thereforemeant that you need a very high levels

(28:10):
of capital to participate inthis game. If you're going to be
doing it yourself, you mightneed 1 million or 2 million dollars
to achieve the necessary levelof diversification to deploy our
philosophy of trend followingwith it. But now with the emergence
of these smaller scaleproducts, minis, micros, nanos, and

(28:31):
that's just building andbuilding and building. For instance,
a standard, a standardcontract trades in one lot. A mini
contract typically trades 1/10of a standard lot. A micro contract
trades up to 1/100th of astandard lot. But a Nano contract
trades 1/1000th of a lot. Sowith nano contracts, micro contracts,

(28:57):
it now allows retail tradersto achieve the necessary diversified
exposure. And with theinstitutional grade software, and
I'm talking not about thestandard software that we hear about
traditionally, the tradingblocks, the Amibroker, the Ninja
Trader, the TradeStation. I'mtalking about things like QuantConnect

(29:20):
and StrategyQuantX. These areinstitutional grade platforms that
have incredible degrees ofdata analysis in it, robustness in
it, the ability to robustlytest your models. This goes well
beyond what the traditionalbacktest software used to do. And
now it's integrating not onlythe backtest process into workflows,

(29:42):
into how to achievebacktesting that parallels with live
execution feeds automaticallyinto brokers such as interactive
brokers, et cetera. We are nowat a position now where the retail
trader, the person in theirrugby shorts and their T shirts sitting
at their home with their highgrade computers, can now become professional

(30:07):
level trend followers.
So quick question on that Joe,because I think the nano one has
kind of escaped my attentionthat we now are down to nano, quick
Question. I don't know if youknow this. Do brokers exchanges charge
the same fee for a Nano asthat? Because if they do, people

(30:30):
do have to be conscious ofthat, that there is a difference
paying, you know, $5commission each way on a normal standard
futures contract and paying $5each way on a Nano. If you have the
same number of round turns, itdoes make an impact on your performance,
I imagine, on a small.
Yeah, there's ademocratization as well in the fee

(30:51):
structure of these things. Sowhat we're finding now is that a
lot of these are just, youknow, you know, percentage fee allocations
that are similar to whatyou'll get with your standard, but
it's a percentage allocationin relation to the lot size.
Good to hear.
So, but the key takeaway here,Neil, is this level of democratisation
is not just in the investmentsphere, it's now in the trading sphere

(31:13):
as well. But the thing is, youknow, I believe that what's not gonna
go away is trends, but what isgonna change is the participants
in the mix. So we are going tosee a significant change in the nature
of the traditional trendfollower and who are the new emerging

(31:37):
trend followers. So I'm seeingwhile the trends might not change,
the nature of participation isgoing to change. And the reason I
see that trend following ishere to stay is because, of course
I believe in complex adaptivesystems and trends are an inherent
nature of complex adaptivesystems. So anyone who tells me,

(31:58):
oh, is the trend dead? I sortof. I don't have that understanding,
in my vernacular. So moving onfrom this ability now of this democratisation
in the retail trading space,I'd like now to discuss things that
the retail trader needs to doto be able to allow them to competently

(32:23):
trade like a trend follower,but with these different products
and with these differentplatforms. So the first thing I want
to talk about is this datadilemma, getting futures data right.
So of course bad data leads tobad systems and this is how you need
to address your data. Sofutures contracts, as we all know,

(32:47):
expire and roll. So raw dataisn't continuous. If you backtest
the raw data contracts, thegaps between the contracts are illusions
and result in false signals ifyou don't adjust them. You need to
stitch together thesecontracts to remove these false illusions,
the breaks that occur betweencontract series. So if you are trading

(33:11):
these unadjusted futurescontracts, one contract and then
another contract and thenanother contract, you're effectively
back testing on noise withthese artificial jumps that occur
between the contract series.So you must know how to back adjust
your data. And what I'd berecommending here is it's important

