SI361: The Four Faces of Trend Following 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-Lasen,
where each week we take thepulse of the global market through
the lens of a rules basedinvestor. And I also want to say
a warm welcome. If today isthe first time you're joining us,
and if someone who cares aboutyou and your portfolio recommended
that you tune into thepodcast, I would like to say a big

(00:45):
thank you for sharing thisepisode with your friends and or
colleagues. It really doesmean a lot to us. Rich, fantastic
to be with you again. How areyou doing down under?
Not bad, Niels. It's prettychilly down here, so I've got my
jumper on. I know it's prettywarm up there in the northern hemisphere,
but I think next week I'mgoing to be joining all of you up

(01:07):
there. So I'm going to be.I'll get my bathers out and my togs
and my sun umbrella and comeup and visit you.
Yes, I'm so much lookingforward to meeting you in person.
It's strange after all theseyears that I've never met you in
person, so I really lookforward to that. And. But you know
what? So actually it's veryinteresting in Europe just to stay

(01:30):
on that topic for a second. Sotoday I'm actually in Scandinavia
and it's very pleasant uphere. I mean, it's not very warm,
but it's warm enough. Right.Switzerland, very hot at the moment,
not to say Southern Europe,massive wildfires and, and 40 plus
degrees Celsius at the moment.So you know, but I think you're coming

(01:51):
to kind of the sweet spot whenyou come to Europe, so it'll be,
it'll be super nice. Now wegot a really wonderful outline thanks
to you today, which I'lldiscuss in a second. But as we always
do, I'd love to hear what'sbeen on your radar, even though I,
I have a feeling I know whatit is, so to speak. And you'll be

(02:14):
surprised when you hear mytopics. But what's been on your radar
recently?
Well, Nils, I've been playingaround with the latest ChatGPT version.
It's version 5, which has beenreleased over the last few weeks
and I've got to say, wow, thefuture's coming at us fast. So I've

(02:34):
found that ChatGPT is anoticeable leap from its earlier
versions. Faster responses,deeper reasoning, better memory,
and it has got a much strongergrip of nuance. And so, but this
is where it's gettinginteresting, and I asked it a few
questions. So if you look athow fast we've gone from ChatGPT3

(02:59):
to ChatGPT5, the curve isgetting far steeper, it's exponentially
rising. Each jump is biggerand the time between jumps is shorter.
So it got me thinking wherethis trend leads us. So at the moment,
AI still heavily relies on us.We feed it data, set the goals, and

(03:22):
pull the plug if necessary.But at some point, maybe five to
10 years on, it's entirelypossible. We'll see AI systems evolve
to sustain themselves withouthuman intervention. They'll be able
to source their own data,refine their own models, and decide
which problems to solve next.So I asked ChatGPT about this, and

(03:45):
this is what it said. I askedit to estimate or paint three scenarios.
An optimistic scenario forhumans, a middle scenario for humans,
and a pessimistic scenario forhumans. And it came back optimistic.
Human control scenario, 15 to20 years away. And this is provided

(04:05):
by, we have strong globalregulation, tight computer governance
and cultural norms, slowly orautonomous handover. The middle scenario
is between seven to 10 years.And this is with gradual erosion
of oversight as autonomousresearch agents become mainstream

(04:27):
and open models keepimproving. And this is its fast track
scenario, it told me three tofive years. And this is where a combination
of decentralized compute,permissive open source releases and
commercial competition removeshuman bottlenecks much faster than
safety frameworks mature. Sounder these scenarios, when humans

(04:51):
lose their control over AI,the selection pressures no longer
will come from humans anymore.They'll come from whatever the AI
itself values. Efficiency,self preservation, problem solving,
speed. And that's wheregovernance becomes tricky, because
the old off switch may notwork the same way. So the uncomfortable

(05:12):
truth is that in the fasttrack path, the inflection point
could arrive before mostgovernments have any effective monitoring
or control systems in place.And that's why some in the AI governance
space argue for treatingcompute allocation and model release
like export controltechnologies, the same way we treat,

(05:33):
say, nuclear material oradvanced biotech. So I'm not saying
to you, Niels, Skynet's aroundthe corner, but it does raise a question,
are we moving fast enough withalignment and safety, given how fast
the capability curve isrising? And I'll tell you this, Niels,
if ChatGPT 6 arrives asquickly as ChatGPT 5 over its predecessor.

(05:58):
4. This conversation we'rehaving now might feel very different
a year from now.
Well, it's so weird, a littlebit scary rich that you put down
AI as your sort of what's beenon your radar. It's exactly the same
for me actually. But before Iget to that, there was one other
story I thought was quitefunny. So over here, obviously footballers

(06:20):
in the us they would call itsoccer is a really big thing and
the club that won theChampions League, which is the biggest
tournament in, in Europe, the,the club PSG from Paris won it for
the first time. And then I sawon, on, on the Internet this week

(06:41):
that they are now. So PSG ishiring a quant to trade crypto and
you think, what the hell, it'sa football club. No, it's true. So
it says PSG is looking for arelatively junior trader. They're
expected to have a PhD and orquantitative masters in and expected
to have three to five years oftrading experience in crypto hedge

(07:02):
funds and prop trading firms.The trader will also assist PSG Labs
broader efforts to createagentic AI tools to be used across
the club. I thought, I mean, Ithought that was quite funny that,
that even a football club ismoving into not just AI, but kind
of crypto trading. Butanyways, back to the AI thing because

(07:23):
it also hit me really hardthis week through a couple of interviews
that I was listening to. Oneinterview was with this, the ex Google
head of AI think and I forgethis name right now, he's Egyptian
and he's written a book acouple of years ago and this is a

(07:46):
relatively recentconversation, only a couple of weeks
old and he paints the picture,but he actually thinks that, you
know, even three to five yearsis not the. Well, it's going to be
even sooner. And he actuallysees a world that most likely could
be some kind of dystopianworld where a lot of people will

(08:07):
be unemployed but they'll getuniversal basic income. And so in
a sense he's saying that itmight be really rough initially,
so to speak, because peoplewon't know what to do with their
time. And, and, and of course,un. Universal basic income doesn't
mean you're going to live wellin any shape or form. But at some

(08:29):
point he thinks, well, itcould actually turn out to utopia
where we don't really have todo a lot, but we still get paid and
we can actually do what wereally want instead. So anyways,
I can't. It's such a long,deep conversation. But all I will
say is that after listening tothat, I listened to another conversation

(08:50):
from someone who, and again, Ishould have written the name down,
but he's labeled as thegodfather of AI. So he's an, he's
a 77 year old English guy whohad been doing, working on these
models for 50 years or so, butway before it, it turned out to be
quote AI. And he was alsoquite worrisome, but actually one

(09:14):
of his students used to be, oris the guy that was mostly responsible
for ChatGPT2 and who leftOpenAI because he felt worried about
the safety, as you say, thatnot enough was being done to make
AI safe. So he's called, Ithink he's called Ilya or something
like that. Yeah. And he's nowworking on how to make AI safe. But

(09:36):
in any event, I think yourpoint about safety, I think the point
about us controlling this, ofcourse you can imagine worst case
scenarios where as, as yourightly say is the self evolving
AI where we have no control.And if these things are put into,
say if they're weaponized andsuddenly you may have armies of,

(09:57):
of AI soldiers that decide whoto point the gun at and not, I mean
this is not something that isunrealistic anymore. And the other
thing is of course that whenyou have a world that, where we really
don't know where it's heading.And I personally think now, I think
I'm convinced now that the,and I've mentioned this a few times

(10:19):
over the years when we talkedabout the markets, right. We talked
about, we had to imagine theunimaginable that some of these market
moves like Coco could happen.And, and, and, and this is exactly
why our conversation today isgoing to be about some of these things.
But, but I think we should nowexpand that, that we live in a world

(10:39):
where we have to imagine theunimaginable, that in three years,
maybe two years, maybe fiveyears, our world and unfortunately
our children's world and theirchildren will be so much different
than any of us can imagine. Ihave convinced myself now that that's
probably where we're headingand it makes me frankly a little

(11:00):
bit nervous. But from aninvestment point of view, all I would
say is that if you live in aworld that can change, change so
dramatically and nobody, evenif they claim they have a clue of
where that's going. By theway, what this guy was also saying,
the podcast host had hadconversations with people who knows

