Trend Followers - The Unwanted Party Guests? ft. Nick Baltas

00:00:00 Intro

You’re about to join Niels Kaastrup-Larsen on a raw and honest journey into the world of systematic investing and learn about the most dependable and consistent yet often overlooked investment strategy. Welcome to the Systematic Investor Series.

00:00:22 Niels

Welcome or welcome back to this week’s edition of the Systematic Investor Series with Nick Baltas and I, Niels Kaastrup-Larsen, where each week we take the pulse of the global markets through the lens of a rules-based investor.

Nick, it is wonderful to be back with you this week. It’s been a little while, I have to say. How have you been? It’s been some time.

00:00:39 Nick

It’s been some time. I’m doing very well, Niels. Glad to be back.

I’ve been traveling quite much over the last few weeks, I would say, but now I’m back in the UK.

00:00:48 Niels

Good stuff. I know it’s been a busy time for you.

We’ve got a very solid lineup of some interesting topics, but also quite a few questions from the same person, same listener. So that’s going to be exciting. But as always, I’d love to, before we dive in, just ask you what’s been on your radar, I was just going to say since we last spoke, but it could also be in the last few weeks, whatever you find interesting at the moment to keep your eye on, but not the topics we’re going to be talking about.

00:01:23 Nick

Yeah, so a couple of things that at least I’m quite focused on or I’m trying to monitor and obviously looking into, I guess, in the trend following space. Certainly, one part of the discussion is this kind of rally that cocoa has had. And I know that you have discussed that quite substantially. It’s also in the news everywhere. I know this big rally that has helped trend followers perform. I guess I’m going to share more of an anecdote.

And that is the following, in the sense that obviously cocoa has reverted a bit a month to date, and I think it’s driving some of the kind of negative performance we’ve seen. But I would say the following, if you look historically, take out 2024. If you look historically, one of the largest negative contributors in a trend following strategy, that is cocoa.

So, the point that I want to make here is that every single time we have those conversations, not just the two of us, but generally speaking, people do ask about why do you need this breadth? Why do you need those markets that historically have never actually trended? Why do you need this asset class that historically doesn’t seem to be trending? And always the defense is like, look, there are times that different markets would actually have helped out. If we were to just choose purely on the basis of the underperformer, at the point that we put the portfolio together, maybe cocoa wouldn’t make the cut. But obviously it did make the cut because we were trying to be agnostic to the underperformer in some respect, because at the end of the day you also play the diversification game.

So, I think point number one is that one. If you had built a portfolio trying to avoid having the underperformers historically from a trend standpoint, cocoa wouldn’t be there. But guess what? You’ll have to answer another question as to why it’s not in your portfolio.

But number two, I think this year more broadly has been a year of delivering trends in markets that historically have not necessarily been the bigger drivers. I think the equity rally is one, obviously cocoa I just mentioned. So I think it’s another good example overall as to how we see those markets exhibiting persistent trends, not necessarily in a way that we can forecast, but certainly by putting them in portfolio, we can achieve to diversify away, but also capture those returns.

So that’s one data point. I mean, other interesting things are, obviously, I guess we’re all on our feet in some respect, figuratively speaking, as to how I guess the macroeconomy is going to evolve. Is it here longer now? Is it going to be no cut and maybe a hike? Is it going to be as is for now? I think this shift in macro sentiment over the first four or five months of the year has been quite substantial.
So that’s another thing that we’re spending a good amount of time.

Beyond that, I know we’re going to touch upon it at the end, but I think this equity market rally specifically on those AI names and this equity momentum kind of driving prices higher and higher and higher, this is certainly something that not only has been driving some returns, maybe outside of the CTA space, I guess when you’re looking at the single names or maybe also within the CTA space, certainly the equity managers, but also this at the same time, it raises some concerns as to whether the trade is overcrowded. So that’s another part of, I think of, I guess of interest, I should put it this way.

00:04:29 Niels

Yeah. You bring up a very good point. I mean, this idea that, you know, cocoa has obviously been the star performer in the last 12 months, but the prior 15 years, at least it was probably the worst performer.