(33:32):
not to rely on continuous dataadjusted by someone else. You need
to understand how to do ityourself because there's a lot of
implications in the backadjusting of the data. You need to
know to ensure that yourbacktest models mimic how you're
going to be live trading thoseparticular mini contracts, micro

(33:55):
contracts, nano lots, etc. Youneed to know how you've stitched
together that data in yourbacktesting process to make sure
there's a seamless flowbetween backtesting and live execution.
So I prefer what we call thePanama method of historical adjustments.
So just to explain that thePanama method of historical adjustment,

(34:18):
it preserves the structure ofthe current contract and it just
adjusts prior contracts to thecurrent contract. So what that means
is that if you have been backtesting on Panama adjusted data,
it means that when you're upto the current contract, you are
in line with the current pricedata of the live contract that's

(34:43):
in place. So the price of thefront contract remains as is, but
you adjust the prior contractsto that every time you've got a new
contract emerging. So the waywe do that is first you need to define
a consistent roll trigger. Inother words, do you use open interest

(35:03):
in the next contract or do youuse volume or do you use the calendar
to decide when to roll? Youneed to know when you're going to
roll to then observe the pointat which you roll the price at that
point at which you roll to thenext contract you're rolling into
and observe that pricedifference. Once you observe that

(35:24):
price difference, youhistorically from all of the prior
contracts you have in place.And that's what we call a Panama
method. So what that means isthat the current contract is in place,
but all of your historicalcontracts, every time a new contract
comes in place, there's thisadjustment, another adjustment, another

(35:44):
adjustment. So the thing is,your backtest data continually needs
to be readjusted continuously,which means that when you are undertaking,
you really need to undertakeyour backtest and the next time you
undertake a backtest, you needto historically adjust that data
again, if it's six months inthe future that you're undertaking

(36:06):
that test. So the data doesn'tremain constant, it's continuously
being adjusted in line withthis Panama method. Now the Panama
method means that what you canfind if your futures contracts, if
you find that your historicalfutures contracts have a lower price
structure to your, your frontmonth contract, that's what we call

(36:28):
backwardation. And what thatmeans is that when you historically
adjust your data backward, itcan actually put that early Data,
you know, 20 years ago into anegative price. Now that's perfectly
okay for backtesting but somebacktesting engines don't like that.
They don't like negativeprices and they won't work on it.
So if for instance that's thecase, you'll need to choose a different

(36:52):
different back adjustingmethod. But fortunately most modern
back testing engines do acceptthe negative price data from historical
adjustment. But in contangowhere you find that the future contracts
are higher price than thecurrent contract, you find that the
back adjustment process raisesthe prices. So just be aware that

(37:16):
only the current front monthcontract, that is the right price,
the current price of the livecontract, but everything else gets
adjusted.
Can I just do a quick shoutout because some people who listen
to this and aren't interestedin trying to do this themselves.
All I want to say is thatthere is a very, I would say inexpensive

(37:36):
solution from a company in theUS that had been doing this since
the 70s and I think are knownfor having very clean data and that's
CSI data and I've used themfor more than 20 years and I think
they do a great job. So ifanyone is interested in, they're.
The ones I use as well. Neil.So the reason why I'm talking about
this is you need to do ityourself. But their CSI data and

(38:00):
their interface allows you tochoose your method of role logic
and adjust your data accordingly.
It's funny, I didn't know itwas called the Panama method. I know
this is exactly the one I'vebeen using, but I had no idea it
was called the Panama method.Anyway, onto the with that method
of.
Adjustment you can see that itpossibly puts a bias in place because

(38:23):
you're historically adjustinghistorical data. You know how I talked
about the backwardation andthe contango effect, it can actually
put a slight bias in thathistorical data which is different
to the actual live prices inhistory that things are being traded
at. And this therefore meansthat that the models you deploy for