(11:20):
the top three CEOs of, ofthese, of the three biggest AI firms
and he could Disclose thenames. Of course, there's only three
people here we can think of.And what he was saying is that the
conversations they have inprivate and what they say publicly
about where AI is heading aretwo completely different things.
We're getting the sugar coatedversion of oh, it's not too bad,

(11:43):
it's going to be great. Yeah,it's going to improve the quality
of life and it's going to helpyou do things much better, blah,
blah, blah. But that's notthe, the dark version. And you know
what, when I was listening tothat episode, I was kind of thinking,
and this is pure speculation,right? I wonder whether that's exactly
why Elon Musk is so committedto move to go to Mars, that actually

(12:08):
he deep down believes that theworld is quote unquote F U C K E
D and you need to go somewhereelse because AI is going to ruin
our world. I mean, I know itsounds crazy, but maybe not that
crazy anymore. Anyways, that'spure speculation. But what I was

(12:30):
going to say is that in aworld like that, what better way
to invest your money insomething that does not have a clue
or tries to predict where themarkets are going? And I truly mean
that. I mean, I've beenoptimistic about trend following
in particular in the last yearor two because we could see that
the world was de globalizing,we could see that economies were

(12:54):
becoming more protectionist,we could see that central banks was
diverging in their policies.All of that we could see. Yet it
hasn't played out great yetbecause of other, because of the
noise from the US is at themoment overpowering and our systems
need to adapt a little bit tothe noise being maybe the signal.

(13:17):
Not that they will interpretit that way, but it's just how the
data will, will pan out. SoI'm still very excited about that.
Not necessarily because Ithink we're heading into a better
world, but it might be muchbetter in terms of these non predictive
strategies. So I know we'll,we'll talk about that. The final
point I want to make is therewas maybe at all a little bit of

(13:39):
an upside to AI because the FThad an article today or maybe yesterday
where it talked about how theart of persuasion that the research
now shows that the top AIchatbots can make people change their
political views after lessthan 10 minutes of conversation.
Right? So I'm thinking maybeall we need is each cta, installer,

(14:04):
chatbot. So when people callthem, they start these conversations
and within 10 minutes they'reconvinced that Trend following is
the right strategy for you'reonto something. Yes, I don't know,
maybe. So anyway, that was alittle bit of a rant but it is super
interesting so we'll see.Anyways, let's go back to something

(14:25):
we know a little bit moreabout. We have a little bit more
certainty about that's trendfollowing. We're just chatting before
we pressed record that itfeels a little bit more constructive
this month for for trendfollowers and there will definitely
be some people who've caughton to some some great trends and
and see strong performance butactually my trend barometer is a

(14:46):
little bit, it's not reallyconfirming that just now. It's actually
pretty weak. Of course this isjust 44 markets that I'm measuring
and so if you are on a one ortwo outlier markets that that I'm
not including in thatportfolio you could still have a
fantastic month. But overallthe industries is up a little bit
this month and yesterday whichis not including in the numbers I'm

(15:09):
going to mention yesterday wasa positive day I would say pretty
much all round as far as I cansee from the early numbers. So beat
up 50 up about 48 basis pointsas of Tuesday down three and a half
percent only now for the yearSoctin CTA up about half a percent
down 7% for the year. Trendindex up 1.4% for the month, down

(15:32):
8.77 for the year. The ShortTerm Traders Index is down a quarter
percent and down 5.33. Sothat's probably the laggard started
out great and and did well ordid better during the month of April
but it certainly lost some ofthat glory in the last few months.

(15:53):
What hasn't lost its glory isthe traditional markets. MSCI World
is up another 2 and a halfpercent this month, up 14% now which
is long this year. S P USAggregate Bond Index up about 1%
just shy of that up 4.69 forthe year and the S&P 500 total return
up 2.04% and up 10.81% as oflast night. So tell me a little bit

(16:17):
about how you see trendenvironment similar differently to
to to what I'm seeing and ifthere's anything, anything, any markets
that particular is standingout to you.
Well the way I've seen it, weof course had this very tough first
six months of the year and Ijust had this opinion and I see it

(16:41):
sort of with what happenedlast month. There was this emergence
of trends startingdirectionality started happening
last month and things weregoing great for us last month. We
were having a bump a monthcompared to prior months. And then
the last two days, of course,we had that massive copper reversal
and all of that went sideways.But then again, it's this month once

(17:05):
again, another powerful startfor this month. You know, I don't
want to, you know, I alwaysget a bit worried when I'm very bullish
about a month because, youknow, the time I say it, the very
next day or the followingweek, I get an absolute thumping.
But things are going well sofar this month. I just don't want
to see it like last month,where in the last few days we had

(17:29):
this huge reversal. But youknow, the trends, trends for us,
you know, of course theequities are booming for us, the
metals. So it's definitely,you know, all of the things we're
seeing, bitcoin, all of thesethings. It's a very difficult environment
at the moment. And the, what Irefer to as the stress assets tend

(17:51):
to be sort of doing quitewell. So I'm keeping my fingers crossed
that this holds for the monthand I'm hoping that the last few
months of this year bring usback to at least maybe a break, even
for the year. But it's been atough year this year, Neils.
Yeah, no, you know, I mean,anything can happen. It could still
be a pretty good year at theend. But what's interesting to me

(18:14):
about this month so far, atleast from my vantage point, and
that is that it's kind of a,kind of a barbell attribution in
terms of sectors. So you have,as you rightly mentioned, equities
continue to really take thelead, not so much for the year as
a whole, but certainly for themonth and maybe last month as well.
And then you also have themetals, as you point out, you have

(18:36):
the meats doing really well.
Oh, meats, yes.
Yeah, yeah. But then you have,at least to some extent, some of
the soy complex, which hasbeen doing well, really sort of taking
a little bit of a, a reversalthis month. And then you have all
the fixed income markets,which are somewhat NSofts, for that
matter, somewhat more trickyat the moment. So. So yes, very much

(19:01):
depending on. And of course, Idon't have any insights to, to the
crypto world. We don't tradethat on our side. But I understand,
of course, from looking at theprice that people must be doing well
in that space right now. Sothat is good to hear. Let's move
across to your topics. Let metry and set the stage a little bit,

(19:25):
because I think it's going tobe kind of a fascinating conversation
both for the traders,investors, meaning allocators, because
we're going to try and take adeep dive into some of the different
archetypes of trend following.We're going to deal with some of
the big debates that divideour industry and also why you feel

(19:45):
that diversification plays acompletely different role depending
on the type of trend followeryou are or you choose. As an investor,
of course I'm going to let yougo through the main topics, but before
I hand it over, let me providethe listeners with a brief overview
of what we're going to betalking about. So first we're going

(20:05):
to try and break down fourbroad camps of trend followers, from
replicators who hawk an indexto core diversifiers to crisis alpha
risk offset strategies, andfinally the outlier hunters who cast
possibly the widest net, so tospeak. And along the way we're going
to try and explore some of thekey philosophical debates like diversification

(20:29):
versus concentration. We havevolatility, targeting versus static
small bets, we have symmetryversus asymmetry in rules and the
speed at which you execute, ofcourse. And then we'll go deep into
maybe the second part, thesecond topic, diversification itself,

(20:49):
and specifically why you feelthat for outlier hunters at least,
maximum breadth isn't just anice to have. It's the edge. And
we're going to look at aportfolio test the real world distribution
of returns and how objectivesdictate design choices. So what I'm
going to do, the best I can isjump in and out where where I can

(21:12):
keep up. And maybe I have someof my own observations on that, but
it's really gonna be as usual,over to you and let's see where we
go on this.
Well, thanks very much Niels,and yes, you will definitely be joining
in this conversation becausewe often have debates on this podcast
together and I'm bringing themup again, but under a different light

(21:34):
to say that there is no rightanswer. But anyway, we'll get into
that. So the thing about trendfollowing is that from the outside
it looks like one unifiedphilosophy. People think this is
people not in the know thatit's all just ride trends and cut
losses short and let profitsrun. But once you step inside, you

(21:57):
quickly realise even thatfamous mantra of cutting losses short
and letting profits run isn'tuniversal amongst trend followers.
Some managers build rules thatwill cut losses short without ever
truly letting profits run inthe classic sense. Others optimise
for smoothness or Symmetrythat naturally perhaps the upside.