And there are probably a few more individual markets we can think of like that. But I also think it’s, at least from my memory, it maybe is important to mention that some of the sectors that we would think would be great for trend, because they’re liquid like currencies, like equities, have not necessarily been that great to trade from a trend following perspective, at least if you trade just the major currencies, and if you trade equities via indices. I’m sure some people will argue, well, that’s exactly why you need to trade them as individual single stocks. I have no idea how that would have made it different. Obviously there have been a few outliers for sure.

So, it hasn’t really been one of those periods where these big macro driven sectors like equities and like currencies has been the best performers because we’ve had these sort of quick corrections, which are somewhat damaging to trend following. But then other sectors from memory, again, that have done well in the last 10, 15 years overall, things like energies, which is still a pretty liquid sector. But it’s just had some better, maybe trending properties. That’s from where I sit, of course.

So that’s what’s been on your radar.

There’s one thing that’s kind of came into my radar simply because suddenly I was being mentioned in a little bit of a, I wouldn’t call it a Twitter storm, because I don’t know how you define a Twitter storm. But there were a lot of tweets where I was referenced in because I think this week, maybe it was that Michael Covell published an episode with Tim Pickering.

So obviously both people have been on the podcast here, where they referenced a discussion that Tim had on TTU with Andrew Beer. So, if people haven’t listened to it, it was a debate about replication that obviously Andrew is representing, and Tim representing his style of trend following. And so, they sort of brought that back to live and had a good, well, they had a conversation, obviously, one sided, because Andrew was not there to talk his side of things.

But, you know, it was an interesting conversation. I don’t think it was too controversial, even though some people might feel that. But I wanted, I couldn’t remember, I’m sure we did, but I simply can’t remember if you and I talked about, back then, the pros and the cons, from your perspective, to replication of these strategies.

So, I just wanted to throw that in there, because it kind of kept popping up on my little Twitter feed this week.

00:07:39 Nick

Yeah, I guess a couple of points. I do remember this episode quite vividly, actually, that was an interesting one, I guess, in a variety of ways.

I certainly enjoyed it. And I believe I was on maybe the week after. So, we had actually, we had discussed a few points around, you know, how do we consider alpha? And how do we define alpha? Is it like a quantifiable measure? Is it more like a heuristic as to how we, I guess, associate investor skill with our performance? And I guess that’s data point number one.

Data point number two, I’ve not actually seen the Twitter conversation. So basically, I’m just going to, I guess, respond to your question on replication, you know, I guess, in a more kind of generic fashion in this regard. Let me put it this way.

Replication is dependent upon its success to replicate by nature, but also, it’s dependent on the success of the underlying instrument we’re trying to replicate. So somehow replicating a strategy suggests that the underlying strategy is in itself quite successful in terms of performing. So that’s, I think, a, I guess, a fundamental requirement for the replication to deliver performance, that the underlying is performing. Now, beyond that, I think some of the challenges of replication, and I believe Andrew has strong views on this one, is how quick you can actually turn the corner in a style drift, or maybe in a repositioning of a strategy.

So, I would imagine faster programs are harder to replicate, or maybe somebody would need to adapt the way that they estimate their models. But to a certain extent, the fact that the CTA industry in some respect is capturing broader macro trends, and maybe nuances beyond those macro trends can allow, I believe, some form of replication to be successful in this regard. Now, the magnitude of tweets or the measurement of success, I think that’s something it’s not easy to, I guess, comment upon without actually looking into the data.

But I can certainly see the reasoning for it. Whether I’m a fan or otherwise, I think that’s another question. But that’s how we’d look at it.

That’s how we’d look at it. I think, if anything, I see the value of grouping together investment profiles and then trying to find a way to achieve that in an efficient manner with a limited set of markets. We can obviously debate as to whether 10 or 20 or 30 or 40 markets are enough. And I’m pretty sure, again, Andrew has specific views on this one. But that’s where I am when it comes to replication.

00:10:13 Niels

Obviously, diplomatic as always, Nick.

But I think you bring up some interesting observations. Now, it’s good for me to now look back at that conversation, because now we’ve had more time. And there’s been some interesting market moves since then, and markets that are not traded by Andrew in his replication.

So, my sort of, and I haven’t sort of done any analysis per se. My general observation is, I’m impressed with Andrew’s ability to keep up with the SocGen CTA index in a time where clearly a lot of CTAs have been making money from cocoa, which is a market he does not trade. I think that is impressive.