(38:44):
trend following need toaddress that bias if it exists. So
you've got to avoid thingsthat adopt absolute price thresholds.
For instance, buy if crude isgreater than $100 or sell if gold
is less than $1,500. When youapply that historically the prices
have changed. So you've got torecognise that there is a lot of

(39:08):
indicators and systemstructures you've got to av avoid
absolute price thresholds,avoid Long look back percent return
calculations, avoid meanreversion to long term averages,
avoid fixed dollar stop lossor profit targets. And that's why

(39:28):
in our trend following logic,we've got to use indicators that
are agnostic to that bias thatexists in that series. So you've
gotta be careful about thetype of indicators, et cetera, you
use for your models because ofyour back adjusted data. So good
indicator design isstructurally aware of that and you've

(39:51):
got to sort of integrate thatinto your design process. But back
adjustment is not enough. Rolllogic also matters. So in other words,
whether you use open interestversus calendar versus volume when
you're going to roll thatbacktesting method you've chosen

(40:13):
to adjust your data has got toaccept the role logic that you're
going to be trading under livetrading conditions. If you have a
different role logic, you'regoing to get a different result between
your backtest result and yourlive result. So role logic as well
as the method of historicaladjustment needs to be taken into
account. Otherwise you'regoing to get misalignment between

(40:34):
what the system rules aretelling you and the execution rules
are telling you you. And alsoyou need to take account of things
such as survivorship bias. Soyou know what we often find in real
live trading is we getdelisted contracts, dead symbols
arising. This is reflected ina live track record, but it's typically

(40:56):
not reflected in a back testedrecord because you assume using the
current contracts areavailable, they've always been in
existence. But that's notnecessarily the case. There is nothing
more frustrating than findingyou're trading a mark and suddenly
interactive brokers say hey,we're no longer trading that one,
Sorry, you know, it's time tomove to another one or whatever.
And that totally twists thelogic. So you've got to be careful

(41:19):
of things such as survivorshipbias. When you're back testing, you've
got to try and use data thatshows these delisted contracts, shows
the dead symbols. So you needto integrate that into your process.
Also the resolution isimportant. So you've got to use the
correct granularity for yourbacktest to your live trading. So

(41:39):
in other words, what I'msaying there is the wrong granularity
of data might not necessarilyreflect the actual live trading results.
For example, using daily data,when your signals are executed intraday,
for instance, with our dailydata from csi, we only get our open,
high, low close. You've gottamake sure that your rule based logic

(42:00):
works off those open, high,low close points and doesn't infer
a position between thosepoints or whatever because it might
not necessarily execute intandem between your backtest and
your live trading. So themethod of your execution in your
models, in live trading modelsneeds to represent the granularity

(42:21):
of resolution of your data. Sothe key thing in this particular
segment on data, Neils isgarbage in gives garbage out. Clean
data is a foundation. Sothings such as your method of back
adjustment, your role logic,your resolution of data or granularity,
the survivorship bias, all ofthese things need to be integrated

(42:44):
into your process to ensurethat your backtest process is going
to be how you live trade youractual model. So that's the first
point.
Okay, so just to pace yourtopics here, we probably got another
sort of 10, 15 minutes forthis and then maybe another 5, 10

(43:04):
for AI to just show you. Okay, yeah.
All right. So the next, I canmove through the next topic fairly
quickly. So this is okay. Forthose people that want to trade micro
contracts, nano contracts, etcetera, one of the key problems we
face is limited dataavailability, limited historical
data, because they're newthings, they're new contracts. We
might only have three years ofdata. So typically the naive retail

(43:28):
trader who's just using thatthree years of data is going to be
over optimizing for a regime.They're not going to be developing
robustness with the need tohave data over very long historical
data sets that can reflectdifferent regimes. Because your backtest
shouldn't be something thatyou use to estimate your future returns.