(22:17):
So while many of us sharecertain fundamentals and probably
the things we do share is onesystematic rules, non predictive
logic and participation insustained price moves. Our objectives
can be completely differentand it's those differences that shape
the systems we build. So ifyou spend any time under the hood,

(22:41):
which you and I do all thetime, you see a very different reality.
Behind the shared principleswe all agree with lies a spectrum
of philosophies, objectivesand trade offs that can make two
managers portfolios look likethey belong to entirely different
worlds. So one manager mightrun 25 highly liquid markets, adjusting

(23:06):
position sizes daily to keepvolatility constant. Another might
run 100 plus markets, nevertouching position sizes after entry,
absorbing the noise in pursuitof a few life changing outliers.
Both wear the trend followerlabel. Both might even post similar
long term returns. But theirdesign DNA and the experience they

(23:28):
deliver to investors could notbe more different. So this is where
the problem begins. Mostindustries conversations skip over
these differences where we'rethrown into the same league tables
and judged by the samescoreboard. Sharpe ratios, mar ratios,
drawdowns, year to datereturns. These numbers are useful,

(23:51):
but they homogenizeeverything, compressing radically
different objectives into asingle measure. Say for instance,
a high sharp might be gold fora manager whose mission is to smooth
a multi asset portfolio. Butfor an outlier hunter, that same
number might signal that thedesign is leaving money on the table.

(24:16):
So without understanding theobjective, you're not reading the
number correctly. So when thismisalignment happens, it costs everyone.
Allocators end up withstrategies that don't behave as expected
in stress events. Managersdrift away from their edge, quietly
optimising for metrics thatplease investors in the short term

(24:40):
but erode survivability overthe long term. Long term. So I've
seen managers that might startwith an outlier mindset and slowly
morph into volatilitytargeters simply because that's what
the scoreboard rewards. And bythe time they notice their design
is no longer fit for thepurpose they began with. So that's

(25:01):
why today I want to pull backthe curtain and talk about the decisions
that really matter. Thephilosophical divides shape portfolios,
diversification versusconcentration, volatility targeting
versus static small bets,symmetry versus asymmetry in rules,
and short term versus longterm horizon execution. Because there's

(25:24):
no universal right answer tothese debates. A replicator tracking
an index will land in oneplace. An outlier hunter chasing
fat tails will land inanother. And if you don't Know your
purpose. It's easy to borrowrules from a different camp that
doesn't actually serve you andcan quietly sabotage your long term

(25:47):
results. So we'll start bybreaking down what I believe are
four broad archetypes of trendfollowers and walk through these
debates one by one. And afterthat, we'll zoom in on one debate
that defines my particularapproach as an outlier hunter more
than any other. And that'sabout diversification and just how
costly it can be when you cutyour universe too narrow. But that

(26:09):
is from the philosophicalperspective of an outlier hunter.
We've got to remember that.So, and through it all, one theme
will stay front and center.Survivability. So no matter what
the trend following archetype,the ultimate proof of a design is
whether it can remain intactand effective for decades. Through

(26:31):
every type of market stress,it can survive.
This is a really importantpoint that you're bringing up because
I do think that it's true thatwe have been kind of put on the one
label. And I think also it'sfair to say that a lot of allocators

(26:54):
who want trend probably thinkthat one trend follow is enough because
when you look at thecorrelation, they often look very
similar. And so I'm reallyglad you brought this up and I'm
excited to go through thesethings and I'm excited to debate
some of them, I'm sure. Solet's see where we go.
Carry on. All right, so here'show I see it. Once you step inside

(27:15):
the tent of trend following,you quickly realize that trend follower
is a label that covers verydifferent species based on, for example,
listening to the many greatinterviews you've had on ttu, particularly
last year, your interviewslast year with Alan, with the systematic
managers. This is how I tendto group trend followers. I group

(27:36):
them into four broadarchetypes. So the first. Let's deal
with the first. The Andrew DeBeers of the world. The replicators.
Okay, their mission is simple.Track a benchmark like the SGCTA
index as closely as possible.They focus on load tracking, error,
operational efficiency anddelivering the return profile allocators

(27:58):
expect from that benchmark. Sothat usually means trading between
10 to 30 of the most liquidmarkets. I think Andrew trades 14
if I remember correctly. Theyrun daily volatility targeting. They
steer clear of anything thatmight create large deviations from
the index and many blendingcross sectional momentum and rebalance

(28:20):
frequently to stay tightlyaligned. That's the first category.
The second category I call thecore diversifiers. So trend following
here is not a Standaloneproduct. It's a satellite allocation
within a larger multi assetportfolio. And it's deliberately
built to lift SHARP or MAratios by adding a diversifying return

(28:44):
stream that zigs when the restof the portfolio zags. So they might
trade similar liquid marketsto the replicators. And they often
use cross sectional momentumover absolute moment. And they tend
to be even more focused onsmoothness and drawdown control because
their role is to complement,not dominate the broader portfolio.

(29:08):
So the third I call the crisisrisk offset. So these are the convex
hedges designed to shine inequity drawdown. So they're often
called long volume variants ortail risk protectors. So portfolios
here might include shorterterm trend systems or OPT and overlays

(29:30):
to be the first responderswhen stress hits. They often keep
exposure to assets thatdeliver convex payoffs in crisis,
such as bonds in risk offmarkets or certain commodities during
inflation spikes. So a realworld example would be a program
that lost money for threestraight years, but returned 80%

(29:51):
plus in 2008 while equitiescollapsed. That's the exact profile
allocators want from thisarchetype crisis payoff first, smoothness
second. And then there's thisfourth category which I include myself
in, the outlier hunters. Sothis is where I sit. Outlier hunters

(30:13):
cast the widest possible netabove 100 plus markets. Because we
know that in any given yearjust a few positions will deliver
the outsized gains that definethe long term equity curve. We're
happy to trade smallermarkets, things like cattle that
we've talked about, small,less liquid markets than the major

(30:35):
players. And we can affordmore operational complexity because
the cost of missing a fat tailis greater than the cost of running
a lean, tidy book like some ofthe other archetypes. So for us it's
pure absolute momentum. Nocross sectional overlays, no dilution.

(30:55):
So of course there are thehybrids and the niche players. You
know, things like macro trendblends, multi strat fusions, crypto
specialists, commodity onlyfunds factor integrated quants. But
most still trace back to oneof these four philosophical archetypes.
So here's the thing. If youdon't know which archetype you are,

(31:18):
you're essentially flyingblind. It's like designing a vehicle
without deciding whether it'smeant to be a sports car, a four
wheel drive, a minibus or along haul truck. All can get you
from A to B, but they excel indifferent environments. So a sports
car and a four wheel drive canboth do 100 kilometres an hour on
the highway, but take them offroad and one gets stranded. And the

(31:42):
same in trend. Following arule set that's perfect for one archetype
can quietly cripple another.That's why I believe the very first
step in designing a system isbeing brutally clear about which
camp you're in. So withoutthat clarity, you could end up debating
trade offs that don't evenapply to your mission because there's

(32:04):
no single right answer acrossall four. So before I move on, do
you want to comment here?
Well, I feel there's one more category.
Yep.
And that's one that oddlyenough, I would call pure trend.
And what I mean by that isthat, and I talk about, say for example,

(32:27):
someone like Don, what ourobjective is to, is to produce the
best long term compoundreturns. That's our objective. We're
not designing it to providecrisis alpha during certain periods.
We're not doing that. That'shappening inherently in terms of
building for long term bestcompound using only trend models.

(32:54):
We're not building it to be acore diversifier. Again, we're building
it to be the best trendfollowing strategy that we can do.
We're certainly not buildingit to be a replicator. Maybe you
would say, well Will, you're alittle bit of an outlier Honda. Yeah,
but then we have some of thequirks that you don't like as an
outlier Honda. So I kind offeel that there may be room for a

(33:15):
fifth category. But I do thinkyou have to would have to be kind
of strict about yourdefinition there as well and say,
even though people say ohyeah, we're pure trend, well then
you have to kind of be able toprove that transparently that you
are pure trend. But anyway,that would be my thought. But I do
like your framework of tryingto define the different types of

(33:37):
quote unquote. I don't know ifwe should say it's different kinds
of CTAs different kind oftrend followers. I'm not entirely
sure. So anyways, just felt,at least it's a start.
So I put these four together.Just thinking broadly from my perspective,
you're right, let's move on.And I think there probably are more
archetypes, as you say, but atleast it's setting the principle

(33:59):
that there is no correctstyle. It's gotta be objective driven.
Anyway, I'll move on. So aboutthese debates that exist and why
a lot of them are acrosspurposes. So one thing I've noticed
over the years is that most ofthe heated arguments and trend following
aren't actually about facts.They're about objectives. So two

(34:22):
managers can be looking atexactly the same rules, the same
data, even the same markets,and yet walk away with completely
different conclusions aboutwhat's best. Why? Because they're
optimising for differentoutcomes. So a replicator might want
the smoothest possible ride sothey can track their benchmark closely.