But the other thought I have in my mind is, what is most important? That he can track it with very high daily correlation or that he can outperform, as he says he will, over time because of the lower fees. That is the point that I don’t know, because, again, from not following this every day or whatever, but there certainly have, and I’ve discussed this with Andrew, there certainly have been periods where the correlation has been that strong, and there’s been some underperformance and then some strong outperformance and so on and so forth. So, to me, obviously, that does ask some questions, because for me, it’s not just a question of, can you outperform because you have lower fees? It is, how closely can you actually track the thing you’re trying to replicate? That, to me, is quite important, if that is your aim.

But I don’t really want to go into replication today, because we have so many other things we need to talk about. But it is a debate that seems that will come back. And now that we will soon have more entries in the ETF space from people we both know very well, soon to be talked about on this podcast, it’ll be a debate that’ll continue.

00:12:20 Nick

Look, looking forward, the one thing I would say, and I agree with you, correlation is very different to performance. Two things can move together, one cannot perform.

So, I think that’s the point that maybe Andrew would need to comment upon. But I guess we’ll leave it here, and you’re going to bring him back as you do.

00:12:37 Niels

We’ll leave it here.

We’ll turn our attention to the month of May, maybe just the last, not just the last week as such, because we did have a pretty soft start to the month of May. A lot of markets turned, not on the same day, but in the first few days of May, it was like one thing after the other. So, CTAs took a little bit of a hit, obviously giving back some of the open profits we made in the first few months of the year.

So, nothing dramatic. I know this week has been a bit better, some stabilization, for sure. But I wanted to, since we do the same thing, I wanted to ask you what you’ve seen in May so far.

00:13:15 Nick

So, May started a bit, I mean, I would not necessarily call it like a perfect storm in a way, but there were effectively a few moves that seemingly were unrelated, but somehow, I guess, turned the corner for most of the trend following programs, I guess, on the negative side, in the sense that yields reverted. That was one move. The JPY rallied, I guess, on the intervention risk expectations.

Then you had some of the energy longs actually suffering a dip. And clearly, we’ve mentioned it, cocoa kind of turned around. So, in some respect, across the asset classes beyond equities that have actually managed to reduce some of the, I guess, down moves, we’ve seen the other three asset classes delivering negative returns.

I don’t think there’s any fundamental reason underlying all those moves between them and across them. But certainly, that was a time, over the first two, three, four days, that negative returns came through and kind of compiled on top of each other, as well as through time. It has been a bit more of a normalization this week, as you said.

So overall, the month still has a good amount of time to actually even go back to positive territory, at least the way that we look at it. Is that how you look at it on your side?

00:43:32 Niels

Yeah, absolutely. I would say the only sector that’s up so far this month is equities, and the rest is down.

So obviously, it will be one of those months where you’re down a little bit more than maybe an average down month. But as I said, nothing dramatic. It’s a great year so far.

00:14:50 Nick

It’s a great year so far.

00:13:51 Niels

Yeah, exactly. It’s a great year. And so, speaking about that, let’s look at the numbers.

So, the BTOP50 is down 2.66% for May, still up 8.18% for the year. SocGen CTA index down 3% for May, or in May, as of Wednesday, I should just add that, up 8.5% for the year. SocGen trend down 3.7% in May, but still up about 10% for the year.

And the SocGen short-term traders index down 70 basis points, up 2.85%. In contrast, the MSCI World is up 3.2% after a bad month, and up 7.63% for the year. The World Government Bond Index is up 86 basis points for the month of May, I should say. And the S&P 500 total return is up 3.58% as of last night, up 9.83% for the year.

Now, before we turn to the questions, I do need to just mention that I have finally published the sixth edition of the ultimate guide to the best investment books. And we now have a collection of more than 400 books that people can look at and get inspiration to dive deeper in and learn more. And the guide is available via the TTU website, so just go there and you’ll find it.

Just scroll down. But also, if you already registered for some resource from us, then you should have had an email in the last few days where I sent it to you free of charge, of course. So just be aware of that.

It may have ended up in your spam folder if you haven’t received it. Otherwise, as I said, you can download it free from the website. Okay.

Now, before we dive into your topics, we actually got quite a few very good questions from Peter. And they are relating to a few things that I know you will expertly guide us through.

So, the first one, Peter writes, have you ever discussed CFM’s paper on agnostic risk parity on TTU? If not, I’ll be really interested to understand whether you and guests believe that signal direction is sacrosanct or whether you believe that sacrificing a degree of directionality can help to create a more robust portfolio.