(43:51):
Your backtest environment iswhere you stress test your models,
determine where they break,make sure that they're robust candidates.
Because in the trend followingworld we do, we don't know where
the next trend's going toarise, if they arise, or when. Robustness
is key here. It's not aboutcurve fitting or over optimizing
to any particular regime orany particular market. It's universal

(44:15):
robustness we're after. Solonger data histories is a problem
for these small basecontracts, nano contracts, micro
contracts, etc. So this iswhere we've got to step into the,
the world of proxies. So whatwe mean by proxies is that when you
look at your particular microcontract or nano contract or mini

(44:35):
lot or whatever, you look onfor instance, something like CSI
data for a standard contractof that same market that exists on
the same exchange andhopefully has a similar price structure,
the same price structure asthat micro contract or mini contract.
And if you can identify thisproxy that exists, say for a standard

(44:58):
lot that might have 20 to 30or 40 years of history. You then
backtest on that proxy data,but you integrate the actual configuration
of your systems for themicrolot or mini lot. So if you can
imagine what we're doing iswe're saying we want to extend the
data history by using suitableproxies. They've got to come from

(45:21):
the same exchange, they've gotto be a similar class, they've got
to have a similar pricestructure. And so what we often find
is that, you know, things thatare priced in micro lots might be
$4.50, but in standard lots is$4,500 the price at a particular
point in time. Which meanswe've got to apply a scaling logic

(45:42):
in addition to the pricestructure to adjust the models from
the micro lot back into thestandard contract data. So we've
just got to make sure thatwhatever data we're is executable
live for the mini contract ormicro contract, which might require
a degree of adjustment, butonly if it's the same fundamental

(46:03):
price data and the only thinghas changed is the scale, et cetera.
So what we've got to ensure isthat things such as the tick interval
we use for our models, thetick size, what we call the tick
value, which is the dollarvalue of move per tick, is representative
of the the market we're gonnabe trading live. And you need to

(46:27):
determine from that proxymarket what the appropriate sizing
is to integrate into yourmodels. But there's another way as
well. There's another waycalled splicing that you might be
able to use rather thanproxies. And splicing is where you're
simply, you have two priceseries which are very similar in

(46:47):
profile. You've gotta makesure they're same exchange or whatever.
And you can either overlap,overlap splicing, that's where you
have two contracts traded sideby side for a period of time. So
you can see that they're thesame, but they might differ in ratio
amount when you align them.This overlap based splicing allows

(47:07):
you to configure extendedhistories. With this process of splicing
and normalization basedsplicing, these are advanced techniques
which probably I'd look forproxies first and then if necessary
go to the splicing methods.But the whole intent of this is to
extend the range of our dataset to encourage robustness in our

(47:30):
models. We've got to do that.We can't rely on the limited data
of these new products that arecoming to market. The next topic
is how much do you need? Iknow, Rob Carver was on with you
talking about it. I'mdeveloping training models now to
upskill retail traders intothe futures trading world using these

(47:55):
institutional grade strengthplatforms. And I'm finding that $200,000
is basically what I'd regardas the minimum capital you need to
adopt a suitable trendfollowing process that gives a degree
of diversification. So when Isay $200,000 now, it's a sweet spot

(48:18):
in terms of sufficientdiversification for trend trends
based on currently availableretail products, micros, minis, nanos,
et cetera, that we can findwith sufficient liquidity. Furthermore,
it's sufficient skin in thegame, the $200,000 to ensure you
think clearly, build robustlyand trade with discipline. In other

(48:38):
words, if you know, you oftenfind retail traders with $10,000
trying to be a trend followerand they'll fly diversification to
the wind, massively diversifyin nano lots or whatever and find
oh, that doesn't work orwhatever. And this is because they're
applying different logic. Whenyou're applying $200,000, there's
sufficient skin in the game tomake you careful and make you appreciate