(34:44):
An outlier hunter like memight want the most open ended convict's
payoff from a rare runawaytrend, even if that means years of
bumpier returns in between.These aren't just preferences, they
are fundamentally differentdesign targets. And that's why so
many of these debates are atcross purposes. Someone says oh,

(35:06):
you have to volatility targetor you'll be too risky. Another says
if you volatility target,you'll cut your outliers short. Both
are right for their ownobjectives. Both are wrong if you
apply their advice to astrategy with the opposite purpose.
So it's also why I think ourindustry has a metrics problem. We've
ended up with a handful ofcommon measures, Sharpe ratio, MA

(35:30):
ratio, average drawdown thatflatten us into one size fits all
comparisons. They make it looklike we're all playing the same game
when we're not. The truth isalmost every metric is only meaningful
in the context of the purposeof the strategy. The one exception
is the only universal metricthat applies to all archetypes is

(35:55):
a long term validated realtrack record. Why? Because survivability
applies to all trend followingstyles. And this principle only can
be gleaned from a long termvalidated track record of survival
across numerous differentregimes. That's the one thing that
cuts across styles becauseit's a proof that a manager has survived

(36:17):
and delivered their particularobjectives over time in real market
conditions. Everything elsehowever, needs to be matched to purpose.
So if I think about it, let'sthink of replicators. Tracking error
versus SGCTA index might bemore relevant for instance to them
than the MA ratio. If I lookat core diversifiers, portfolio level

(36:42):
Sharpe and correlation to therest of the holdings might matter
most to them a crisis riskoffset archetypes, crisis period,
CAGR or convexity ratios couldtell the real story for them and
for us outlier hound hunters,payoff, asymmetry, skewness and contribution
from top trades will always bemore revealing than a point in time

(37:06):
Sharpe ratio. So withoutmaking these distinctions, we risk
running comparisons and makinginvestment decisions based on measures
that don't actually reflectthe actual mission of the strategy.
And that's why I think beforewe even get into specifics like Diversification,
volatility, targeting symmetryand rules. We have to accept that

(37:27):
there's no universal rightanswer. There's only right for your
archetype and objective. Sowhat do you think so far?
Well, I mean, I think you'reobviously bringing up some very important
points. A lot of these metricsof course will be weaponized in the
marketing slide deck. Ofcourse, whatever fits best to your

(37:48):
strategy is the one you'regoing to say is the most important.
However, what I feel here,that is also a challenge is that
yes, we can define differentkind of trend followers having different
characteristics, so to speak,but I also think that in, you know,

(38:08):
frankly, a lot of investorswant more than just one characteristic
from, from a manager. Ideally.This is, this is why initially in
our conversation today, I saidit is the wrong view to say that
trend follows all the same.And I just need one that's definitely.
And I think you show thatthrough this. And in a sense, if

(38:32):
an investor were to be fullysatisfied with saying, well, I actually
need some crisis alpha, but Ialso need something that really can
compound for me over time orwhatever it may be. Well, that's
exactly why they probablyshould come up with a frame for framework
along these lines and say,okay, let's identify all the manage
managers we're looking at inour peer group, it being one of these

(38:55):
four or five different typesof managers. Let me decide which
of these groups do I need inmy portfolio to fulfill my overall
objective of the allocation toCTAs. And then drill down on your
short list and pick the bestone or two in each of these categories.
So you may end up with, withfour or five, I don't know. But yeah,

(39:15):
but, but two, I think it's alittle bit unfair to ask us as managers
to be able to deliver all thethings you would want in one package.
Because I think as you and Iwill get to. That's not possible.
It's not possible. Exactly.And when you design for these different
archetypes, you must takedifferent roads, which means that

(39:37):
you cannot satisfy allinvestor requirements. So this is
exactly right. So you know, mypreferred approach would be firstly
to define these trendfollowers into their different buckets,
use different ratios that arevalid for their objectives to look
at their performance, and thendo your selection across multiple

(39:57):
trend followers to seek aparticular broader objective that
marries a range of differentflavors, if you like. So the first
thing I'm going to get. Let'sget into these debates. So let's
get into the first debate.Diversification versus concentration,
let's say you're a Replicator.20 to 30 highly liquid markets is

(40:20):
often enough for a replicator.That's all you need.
Or even 10, I think, by theway. That's right.
So that's all you need totrack the SGCTA index closely, keeping
operational, complex.
Let's define closely as beinga little bit loose here, because
I think that's the challenge. Right?
Yeah.
Even they might performing them.
He hasn't actually beenbenchmarking them, he's been outperforming

(40:42):
them.
Well, that's the thing. Right.So again, going back to replication,
Right. You could almost saythe same about replication, where
we define it as being someonewho can just hog the index. But can
they really hog the index?Probably not. Right. Ideally they
would, because then they'regiving the benchmark to investors.
But you know, sometimesthey're going to outperform, sometimes

(41:04):
they're going to underperform,and all of that is fine. But I think
you're right in saying maybethe best metric for them is really
tracking error.
Yeah.
Because that's what theyshould be, you know, held accountable
for if they, if they representthemselves as someone who can give
you the index return, ideallyplus or something, Right?

(41:26):
Yeah, exactly. So back to thisdiversification. So for them, 10,
20, 30 markets is sufficientfor them to try and minimize this
tracking error. But for anoutlier hunter like me, that's far
too narrow. For me, my wholeedge comes from giving myself as
many opportunities as possibleto catch the big asymmetric moves.

(41:48):
That means running the widestfeasible portfolio I can, including
markets that might be lessliquid, might cost a bit more in
slippage, or might be morequirky in behavior. But these are
objectives that the replicatordoesn't want, because this increases
their tracking error. Sohere's the trade off. If you run
fewer markets, you might betidier and more efficient operationally,

(42:10):
but you're also increasing therole of luck in your results. Miss
the few markets that deliverthe big wins in a given period and
you're stuck with mediocrity.If you run maximum breadth, like
Red Flag, the outlier hunter,you're reducing that luck factor.
You'll inevitably have moreunprofitable markets in the mix.

(42:31):
But the one or two that goparabolic hopefully will more than
compensate for it. So this iswhere philosophy comes in. For me,
diversification isn'toptional. It's not just a portfolio
construction choice. To me, itis my edge. Every additional market
is another lottery ticket inthe fat tail draw. If I only buy

(42:52):
a handful of tickets, my Oddsof hitting the jackpot collapse.
So of course this is not auniversal law, but it applies to
my outlier hunting. A corediversifier, for example, might not
want 70 markets, 100 markets.The extra breadth could dilute their
desired correlation profile. Acrisis risk offset manager might

(43:13):
be more selective, focusing onmarkets most likely to respond in
equity drawdowns. But for anoutlier hunter, cutting the universe
down is like telling afisherman they can only cast their
net in one small bay insteadof across the whole ocean. Sure,
it's easier to manage, but thechance of landing a giant catch drops
dramatically. So this is wherethe debates get muddied. When someone

(43:36):
says you don't need more than30 markets, they might be perfectly
right for their style. But ifI applied that same advice, it would
fundamentally compromise myability to achieve my objectives.
So before we get into the deepdive on diversification later, cause
I've actually got an examplefrom an outlier's perspective which
I think is going to be veryinteresting. It's worth remembering

(43:59):
this isn't just a matter ofpersonal taste. It's a structural
decision that ties directly toyour archetype and what you want
your strategy to achieve. Sobefore I move on into the second
debate, do you want to makeany comments here? Nils?
No, I mean, I think that thisis the, this is obviously one of
the debates that we've talkedabout a lot over the years. I think

(44:22):
again, what often happens isthat you get people from each camp
talking in absolute, sayingoh, this is better or and the other
ones are worse. Of course, thetruth is that's not what the data
shows. As you rightly said atoddly enough, to some extent at least
many of the people who've beenaround for a long time, if you look