Simple example, if you have two assets with very high historic pairwise correlation, and one asset has a very strong buy signal while the other one has a very weak buy signal, would it be scarcely to be very long the first assets and slightly short the second to offset some of the volatility? And then Peter goes on to say maybe a slight rephrasing of the question above would be, should signal direction always trump signal strength in portfolio construction? So let’s start with that.

00:17:49 Nick

Super. So, I think this touches upon maybe a fundamental debate when it comes to building trend following strategies. And that is, I guess, starting from the rephrasing of the question, should I always go long appreciating markets and go short depreciating markets or not? Some people say that this is your pure trend, maybe with some scaling related to that purity argument. But is the direction important or are there cases where we can sacrifice some of that kind of directional bet to promote diversification? That’s basically the question.

I know the paper by CFM, it’s about six, seven years back, I believe, on agnostic risk parity. I mean, I can explain exactly what this paper talks about, a bit more like heuristically, I guess. I think there’s a lot of value in what they put forward. I don’t generally have a strong view in favor or against, aside from acknowledging the good and maybe not necessarily the bad things about it, but at least the considerations that somebody should have about it.

So, let’s have two extreme examples. Extreme example number one is that you have those two markets that are highly correlated, and they have very good and positive trend, for instance. I don’t know, let’s take NASDAQ and S&P this year.

So, case number one is that you deploy long exposures in S&P and NASDAQ and maybe you compensate for the fact that they’re highly correlated by reducing the overall gross exposure to the two markets jointly, but you still have two long exposures. In some respect, basically we’re betting on equity markets beyond the difference between NASDAQ and S&P. So that’s case number one.

Case number two, the other extreme, would be to say, and that’s more like how mean variance typically operates, I observe two markets, very high correlation, let’s call it 95%. One has a slightly better trend than the other. Maybe the signal is 1 and 0.9, for the sake of argument.

So I’d rather just go long the one that has a signal of 1, go short the one with a signal of 0.9, I get a differential return of 0.1, but at zero risk, assuming very high level of correlation and same volatility. So now I build an RV portfolio with almost infinite Sharpe ratio, and then I can start levering up as much as I want to gross that out and get this tiny return, which is the spread between the two, at least historically, as a signal magnified under the assumption that whatever I observe as a signal is going to materialize going forward.

So two extreme cases, both of them long, moderating the exposure because they correlate highly, or one long, the other short, and I gross it up as much as I can, at least from a tradability standpoint. Now, the latter case is heavily, heavily, heavily dependent on how those two signals play out on a forward looking basis.

If suddenly NASDAQ outperforms S&P, it can be a disaster, right? Because I built a small RV trade, but then I gross it up as much as I can. This is how mean variance operates. Mean variance basically says, forget about the beta moves of the equity markets and really focus on, from a technical standpoint, the smaller eigen values of the decomposition of my correlation structure. That’s the technical side of it.

I guess the more heuristic side of it is focus on those relative value trades, try to isolate those highly correlated pairs, go long short, and basically reduce your risk by maximizing your expected Sharpe ratio. Heavily, heavily, heavily dependent on those signals playing out exactly as they were observed, as in the outperformer outperforming the underperformer, basically. Now, beyond the fact that this can, at times, create heavily concentrated positions and maybe become too exposed to what we observe not materializing, we can actually debate whether it’s called trend following or not now anymore. Because I observe two rising markets and I go long and short both of them, as in long one, short the other, and I gross it up as much as I could possibly do. So, observing those two extremes, somebody can argue that there is a middle ground.

The middle ground is what this agnostic risk part is all about. It basically says, again, I’m going to introduce some technicalities here, but if I do a principal component analysis between S&P and NASDAQ, my first principal component is basically the equally weighted portfolio between the two. It’s basically the market.

That’s my capel, that’s my market factor. Now, the second principal component that has to be orthogonal to the first is nothing more than the spread between the two, NASDAQ versus S&P, this RV trade. Okay? So, the two examples I gave go in either the first principal component and buying both markets, or in the second principal component, and they just do the RV.

The agnostic risk part says, let’s try to blend the two. In other words, let me hold long positions in the markets, as in both of them, or one of them, because if one is like an imperfect subject of the other, basically hold the market. But then I also add to it a small RV trade.