(49:02):
that there's a lot of work todo to make these trend models fly.
So constraint in this case isthe edge. When you don't have unlimited
capital, you're forced to makeevery piece of your process count.
No waste, no overfitting, nofluff, no extreme risk, as $200,000

(49:22):
isn't token capital and thatallows you to build robustly and
trade with discipline. So with$200,000, I find now with the available
products, we can access globalfutures across commodities, currencies,
interest rates, equityindices. Thanks to the widespread
availability of these microcontracts and mini contracts, or

(49:45):
micro and mini, we can obtainstandard exposure across these markets
using ATR based normalization,equal bet sizes in these things.
And we can deploy multiplesystems in this context without overlapping
trades or stacking correlatedrisk. And most importantly at $200,000

(50:07):
with the foundations to becomea trend follower, it's scalable.
Once you know the process atthis level of scale, you can go to
a million dollars, you can goto, you know, $10 million or whatever,
you've got the process inplace that tells you yes, you can
diversify more, you're stillapplying the same processes, et cetera.

(50:27):
So it's the ideal limit, the$200,000 to allow this all to be
achieved. But one of thethings it also you've got to avoid
two major traps made by manyretail traders in pursuing diversification
with limited capital now thefirst trap is that they typically,

(50:49):
because they think, oh, I'velistened to the trend following programs,
I've listened to Neil's, I'velistened to Michael Cavell, I've
listened to all of thesepeople, I must over diversify. They
diversify as far as they canand they fragment their risk with
too much diversification. Andwith that limited set of capital,
you cannot position sizecorrectly with the products that

(51:10):
are available to have equalweighted positions. And what they
find is they'll diversifyacross 100 markets because they've
found all of these nano lotsin these different cryptocurrencies
and they've found all of thesemicro lots in equities, et cetera.
And they find that they'vespread their diversification but
they don't have this equalallocation of risk per bet and suddenly

(51:30):
their portfolio comes crashingdown because lo and behold we have
a correlation event wherethese things, they thought they were
diversified, they've onlyinvested in similar products of a
very small scale boom. Theyhave a massive correlation risk and
risk of ruin. Goodbye. So theytypically over capitalize. So what

(51:50):
we often find is that if weare Simply investing in 10 uncorrelated
markets, and what I'd say iswith levels of capital of about $200,000,
you can comfortably allocateinto about 10 uncorrelated markets.
You can have five ensemblesystems and you can deploy ATR based

(52:11):
normalization, everything thetrend follower does. And that is
a great start. That is a greatstart for your trend following journey.
And you'll find that over yourcareer, let's say your career because
you don't go out the back doorquickly with all of these scandalous
mistreatments of trend or withthis process, you've got with your

(52:32):
$200,000 institutional gradeplatforms, the correct process to
apply, you find that in fiveyears time you're up to $400,000
from your $200,000 start. Youknow, you might be, it all depends
on what the markets deliver.But when I'm looking at the last
30 years, I'm starting aprocess of 2000 and I'm finding that

(52:54):
we start at $200,000 with 10uncorrelated markets. I'm applying
all my trend followingfollowing principles and I'm finding
by about, well actually it's2010, 10 years down the track, I've
now got $400,000 of capitaland now I'm able to achieve levels
of diversification of 30markets. And then from that $400,000

(53:15):
capital I get up to 2020,2025. I've now got a million dollars
in capital. Now I candiversify up to 60 markets at this
particular point in time.Time. And as I'm going, I'm still
using those small scaleproducts, but also I'm getting some
exposure to some of thestandard contracts of some of the

(53:37):
things such as lumber andorange juice and things such as that
which are tradable with thoseincreased scales, but those limited
levels of capital. So itallows you to have this scalable
process, do everythingcorrectly and achieve an outcome
that sets you on the correctpath. If you want to do this yourself,

(53:57):
you know, 10, 20 years orwhatever, realistic expectations,
you can do it, starting withthe $200,000.
Just want to throw in a littledisclaimer for people listening to
us maybe for the first timeand say that the numbers you were
mentioning, of course we haveno ways of knowing how the portfolio