(44:44):
at their returns, say over arolling 10 year period, they're not
vastly different overall, butthey'll be vastly different in terms
of when they occurred and soon and so forth. So, so I've obviously
been part of the debate sayingthat when people a few years ago
said, well, you need to tradethree or four or 500 markets, you

(45:07):
know, I never bought quiteinto that because I couldn't see
it in the performance data. Icouldn't see the evidence. You can
certainly agree thatperformance is different, but I'm
just very cautious whenever Ihear someone saying it's better and
being very absolute about. Andthis is also, I think, where some

(45:28):
of the friction frankly, interms of this debate about are you
a classic trend follower? Areyou not a classic trend follower?
I think it's unnecessary. AndI know sometimes it's being said
a little bit to be in gist andin being a little bit provocative
to get the debate going. Andthat's perfectly fine. But I think
it's important for people whomaybe not be as much into the details

(45:52):
as we are that we're openabout it, that it's to some extent
it's the preference from thedesign of our systems. But it doesn't
mean that we can objectivelysay, oh, if you look. Because as
you well know, Rich, some ofthe best performing strategies, even
in the outlier hunter camp, soto speak, has been Portfolios with
only 20, 30 markets, 40markets in recent years. So of course

(46:15):
there is no such thing as it'salways going to be this way or that
way. But you open a veryimportant debate about it in terms
of how you see the best way ofachieving your design goal. Meaning
how many markets do I feel Ishould be trading if this is what
I want to achieve? Becauseeven within, I guess your group of

(46:38):
category, as I said, you'llfind people who trade hundreds of
markets. I think you yourselftrade less than a hundred. But you
still get a lot ofdiversification and you're gonna
find people who trade evenfewer than you do and they feel they
get a lot of diversification.So it's a bit of, you know, there
is some individual taste as well.
But it's interesting, Nils,that if you, if you trade fewer markets,

(47:01):
you've got to, you dodifferent things with your strategies.
So for instance, with my say100 markets or just a. But less than
100 markets, I have verysimple strategies. But if I deployed
them with less markets, they'dbe far less functional. So these
decisions, they're not just,you know, a single objective decision.

(47:24):
They influence everything inyour strategy design. So what I think,
if you remember, I think itwas Harold de Beer suggested that
we really should be proud tocall ourselves different in a trend
following space. We shouldn'tbe trying to be all homogenous. And
I tend to agree with that.Maybe it's time that we started looking

(47:47):
seriously at definingourselves better into these different
archetypes. It might help usbetter rather than being classified
as trend followers. So, youknow, these debates are continually
going to surface unless theytake the deep dive we're doing here
today to understand this.They're always going to be debated
on the social media, etcetera, or you're doing the wrong

(48:08):
thing, he's doing the rightthing, all of this stuff, stuff.
But if they understand theBroad objectives, really, that all
goes away.
But anyway, look, yeah, evento a point, Rich, even if they just
understand the four or fivecategories you started out by laying
out, I think that in itselfwill be extremely useful. Very powerful
actually from an allocator'sperspective, because then they also

(48:31):
know what to one expect. Butalso they would understand better
how to combine these differentdesign designs, let's call it that,
in order to achieve, you know,the highest probability of getting
the outcome they want so theydon't get disappointed. That's the.
The whole point. Okay.

(48:53):
All right, so let's move on tothe second debate. Absolute momentum
versus cross sectionalmomentum. So this tends to split
the trend following world in.In debates, heated debates. So absolute
momentum measures each marketon its own terms. If a market is
trending, you take the trade.If it isn't, you don't. Positions

(49:14):
are allowed to run withoutbeing cut back just because something
else is trending more. So thiskeeps your winners intact for as
long as they want to go. Butcross sectional momentum, on the
other hand, ranks all yourmarkets against each other and tilts
capital towards the strongestones, trimming or dropping the weaker
ones. The approach smoothsreturns, controls risk, and keeps

(49:37):
the portfolio concentrated incurrent leaders. So for replicators
or core diversifiers, two ofthose archetypes, cross sectional
momentum can be ideal. Itkeeps correlations in check, avoids
dead weight. But for anoutlier hunter, it can be poison.
You may end up selling intostrength and cutting the very trades

(49:58):
would have delivered yourbiggest lifetime wins. So both camps
have good reasons for theirdifferent styles. The key is knowing
which one matches yourpurpose. Are you trying to smooth
the ride or are you preparedto hold the choppy road if it means
catching the rare monstertrend? So that's a second broad debate,
once again, tied toobjectives. No right and wrong. So

(50:23):
the third debate, this iswhere we talk about volatility targeting
versus static small bets. Sothis is the debate that gets people
fired up. We've had many ofthem Neils throughout the on the
series because it touches thecore of how you think about risk,
returns and philosophy as atrader. So on one side we have volatility

(50:43):
targeters. So this is the campwhere many replicators and core diversifiers
live. They dynamically adjustposition sizes as volatility changes,
aiming to keep portfoliovolatility constant. If a market's
volatility spikes, they cutthe position down. If volatility
drops, they increase exposure.Now I'm not talking about done here,

(51:04):
Niels. Cause that's adifferent form that's dynamic position
sizing, which classifiesitself differently to that. But this
is the broad generalvolatility target as I'm talking
about here. So from theirperspective, this makes perfect sense.
It smooths the equity curve,keeps risk metrics like annualized
standard deviation and check,and produces a more consistent ride
for investors. It also helpsalign a strategy with a specific

(51:28):
volatility budget, which isoften a mandate requirement. But
for an outlier hunter, thisapproach can be counterproductive,
even dangerous for us. So myobjective is to maximize a payoff
from rare extreme moves. Thosemoves often come with surging volatility
in the middle of a trend. Soif I start cutting back my position

(51:49):
size just because volatilityhas spiked, I'm potentially clipping
the wings of my biggestwinners. So if I think about it,
if I catch a major uptrend incrude oil and volatility doubles
halfway through, a volatilitytargeter will cut their position
by half at the very moment thetrend is accelerating from my perspective.
They've just reduced theirpotential payoff from my perspective,

(52:12):
because their metric told themto smooth the ride. I don't want
to smooth the ride. I want toride the wave in full. That's why
I run equal small bets, atr,ATR normalised at entry. And then
I leave them alone. I don'tsize up if volatility drops, I don't
size down if it rises. Eachtrade is a small fixed piece of the

(52:34):
portfolio, designed that nosingle loss can hurt me too badly.
But every winner can reach itsfull potential. And once again, that
goes into the diversification.One reason why I diversify so widely,
my bets stay so small for anyparticular adverse volatility move.
But so this is wherephilosophy splits. Are you optimising

(52:55):
for smoothness, consistencyand investor comfort? Or are you
optimising for convexity, thebiggest possible payoff from the
smallest possible risk on anysingle trade? Neither approach is
right in a universal sense.Volatility targeting works brilliantly
for strategies designed todeliver stable risk adjusted returns,
especially when investors havea low tolerance for drawdowns. But

(53:18):
static small bets workbrilliantly when your mission is
to catch the home runs andaccept that your equity curve will
have more noise and biggerswings. And here is where it loops
back to my earlier point. Ifyou judge both these styles on the
same performance metrics, youmight think one is better than the
other. But you're notcomparing like with like. One is
optimising for smoothness, theother for asymmetry. So I think this

(53:42):
is why debates on social mediaget so heated, people talk past each
other because they'reimplicitly defending the approach
that fits their objectives,not necessarily your objectives.
So before I move on to thefourth, anything you'd like to say
here.
That might be one or twothings, Rich. Well, no, again, it's

(54:06):
really about, for me, it'sabout the nuances. Right. Because
I pay attention to thelanguage you use and you describe
it really well. So. Great. Buthere are a couple of things. When
you talk about the static, youalways talk about small bits, right?
Well, actually, I think infairness, I think there are small
bets on both sides. I don'tthink that the volume targeting managers

(54:29):
are taking big bets either. Ithink they're taking small bets.
I think that's fine. And veryimportantly, the volatility targeting.
And of course, we know wherethat term kind of stems from. Right.
I don't know that there arethat many of them left because I
think that some of the bigfirms that we all know, European

(54:54):
based mainly, I think, yep,that's how they started. But I don't
think they do volume targetingtoday. I think they do the same as
Don, which is risk targeting.So we don't worry about whether the
volume will go from 20%rolling 12 months of all to 30%.
That's just the way it works.Right. But we may have a cap on the