In other words, instead of me going long one too much, going short the other too much, maintain longer exposures in both, but then create an RV between the two.

00:23:28 Niels

I know that obviously doesn’t sound like trend following when you say, well, I’m actually going to go short one market that’s trending upwards just because there’s another market that has a stronger signal. But of course, the way we would do it, most of us, is just to say, well, okay, if we have a stronger signal from the NASDAQ, we’re going to have a few more long positions or contracts in that versus the one in the S&P.

00:23:28 Nick

Agreed, agreed, agreed, agreed, agreed. So certainly, within this long position between the two, somebody can scale relative to the respective signal strength by shorting…

00:24:05 Niels

Because, sorry to interrupt you again, because I lost my train of thought.

In a sense, that is also a little bit of an RV position because if S&P starts to catch up in terms of signal strength, we’re going to be adding to it.

So, I know it’s not RV the way you described before, but there is some relativity in it.

00:24:25 Nick

100%. So, to the extent that we move away from a binary signal, we’re implicitly introducing some sort of cross-sectionality.

We just do not force a shorting exposure for a long that it itself by nature is going to (and that’s the agnostic risk parity) allocate equal units of risk across all the dimensions of a principal component analysis. It basically allocates across all the eigenvectors or eigenvalues (it’s the same thing at the end of the day), allocating equal risk to those dimensions of risk that contribute risk to my portfolio. And the two extremes I gave do not balance that out.

Either they go with the dominant components, which is basically a more kind of pure trend following, or would go with the more RV trades, but explicit RV trades, and that’s more like a mean variance. So agnostic risk parity somehow sits between the two. And I think what they tried to put forward is to say, there is a limit beyond which concentration on the big macro moves brings value from a trend-following perspective, and maybe we can add to it a diversifying component.

Now, maybe we can argue that this is trend plus, right? So some of the work that we do in terms of enhancing with carry or reversion or whatever other type of characteristics, a trend-following strategy, is by adding components that don’t necessarily agree with the direction. Whether we call them agnostic risk parity that come from within, I guess, an eigenvector decomposition, or by looking into other premia, we can have a debate upon it. So I guess concluding on that part, I don’t think there is anything right or anything wrong here.

Maybe what is more concerning to me would be to do the one extreme that says, I’m going to just focus on this differential of the signal, and I’m going to build an RV and gross it up. Because I think that’s more dependent on history kind of realizing 100%, and therefore there’s too much dependence on noise. As well as concentration risk and all that lot.

I think the middle ground, or the more, I’m not going to fight my signal direction, are equally plausible. And I think at the end of the day, it’s also like a philosophical point. I can see arguments for both, let’s put it this way.

So that’s, I’m not sure if I’m answering the question, but hopefully I’m clarifying what are the data points that somebody should look at, and how this agnostic risk parity plays out in the trend following context.

00:26:56 Niels

Yeah, I mean, I think you did a very good job. I was just going to make my own little observation that says, and that is, since I’m a trend follower by heart, anything that’s not pure trend, I wonder why do we call it trend plus? Why not just call it trend minus? Because in the long term, trend following hopefully will outperform all the other ones. So, there we are.

00:23:28 Nick

It’s a matter of cannibalizing or enhancing it, I guess. But I’m with you in the sense that I typically prefer to, I guess, focus on the direction and then let other things playing out beyond that part. But anyhow, that’s a good one. It’s a very good paper.

00:27:39 Niels

All right. Okay. So that was the first two of five questions from Peter.

The next one, and we don’t do it in the order that we got them, we just kind of put them together where we think they fit. The next question is, is it reasonable to aim for broadly equal risk allocation across different asset classes? I must confess that to me, it feels like a suboptimal portfolio construction solution given, for example, the lower pairwise correlations of individual commodity markets than those of individual equity markets. However, very mindful that correlations change over time and that it’s easy to overfit models by adding complexity.

Thoughts?

00:28:18 Nick

So, this basically says, how do I split my risk? Now, the first thing I would want to comment upon that is the following. Any weighting scheme we apply between asset classes or between markets, even if it’s not explicit, it has an implied expectation of return. That goes back to Black Littleman, the 90s framework as to how you introduce your views. If you work around the algebra, you can also express any weighting scheme as a forecast of returns.