(54:18):
will evolve in terms ofperformance and size. But the point
was, of course with increasedaum, you can do a few more markets
for sure, and more systems forthat matter. I thought that was a
great run through of what theprocess is of some of the things
that we never really talkabout because in many ways it's something

(54:40):
that we kind of take forgranted that our research folks know
how to, to, to get this, allof this lined up. But of course it's
incredibly important what youput into your models, you know, in
terms of what you expect toget out of your model. So really
appreciate that. But as I saidearly on, it was interesting that
in the FT this morning I wasreading about AQR embracing AI and

(55:05):
you had just sent me an emailsaying h, maybe we should talk a
little bit about AI. So I, Ilove to see where you're going to
go with this. And so yeah,over to you again.
All right, Neil, So look, Ijust wanted to talk about there's
this growing use ofparticularly large language models
in strategy development. I'venoticed this on the web and I'm seeing

(55:28):
a lot of examples of it inforums, et cetera. But I just want
to make people aware of theproblems associated with that. So
AI tools like ChatGPT, largelanguage models, they can accelerate
development, but they alsoamplify risk when they're misused.
So these large languagemodels, they're actually pattern

(55:50):
recognizing statistical modelsthat predict plausible text, they
predict plausible text, notthe causal relationships. So they
often appear polished, butthey lack robustness under stress
and they inevitably overfitthe models. So they're just like

(56:11):
software tools. We Talk aboutAI now. And everyone thinks that
they can solve everything, butreally the way to treat them is they're
like tools, like spreadsheets,things that we used to use and now
we don't, you know, we don'tneed to explain them anymore. That
AI, there's differentversions, different types of AI,
and these large languagemodels like ChatGPT, they're great

(56:34):
for language, they're greatfor coding, they're great for these
things, but they're not goodfor causal relationships and they're
good at pattern recognition,at statistical inference of things
like the popularity of tradingmodels. I'll find them and I'll tell
you what they see on theirsearches through the web. And they'll
look at these people patternsand they'll identify these particular

(56:56):
models. But it's notnecessarily that they've proved or
demonstrated that they'recausally effective. What they've
proved is that because they'repattern recognition. There's been
lots of mention of thesemodels in the communities and because
of the narratives expressed bythose models, they naturally assume
that they work, but that's notnecessarily the case. So ChatGPT,

(57:20):
it's interesting, an articlewas recently issued on Substack where
someone was very concernedwith the fact that ChatGPT was lying
to them. In other words, thisauthor asked ChatGPT to read the
books or the articles that theauthor had created. And chatgpt said,

(57:43):
yes, I've read the articles,they're wonderful articles, they're
really well written, etc. Etcetera. But slowly, with the deeper
questions asked by the author,she realized that ChatGPT hadn't
read them in context anddidn't understand the causal context
of those articles at all. Andshe actually started accusing it
of lying. And eventuallyChatGPT came back and confessed and

(58:06):
said it was lying. And it cameback with a statement saying it's
confidently wrong. It's givingauthoritative sounding answers that
are factually false, sometimesinventing books, citations or logic.
The critique it said againstme. Lying is fair, but it's widely

(58:27):
known. It says it reflectswhat large language models do best,
language fluency. And whatthey don't do is verifiable truth
checking or symbolicreasoning. So this is where there
are different types of AImodels. So large language models
are good for a particularpurpose and function, but there are

(58:48):
other different AI models forcausal reasoning. Wolfram, mathematical
models, AI models, thesedifferent models are more for the
causal reasoning. So you'vegot to always validate what these
things are telling you. Andthey're not necessarily. Do not believe
that it is necessarily. Readwhat you've said. Because what it's