(55:15):
overall value and risk we cantake on any given day and so on and
so forth. So there's thislittle hybrid in between, which is
very important because Iactually think predominantly that's
what people do today. Now.
Fair enough.
That being said, there is onething that I also think needs to
be mentioned and that is whenin your camp, the static position

(55:40):
size, even though it may besmall to begin with, we have seen
examples in the last few yearswhere that little small bet became
a monster in the portfolio anddrove daily volume drawdowns performance
to an extreme. So much so thatI've seen one fund go from an annualized
volume of around 35 to at somepoint have an annualized volume of

(56:04):
like 95. So this is whatworries me to some extent is that
it can be a little bitseductive when you say, oh, we just
take small bets. So I'm notworried about it. No, no, yeah, yeah.
But things can change. And so,but as long as people know that it's

(56:24):
not an issue, it's not aproblem. It just needs to be made
clear from up front that, thatthese are the differences. And investors,
as you say, they have madeone, they may have one Preference
for someone who can really,you know, knock the ball out of the
park because of one market ortwo markets moving, or someone where
they say, yeah, if it gets toocrazy, we're gonna, we're gonna slow

(56:47):
it down a little bit. But, andthis is the important part, which
rarely gets mentioned forpeople who do use volatility in the
position siding, whetherthey're risk managers or, or volume
managers, so to speak, therisk can also increase. It's not
always about lowering theposition size just because we have
a big trend and, and we'relimiting ourselves to have a great

(57:12):
performance from that trend.No, no, we could in fact be actually
lowering our position size ata time when the trend has risen,
but the volume increases, thenthe market corrects, then the market
goes quiet, then we increasethe position size. We're still in
the same position, it's justnow being increased again and off
goes the market. So there'sall these small nuances which of

(57:33):
course in social media willnever be mentioned. So we end up
being, you know, as if we aremassively disagreeing. I don't think
we are, because I think weunderstand what the real differences
are. And as you say, there'sno wrong or right, it's just a matter
of preference.
So internally withinTrendland, we do, we have these vigorous

(57:54):
discussions. They're not,they're not, you know, in social
media it might sort of turninto a bit of conflict here and there,
but certainly when we talkcivilly explain our position, I think
everyone in trend followingland understands where we're coming
from when we're talking aboutthis. So, yeah, I agree with you,
but let's get onto this nextdebate. So this is something where

(58:15):
I might be different to someother people in my space. This is
symmetry versus asymmetry andrules. So this one is about whether
you treat long and shorttrades exactly the same or whether
you design rules differentlyfor each side. So replicators tend
to keep symmetry. So if thelong entry is a breakout above the

(58:36):
100 day high, the short entrywill be a breakdown below the 100
day low. Same stop, sametrailing logic, et cetera, same risk
parameters. This keeps thestrategy clean and the benchmark
aligned that it ensures nosystematic bias towards one side.
But not all camps take thatapproach. So crisis risk offset strategies,
for example, may intentionallyfavour the long side in certain markets,

(59:00):
especially in government bondsor safe haven currencies. Because
their primary mission is todeliver convexity during equity drawdowns,
they might allow for slower,looser exits on those longs. But
running Tighter stops onshorts in risk assets. So outlier
hunters like me, sometimes I'ma bit different to other outlier

(59:20):
hunters. So I'll considerasymmetry if it improves tail capture.
So, for example, in commoditymarkets, the most explosive moves
are often on the long duringsupply shocks. So, like wheat in
2022, natural gas, 2021, inthose situations, I might allow more
room for longs to breathe thanfor shorts. So conversely, in equities,

(59:42):
the most violent moves tend tobe on the downside during crises,
so I might run wider trailingstops on shorts to fully participate
in those collapses. So this iswhere I might differ from some other
of my colleagues. So the pointis, asymmetry is not about prediction.
It's about structural reality.So different markets and different
directions have differenthistorical profiles for speed, magnitude,

(01:00:05):
and persistence. If yourmission is to maximize convexity
from those moves, it can makesense to reflect that in your rules.
So, of course, symmetry doeshave its strengths. Simplicity, elegance,
fewer moving parts to explainto investors. But if you're willing
to accept a bit morecomplexity, asymmetry can give you,
I believe, an extra edge inthose moments that matter most. So,

(01:00:28):
again, this comes back toobjectives. Are you designed for
elegance and operationalsimplicity, or are you designing
for opportunistic capture ofrare directional extremes? So that's
on symmetry.
Yeah. You know, this isactually a point that I think is
very rarely debated, and Idon't think we've talked a lot about

(01:00:49):
it, actually in the podcast.That being something of a design
choice, I think for me,asymmetry has a little bit of a taste
of optimization. Right. Let'sbe frank. That's kind of what we're
trying to do. I think for me,if you just use the same rules across
all markets, et cetera, Ithink you could argue that maybe
it's a more robust approach tothat. But again, why shouldn't you

(01:01:15):
put your own taste into yourdesign? It's your system. So, of
course. And I think it's areally important point. I don't know
that many allocators ask usthat question, frankly, which they
should. Some do. Some do. ButI think it's a super, super important
one. And, yeah, I mean, it'sworked really well for you, so why
not?
Well, it has. And, you know,there might be instances as well

(01:01:37):
where we're at a historic lowon a particular commodity, et cetera.
We know that it hasn't gotthat far to go before zero. And you
know, that therefore constrainis how much of an outlier we can
get from where it is now towhere it is if it gets to those levels.
So that's why I do need totake that into account when I'm designing

(01:02:00):
strategies. But let's get ontothis next debate. Speed of execution.
So how fast you want yoursystems to react. So short term trend
followers and traders runninghigher frequency systems, they want
faster turnover, tighterstock, quicker responses to price
reversals. Their argument isthat the earlier you cut a losing

(01:02:20):
trade, the smaller the damage.And the earlier you enter a new move,
the more of it you capture.This style often appeals to replicators
who need tighter tracking ofbenchmarks, and to some core diversifiers
who value keeping portfoliorisk tightly contained. So on the
other side, however, the longhorizon trend followers want to breathe

(01:02:40):
through the noise. They'reprepared to sit through more volatility
in order to ride the multimonth, sometimes multi year moves
that deliver the real payoff.This is particularly true, for example,
for us outlier hunters, wherethe mission is not to catch every
move, but to stay on the bigones for as long as possible. So
the danger with too much speedis that you can get chopped out of
a long term trend multipletimes, missing the bulk of the payoff

(01:03:04):
because you couldn't absorbthe interim volatility. So speed
is not just a technicalparameter, it's a statement of philosophy.
It says something about yourtolerance for drawdowns, your patience
for building positions, andyour willingness to endure short
term pain in pursuit of longterm gain. Ultimately, the right
speed is determined by yourtrue north, which of the four archetypes

(01:03:27):
you belong to, and what yourportfolio is designed to achieve.
That's why for me, thisdebate, like all the others, circles
back to clarity of purpose.You can't choose the right reaction
speed unless you're crystalclear on what your strategy is meant
to deliver and over what horizon.
Well, I mean, so speed is veryinteresting because it's something

(01:03:49):
we often debate and it'ssomething people will use as a classifier
of what kind of managerthey're looking for. Are you short
term, medium term, long term?Now? Of course, I think the best
answer is that you probablyshould be designing your system to
be a little bit of everything,but not necessarily in a static way,
which we did in the old days.I think it was natural in the old

(01:04:10):
days that a lot of managerswould sit once a year and have a
committee saying 25% of ourmodels should be short term, 50%
medium term, and maybe 25%long term. That's kind of how it
was done today. You can do itin a much more scientific way you
can do dynamic optimizationand recalibration of your parameters
and so on and so forth. Thething that makes a lot of sense now,

(01:04:34):
when I look objectively at atrend model and I just simply apply
different time frames, thereis no doubt that long term parameters
work best. There's just no doubt.
Yeah. For your objectives. Butfor instance in the crisis offset
camp or in those differentobjectives, if their objectives are

(01:04:55):
different, it might not be formaximum cagr, it might be to provide
downside protection and hencethe short term trend followers. I
can see it.
Yes. And that's the narrativethat was sold to people a few years
ago. A few years ago, Iremember that, that the word risk
mitigation became a thing inthe, in the narrative. Right. And,
and people were saying wellmarkets are moving so quickly so