So let me give an example. If I have a 50/50 portfolio between equities and bonds, every single time one of the two is outperforming, I need to sell it to buy the other. So, in itself, a fixed weighted portfolio has an implicit value bet within it because I’m selling what goes up and I’m buying whatever goes down so that I can go back to this fixed weighting scheme, whatever that is, 60/40, 80/20. So, there’s an embedded valuation as a prediction of my return every time I follow a scheme that follows a fixed weight structure. So doing 1 over a vol, in principle, between asset classes implicitly suggests that I expect all of them to deliver the same Sharpe ratio, pretty much. So that’s the optimality here.

Now, is that the right thing to do? Here’s another philosophical point. Do I get agnostic about what is the contribution one asset class is going to bring to my overall portfolio? Or I would want to allow my portfolio to navigate not only within assets, but also between asset classes into those that trend more and those that do not trend as much. So, if you go back in 2022, I mean, with perfect foresight, we did not want to be in equities at all, because they faced this kind of downside, whipsawing dynamic, did not add any value by just being short bonds would be good enough. But equally, we wouldn’t have the chance to move out of commodities post-March and get into the dollar.

So, I think those dynamic risk allocations between markets, at least in my view, produce a, I don’t know, even if I were to say better portfolio, I need to define what better means here. But anyway, I think at least in my mind, allocating risk where the opportunity lies is prudent. Now, beyond that, and I think the question, you know, in some respect, maybe incorporates that subtle point, that being, it talks about correlations. If I say that I’m going to allocate equal risks or whatever those risks I can determine, it’s not a volatility scaling exercise only. It is also a correlation exercise.

So, in some respect, unless I take into account correlations, I do not actually deliver the risk contribution I’m expecting to deliver. Let me give an example. If I have two markets with very high correlation and two markets with very low correlation, all of them having the same vol, if I allocate equal dollars to those four, I’m heavily more exposed to the two that are more correlated with each other. So, I somehow, by dollar allocating equal units, because simply they happen to have the same vol, I’m ignoring the fact that they are actually correlated.

So, all else been equal, for me to deliver those equal risks, I need to downscale in dollar terms, not in risk terms, the highly correlated markets, because their own correlation will bring that risk up to balance it out with the rest. So, another way of interpreting the question is more about when I allocate on some risk basis, do I effectively do so by simply scaling by vol? My answer is no, because then you’re creating blind spots on correlated clusters, as well as loosely correlated clusters.

The former are over-represented to what you thought they are, and the latter under-represented to what you thought they are. And I think at the end of the day, translating risk into dollar is ultimately the most important part when it comes to building those multi-asset portfolios, in my view.

00:32:16 Niels

It’s worth clarifying, of course, is that, first of all, I think CTAs or trend follows specifically, I can’t speak for everyone, of course, but a few at least that I know, I don’t think it’s like we sit down and say, OK, we’re going to have the same risk in currencies, bonds, equities and commodities at the same time.

I think a lot of managers basically take the approach and say, OK, these are the markets that we want to trade. And yeah, there’s going to be some sectors that have more markets than others, of course. There are not 20 meat contracts, there are only two of them. So, you know, for example.

But I think what many people do is simply just to say, we’re going to treat them equally in the sense we’re going to give them the same opportunity. So, if LeanHawks has the same strength of trend as the S&P, we should be risking the same amount.

But it’s always dependent on the strength of the trend of the individual market. We’re not forcing a risk allocation onto a market that’s not trending just because we want to have equal risk allocation between sectors. We don’t do that.

So, I think that’s important because giving the same opportunity set is not the same as saying that we have the same risk allocation at all times. We don’t. And that’s what I like about the dynamic way that trend following works is that it really does seek the opportunities and will obviously chase them and sometimes will chase them without any joy.

But at least we have that adaptability built in and not trying to predict what one asset class or market is going to do versus another based on some historical data, which may not be true. So, I think that’s what I like about a very open minded type of approach to allocating the actual risk.

00:34:37 Nick

I think that’s a nicer way of basically saying my initial point, which is forcing some risk budgets, in itself, implies expectation of returns.

00:34:48 Niels

Exactly.

00:34:49 Nick

And your point is by being more dynamic and letting those budgets being more dynamic, to the extent possible, you allow yourself to capture the opportunity wherever it lies, as opposed to forcing it to be allocating into maybe non-trending markets.