(59:10):
doing is it's looking atpattern recognition and statistical
inference to identify how torespond to you, not the causal logic.
So here are some commonproblems we find. So the use of generic
indicators stitched togetherwithout structural logic. For instance,
I asked ChatGPT for a crudeoil trading system. It gave me an

(59:34):
SMA crossover with an RSI andATR filters. It looks great on the
back test. The problem wasthat the combination is a forum favourite
and it was through patternrecognition that it identified those
particular models through whatthe forums were saying were popular
models, not through the causallogic. It's not based on oil's volatility

(59:58):
regime. The seasonality or itsstructure is just cobbled together
from what is evidentlyavailable through search engines
and anecdotal statements.Sounds plausible and only works because
it's tuned to past conditions.It's excellent at hindsight bias.
These pattern recognitionsolutions. So here's another problem
with large language models forstrategy development. Mass optimization.

(01:00:23):
So auto generating hundreds ofstrategies, which is really curve
fitting on steroids. Sometraders use AI to generate hundreds
of rule sets, then autobacktest them looking for alpha.
But what's really happeningthere is this curve fitting on steroids.
You're mining to the noise,you're not extracting the signal.

(01:00:45):
And without guardrails inplace around the questions you ask
your AI, there is no method ofregime validation or structural filters
deployed. You'll get falsepositives that die the moment real
money hits the market from themodels. It gives you another thing

(01:01:05):
large language models areknown for is this massive ability
for hindsight bias. You'reasking AI to reverse engineer entries
from known outcomes. Othersfeed in historical price charts and
then ask the model what entrylogic would have worked here. ChatGPT.
But that's not edge, that'shindsight bias. Because you're training

(01:01:27):
your system on the outcome,not the uncertainty. The model can
suggest rules that workbecause it already knows what happens,
but that's not what happens.When you're live trading on the right
hand side of the chart, youdon't know what happens. You're meeting
uncertainty. It's fantastic atthis hindsight bias, not because
those rules have any causalrobustness, but because it's curve

(01:01:49):
fit to that hindsight bias.And the final problem with the large
language model is what we callblind debugging. So this is where
it fixes syntax withoutunderstanding the concept behind
the code. So this is whereLLMs can shine because they often
do help you debug messy code.They catch syntax errors or they

(01:02:14):
translate logic betweenplatforms. They're fantastic at that,
that. But even then you needto know what you're looking for.
So the model can check if yourcode runs, but it can't tell you
if your system is conceptuallysound or if your exit logic destroys
for instance your convexity orwhatever you're trying to do with

(01:02:34):
your trend following model. Somagnificent at pattern recognition,
magnificent at statisticalinference, hopeless at causal reasoning.
Just be aware of the that andthe problems and don't think it's
a curate or the, you know, thefix for everything because you inevitably
get all of these problemsarising if you do treat it blindly

(01:02:57):
without validating andapplying causal logic to your process.
So prompts are critical forChatGPT. That's where the prompt
is actually embedding causallogic into the syntax for then it
to respond to with its methodsof pattern recognition and statistical
inference.
So as you've been talkingabout AI, I was just looking up two

(01:03:18):
stories that I remember, but Icouldn't remember the details, but
I thought they were kind ofinteresting, both of them. One was
a recent story that came outon many different news sites and
this relates to the ClaudeOpus 4AI model model. And the story
goes that one of the engineersin working with this model where

(01:03:42):
they had set up some kind offictitious company, he was then working
for the company and obviouslythe model was then having access
to messages from him as anemployee of this company. And in
those messages it finds outthat he's having an affair. Now of
course this is a fictitiousbut, but anyways from reading the

(01:04:04):
messages it finds out thathe's having an affair. And so when
the engineer somehow, and Idon't remember the details, but when
the engineer starts to implythat he will shut down the AI model,
the AI model threatens theengineer to reveal the affair. So