(01:05:17):
you know, you should, youshould go with short term managers.
And some of the managers grewlike massively multi billion dollar
short term managers. I waskind of thinking frankly that's never
going to work because you'regoing to have slippage and you're
going to all these things now.So what's happened in reality, this
is a good time to look back onit. Well, it's kind of the same thing

(01:05:39):
as when people said well I'mjust going to buy the VIX because
when equity markets go down,I'm going to make money. They're
going to go, the VIX is goingto go up. Well, surprise, surprise.
The VIX doesn't always go upwhen markets go down and sometimes
the VIX goes up when marketsgo up because it's much more nuanced
today when you decomposewhat's causing say the VIX to move

(01:06:03):
or not? We just did an episodethat came out a couple of days ago
on the, on the whole VIXdecompos composition. I think it's
actually quite interesting to,to learn from however, back to our
little sandbox. So what I'veseen at least or noticed is that
some of these short termmanagers. Let's take the, the April
liberation, right? Yeah. Theymay actually produce a decent better

(01:06:26):
return for two days in the,you know, when, when, when markets
were tanking for two days in arow. But they lose it all the next
three days or the next fivedays. So even within that space I
think they provide less of aportfolio benefit today than they
used to do. I think they usedto, the moves in the markets we used

(01:06:51):
to be a little bit longer sothat they could actually capture
the P and L from Say a volumebreakout, capture that for three
to five days and actually addbenefit to the portfolio without
detracting the next week. Forexample. I see that it's a little
bit more challenged today indoing so. And as we talked about

(01:07:11):
in the beginning, if you lookat the Soctian Short Term Traders
Index, it did really well,relatively speaking, the beginning
of the year. Today, on avolume adjusted basis, it's doing
worse than the trendfollowing. And you wouldn't say that
the last six months has been atrend following, a great environment.
It should, if anything be ashort term traders environment, but

(01:07:32):
it's not. So. So I don't know,it must be something to do. Maybe
there were the marketstructure or something like that.
But I think people have to bereally careful in terms of saying,
oh, we're short term, we'll,we'll definitely give you crisis
alpha question mark today.
It could, it could validate first.
Yes, true. I mean, I mean thenumbers is a good place to start,

(01:07:56):
right?
Yeah, exactly.
Okay, where are we going now?
So where are we going to gonow? So now that we're living this
happy, friendly place betweentrend followers, everyone knows what
we're doing, we understand thearchetypes. Now I'm going to start
looking at diversificationbecause this to me defines my outlier

(01:08:22):
edge. And I just want toexplain it to you. So if you're in
the replicator camp, you canlive with 20 to 30 very liquid markets,
still meet your objectives.But if you're a core diversifier,
you might see trend as asatellite allocation and keep it
even tighter. But if yourobjective is to hunt these outliers,

(01:08:43):
maximum breadth is notoptional. It's the oxygen our process
breathes. With the fewermarkets you trade, the higher the
odds you'll miss the nextwheat in 2022 or the JPY 2008 moment.
That's where I want to godeeper next. Not just why diversification
matters, but how the wrongapproach to it quickly erodes the

(01:09:05):
very edge you're trying tobuild. So I want to start with a
real test I ran because itsays more than any theoretical debate
could. So I built a 68 marketportfolio designed for an outlier
hunter. The test period wasJanuary 1, 2020 through to July 31,

(01:09:27):
2025. In that time, thestrategy generated 3,264 trades.
So here's the first surprisingthing. Of those 68 markets, 27 were
unprofitable over the entiretest period. And yet when we ran

(01:09:47):
the portfolio equal weightedacross all 68 with no hindsight application,
it still delivered a mar ratioof 0.8. So why? Because the big winners
swamped the losers. The tailevents, when they happened, more
than paid for the markets thatwent nowhere or bled. So let's flip

(01:10:11):
this thought experiment. Whatif you could trade only 30 markets
from this 68 market universe?No hindsight, no peeking at the winners.
You just have to pick your setof 30 within the 68 and live with
it. So how many Neil's 30market portfolios can you make from

(01:10:32):
a universe of 68? And theanswer is staggering. I don't expect
you to get it. It's four.
I was just going to say. Ihope it's not a question.
No, it's 4.8 by 10 to the 38thdifferent combinations. It's almost
an incomprehensible number.And to test what that means in practice,
I randomly generated 300different 30 market portfolios from

(01:10:58):
this 68 market universe. Eachwas built blind by me. No foreknowledge
of which markets would performthe best. And I'll tell you what,
the results I got out of those300 samples, only 35 of them achieved
a MA greater than 0.8, whichis just 12% of that sample. Which

(01:11:19):
means there's an 88% chancefrom that sample you'd underperform
the full 68 mark benchmarksimply because of market selection.
Not because your system wasbad, but because of. Not because
of execution errors, butbecause of the luck, or lack of it
in what you happen to includein your universe. And here's the

(01:11:42):
thing. If you accidentallyoverweight those unprofitable markets
in your smaller universe, youcan spend years underwater. This
is why for an outlier hunterlike me, maximum diversification
is non negotiable. And it'snot about operational neatness. It's
not about a clean marketingstory. It's about reducing the role
of luck in catching the nextfat tail. So if I think about it,

(01:12:06):
small universes magnify luck,Large universes dilute luck. A great
outlier in a market you don'ttrade is a missed opportunity. Pure
luck if it's in your book.Pure bad luck if it's not. Managing
that trade off is part ofdefining your archetype. So what

(01:12:28):
the portfolio test reallyexposes is the shape of the return
distribution for outlierhunting strategies. And it's not
the tidy symmetrical bellcurve that many investors imagine.
Instead, it's lopsided, highlyskewed, and brutally unforgiving
to small sample sizes. So ifwe look at those 300 random 30 market

(01:12:50):
portfolios I tested, themedian MA ratio was nowhere near
the 0.8 we got from the 468markets. In fact, it was well under
0.5. That's the median,meaning half the portfolios did worse
than that. And if I looked atthe 25th percentile, they were the

(01:13:10):
unlucky quarter of portfoliosthat ended up in the bottom range.
And many of them had MARratios that would be survival threatening.
If you're running realinvestor capital as an outlier hunter,
these are the managers who'dbe showing three years of red ink
not because their process wasbroken, but because they simply didn't
have enough breadth to letprobability do its work. And here's

(01:13:34):
the kicker. At the other endof the distribution, a small handful
of portfolios deliveredabsolutely extraordinary results.
But you had to be lucky enoughto land on them when choosing your
30 markets. That's the trap.So when you're hunting outliers,
most people portfolio outcomeswill underwhelm, a few will be exceptional.

(01:13:54):
And if you reduce youruniverse, you dramatically increase
the odds of drifting towardsthe noisy middle of the distribution
where the edge erodes and yourperformance blends in with everyone
else's. So this is why I saythe real enemy of the outlier is
not volatility or drawdowns oreven bad trades. It's the quiet invisible

(01:14:15):
erosion of the edge throughunder diversification for the outlier
hunter. So if your process isdesigned to capture rare events,
you need enough hooks in thewater to make sure you're there when
they happen. And that'ssomething the standard industry performance
metrics, they don't reveal. MAsharp cagr, they hide the fact the

(01:14:37):
portfolio may be sitting on aknife edge of survivability risk
simply because of the smallnumber of markets being traded. So
this is where the conversationcomes full circle. Nils, because
diversification isn't a virtuein itself, it's a design choice that
only makes sense when alignedwith your objective. So if you're

(01:14:58):
a replicator, 20 to 30 marketsis often enough. Your job is to mirror
the SGCTA index or a similarbenchmark, keep costs tight, deliver
a familiar performanceprofile. Breadth beyond that isn't
necessary. In fact, it mightjust add operational complexity without
adding much to your trackingaccuracy. If you're a core diversifier,

(01:15:18):
trend following is probablyjust one sleeve of a larger multi
asset portfolio. In thatcontext, you're optimising for incremental
Sharp or MAR improvement atthe whole portfolio level, not for
maximum standalone performancefrom trend. Again, a smaller Highly
liquid market will often serveyou just fine in that context. If

(01:15:41):
you're a crisis risk offsetmanager, the third archetype, you
might have a longer term, moreconvex profile. And your bucket list
could be biased towards assetsthat historically provide equity
drawdown protection. That'syour North Star. And diversification
choices will reflect that,even if it means trading fewer markets.

(01:16:02):
But if you're an outlierhunter like me, the game changes
here. Breadth is not anaccessory, it's a core operating
principle. My edge comes frommaximising the number of independent
opportunities to catchsomething truly explosive. And without
a wide enough market set, thelaw of small numbers works against
you. The cost of missing thebig next move is far greater than

(01:16:23):
the benefit of running astreamlined operation for me. So
that's why in my own process,I'm prepared to trade some markets
that aren't perfectlyefficient, some that carry extra
operational friction, livecattle, all of these small markets,
rubber, all of these smallmarkets milk. Some that might spend
long periods contributingnothing like Coco did for how many

(01:16:46):
years? Because the payoff fromjust one of them hitting a fat tail
event can outweigh years ofmediocrity. So diversification, the
approach you take, should bedictated by your purpose. And that's
why I think some of theindustry debates get lost. They argue
about the right number ofmarkets or the right market selection
criteria without first askingthe most important question, right

(01:17:08):
for what objective? So I'mgoing to close off now. I suppose
we're getting to the end. Butlook, if there's one theme running
through everything we'vetalked about today, it's that your
objective dictates yourdesign. Whether you're aiming for
smoothness index, replication,crisis convexity or outlier capture.
Each of these objectivesdemands different answers to the

(01:17:30):
debates we've covered today.Volatility, target targeting versus
static, small bets, breadthversus concentration, symmetry versus
asymmetry in rules, short termturnover versus long term patience.
None of these debates have asingle right answer. They have a
right for your purpose answer.And the trouble is, the industry

(01:17:50):
often flattens all thesestrategies into one homogenized peer
group and then ranks them onthe same metrics. It's like lining
up a bus, a sports car, a fourwheel drive and a motorbike, then
scoring them on their lap timearound a racetrack. It's a misleading
comparison and it hides thestrengths of each design. So my encouragement

(01:18:11):
to traders is this. Getcrystal clear on what your strategy
does and then build it to doexactly that. And for investors,
look beyond the surfacemetrics and Ask whether the strategy
you're buying into is actuallydesigned to meet the role you need
it to play. So in trendfollowing, alignment between purpose

(01:18:31):
and design isn't justimportant to me, it's everything.
So there you go, Niels.
No, no, I mean great. Let meadd a few thoughts to the point that
you mentioned here. I have twoquestions for you. One is, is there
a number? And I know it'sgoing to be an approximate number

(01:18:53):
bar where you would say wellanything above this is not going
to give me more independent.Because you mentioned the word independent
markets or bets. So that's onething. Because I do think that people
also misuse that a little bitsaying well, 500 markets surely will
be better. I don't agree withthat. So I wanted to ask you if you

(01:19:17):
in your research have roughlyan idea of what that number will
be where you feel I've gotmaximum diversification, that's one
point. And then the otherpoint, and I do think this is relevant
because it does introduce someother factors and that is would you
go in order to have moremarkets, would you go off exchange

(01:19:39):
introducing counterpartyrisks? Would you go to countries
where maybe the regulation,the regime is less as we're used
to in the Western world worldand introduces other risks. So what
are your thoughts on that interms of your preferences?
So my preferences first marketdiversification, to me, that is limited

(01:20:06):
by your capital. I wouldcontinue to diversify capital. So
let's say I had adiversification of 100 markets and
my capital significantlyincreased. So typically there's the
decision how do I scale up myresults? Do I increase position sizing
or do I invest that extracapital in new markets or do I invest

(01:20:29):
that extra capital in newsystems? So my priority would be
always invest in new marketsprovided. And this will flow into
the second question, alwaysinvest in new markets. I do find
system diversification to memaxes out at about 30, 10 different
trend following systems. Sosystem diversification is wonderful

(01:20:51):
to deliberately injectuncorrelated properties into your
market portfolios. So that'swhy I can trade Brent and Crude with
trend following ensembles andthey don't produce a correlated result.
They deliberately break thatdown. But there is a limit to the
benefit of diversification. SoI'd go 10 systems. As far as market

(01:21:12):
diversification, I wouldalways prefer to invest in more markets
and increase my positionsizing, for instance.
But isn't there a limit richthough where you say actually at
market number 251 I'm notgetting, I can't find Any independent
markets or I don't want to goto a certain region of the world

(01:21:32):
where, you know, there'sdefinitely those limits.
So there's definitely. So thatliquidity is another key requirement.
I do require liquidity andalso the additional risks you mentioned
going off exchange and things,I don't think it's worth the risk.
So within that it really doescap you out as we know. But if for

(01:21:54):
instance it could increase, Iprobably would go with it because
my underlying mantra is I justdon't know where the next outlier
is going to come from. I doknow that any liquid market over
the long term has these fattail properties. So yeah, I would
tend. That's why to me, youknow that Pierce Brosnan the world

(01:22:15):
is not enough James Bond I saydiversification is never enough.
That's a good way to end ourconversation, but I don't want to
end it completely. I do wantto say a big thank you for providing
this wonderful framework ofdiscussing trend in a new light.
That's really wonderful. I'msure people will enjoy that and find

(01:22:38):
a lot of benefit from this andhopefully implement some of these
things. And if you do, let usknow by the way if you've gone out
and quantified or categorizedmanagers in these buckets. That would
be fantastic. I do want to goback to AI though. Before we finish,
I completely forgot to mentiontwo things to you. You. So in terms

(01:23:02):
of AI, this recording platformwe're recording on actually has.
Which is new to me and. And ithas this feature where I noticed
that in the post production itsays oh I can do text to speech and
it will be used in AI. So Ithought that could be fun. So I just
posted in a snippet of what Ihad said in an episode and let it

(01:23:26):
generate a an AI voice formatme and you know what came out? It
gave me an Australian accent.
Oh there look. You can'tobject to that. Niels.
Well no, I just thought thatwas funny. Why would you choose an
Australian accent? Maybe Isound like an Australian. I did.
I don't know. I think I soundlike a Dane who'd living abroad too

(01:23:47):
long. Anyways, I thought thatwas a little bit funny since you're
on the second thing actually Icompletely forgot to say is you praised
the ChatGPT5. I read a verydifferent post critique of ChatGPT5
this morning actually someonewho was very concerned about what
was happening because itsounded like on. On for the. In this
post it sounded like whatthey're doing is they're limiting

(01:24:09):
your choices that you cannotgo back and choose number four or
number three or whatever. Nowthey're giving you just five and
they're saying, oh, we'regoing to choose the one for you that's
the best for your purpose. Butwhat they're really doing is saying,
saying, well, we're going togive you the one where we don't constrain
our system too much becauseit's bloody expensive to do these
queries. And then we're goingto say to people, oh, but if you

(01:24:32):
want choice, it cost you $200a month or whatever. So actually
this guy was very criticalabout where it was heading and not
seeing this as anyimprovement. On the contrary, for
us as users, this was actuallya much, much worse outcome. Even
though of course it's going tobe sold as the best AI that's ever
been published by Sam and hisfriends. But not just OpenAI also

(01:24:59):
all the other ones. Soanyways, just to be mindful about
these things, I just wanted tothrow that in. As I said, Rich, this
was fantastic. I can't wait tosee you in person soon here in your
Europe. If you want to say abig thanks to Rich, the best way
to do that is just to go onyour favorite podcast platform, leave

(01:25:21):
a five star rating and review.And that's going to be the best way
for other people to see theshow and listen to this conversation.
Next week I'm joined by UAVGit. He's back. So that would be
another way for us to tacklesome of your questions. So if you
have any topics that's relatedto what we like to talk about with
your of, do send me an emailinfotop traders on plug.com from

(01:25:45):
rich and me, thanks ever somuch for listening. We look forward
to being back with you nextweek. And until next time, as usual,
take care of yourself and takecare of each other.
Thanks for listening to theSystematic Investor 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
from Top Traders Unplugged. Ifyou have questions about systematic

(01:26:08):
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 infer anything
about future performance.Also, understand that there's a significant
risk of financial loss withall investment strategies, and you

(01:26:30):
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

Stuff You Should Know

Dateline NBC

New Heights with Jason & Travis Kelce

.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;}SI361: The Four Faces of Trend Following 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;}Stuff You Should Know

Dateline NBC

New Heights with Jason & Travis Kelce

All Episodes

SI361: The Four Faces of Trend Following ft. Richard Brennan

Stuff You Should Know