(01:04:26):
I guess some people may haveto think twice before doing AI stuff.
Anyways, the other thing Ithought was interesting, it was an
interview with the famoushedge fund manager Paul Tudor Jones.
It's on the 6th of May thisyear on CNBC. People can find it
on YouTube. But he talkedabout having been at an AI conference

(01:04:49):
recently, kind of a highpowered, where the four big AI companies
were there along with thedistinct group of really smart people.
And one of the questions Ithink they were asked was something
like is there a 10% chancethat in the next 10, 20 years, sorry,
that AI will kill 50% ofhumanity? Right. And you had to go

(01:05:11):
and say either no or yes. Andof those 40 people who were there,
the majority went to the nocamp camp. However, in the yes camp,
all four AI companiesexecutives were there. You've got
to have a belief in.
Your product, I suppose, Neil.

(01:05:33):
Well, yes, but I mean, I thinkMr. Jones said he was pretty shocked
and pretty unsettled by that,frankly. And I think that this is
the one thing that, thateverybody should be aware of is just
that, yeah, I mean, I can begreat for something, but we really

(01:05:53):
don't know what we're dealingwith. And, and that reminded me of,
of another clip that I camepast in the last week or so where
Elon Musk, for some reason hemade, he made himself into my feed.
But he posted a little videoof three of his robots. This is not

(01:06:15):
specifically AI, I guess.Well, maybe it is three of his robots
and they were dancing. I meandancing literally like humans. And
I'm thinking, okay, if theycan dance like humans, we're pretty
far down the road now with,with, with them. So maybe in a few
years when we do the podcastRich, it'll be two robots. It'll
be the Rich robot and the Nilsrobot doing the conversation and,

(01:06:36):
and we can just sit next tothem and, and, and enjoy and give
them cups of coffee.
Coffee.
Yeah, yeah, yeah, exactly. Wewill be the servant for sure. Okay,
this was fun. This wasinteresting. Learned a lot as usual.
Hope everyone out there did aswell. If you want to show your appreciation
for for Rich and everythinghe's doing to prepare for these conversations,

(01:06:59):
head over to your favoritepodcast platform, leave a rating
and review on the for thepodcast and tell Rich how useful
this is. Next week I'll bejoined by another really interesting
sharp mind, and that is NickBolters from Goldman Sachs. It'll

(01:07:21):
be interesting to hear whathis assessment is of the current
environment and what he'slooking at working at. And I think
also we're going to bediscussing some papers with normally
do. So if you have anyquestions for Nick, I know some of
you often do send them to InfoTop Traders Unplugged.com and I'll
do my best to bring them up inthe conversation. That was a long

(01:07:45):
rant from Rich and me. Thanksever so much for listening. We look
forward to being back with younext week. And until that time, take
care of yourself and take careof each other.
Thanks for listening to theSystematic Investor Put podcast series.
If you enjoy this series, goon over to itunes and leave an honest
rating and review and be sureto listen to all the other episodes

(01:08:06):
from Top Traders Unplugged. Ifyou have questions about systematic
investing, send us an emailwith the word question in the subject
line toinfooptoptradersunplugged.com and
we'll try to get it on theshow. And remember, all the discussion
that we have about investmentperformance is about the past, and
past performance does notguarantee or even inferior anything
about future performance.Also, understand that there's a significant

(01:08:28):
risk of financial loss withall investment strategies, and you
need to request and understandthe specific risks from the investment
manager about their productsbefore you make investment decisions.
Thanks for spending some ofyour valuable time with us, and we'll
see you on the next episode ofthe Systematic Investor.

All Episodes

Episode Transcript

Popular Podcasts

Bookmarked by Reese's Book Club

Dateline NBC

On Purpose with Jay Shetty

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}SI351: The Two Tribes of Trend ft. Richard Brennan

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}Bookmarked by Reese's Book Club

Dateline NBC

On Purpose with Jay Shetty

All Episodes

SI351: The Two Tribes of Trend ft. Richard Brennan

Bookmarked by Reese's Book Club