So, not forcing effectively expected returns in the weighting scheme, even if you think that you don’t impose them, you actually do.

00:35:08 Niels

Which, of course, risk parity does. Exactly.

00:35:12 Nick

Yeah. Risk parity basically says I’m forcing or I’m expecting that they are going to deliver equal operations.

Pretty much. That’s what it says.

00:35:21 Niels

All right.

Next question. If you were an institutional investor with an ultra-long time horizon, a stomach for short term volatility and a mandate to deliver the highest possible KGAR over the long term, how much of your portfolio would you allocate to systematic trend following? Nick Baltas,, what is your answer?

00:35:43 Nick

Here’s the question. I mean, we can look at it.

If you look at it mathematically, right, maximizing the returns over the longer term, all else being equal, no constraints applied. If you have trend following and maybe other liquid assets, you would do 100% trend following. That’s just basic exercise of looking into the higher historical expectation of return vis-a-vis a 60/40 portfolio.

Now, that, in itself, would make the institutional investor, however, the odd one out, because I believe there is a lot of, I guess, fixation to some policy portfolio. So, what we disregard here, and maybe have some views, is that institutional investors, they don’t just focus on maximizing long term returns, because if they did, that’s a very different utility function to making sure there’s liquidity in the structure, or maybe there is some tracking error to a policy portfolio that they’re, at the end of the day, assessed against. So, there’s all this kind of intermediation, as well as (how should I put it), I guess, a relative risk to whatever I could anyway hold, being a 60/40 portfolio.

And this active risk is adding career risk, and this career risk is adding aversion to deviating from it. So, in that context, there is no optimal allocation, and it all becomes a utility function between how much I believe I can help in the downside scenario of my 60/40, whatever, policy portfolio I have, and how much tracking error I can allow myself to deviate from that 60/40. So, it’s a balancing act between improving downside risk, and obviously, in that case, maximizing cargo over the longer term, and tracking error.

So I don’t have a straight answer, because there is none, in my view. What’s your view?

00:37:49 Niels

Well, I was a little bit surprised when you said, if you just look at the math, you would take 100% trend. I didn’t expect that.

And the reason I say that is that…

00:38:02 Nick

Maybe it’s 70% or 80%, but it’s very, very high. Let me put it this way.

00:38:06 Niels

Sure, sure. I agree. I agree. I agree with that.

The only reason I say that is that if there is actually one thing I do track on a monthly basis is we have kind of a web-based portfolio tool that our clients and prospects can use if they go to the DUNN website. And when you do that, a simple 50% S&P and 50% DUNN actually outperforms even our own returns. And that’s obviously because you get some correlation benefits, so you’re compounding at a higher level at all times. So, I couldn’t put my hand on my heart and say, well, actually, 100% DUNN would have done better. It wouldn’t. But obviously, these are historical numbers and so on and so forth. But anyways, it is a much higher number than most people would accept.

And that’s actually also something that many years ago, I had a confirmation. It’s one of the early episodes I did with three of the largest consultants in the world, actually. And we talked about this. And they also admitted to saying, well, if we just looked at our analysis, we would have to pitch to our clients a much, much higher allocation. I think they mentioned like 45%, 50% back then. And of course, we know that that’s not going to fly.

All right. A final question from Peter. And thank you very much for these questions, Peter. Very helpful. He writes, I enjoyed your conversation with Anthony Todd from Aspect recently, and in particular where he described trend alpha coming in bursts. I recently read an interesting paper suggesting that trend following works particularly well during periods of higher than average volatility for a given asset, and that its Sharpe ratio is much in periods of benign volatility for that asset.

This was then used as a justification for trend following in alternative markets, given the greater frequency of high volatility environments in those markets. Would you agree with this?

00:40:09 Nick

So, maybe I would agree with some additional conditions in the following sense. I mean, the question initially reminded me of some of the work we had done back in the days.

As you maybe remember, the first time I came in was about this Financial Times article on inflation and trend following strategies that was back in 2022, where we were making kind of the point that before 2020, part of the reason why trend following was not necessarily delivering returns in line with history was the increased fundamental certainty and subdued fundamental disagreement in the sense that when we all agree about valuation, and our views are very tightly linked with some consensus, any price deviation from an exogenous shock would make us more, I guess, keen to play the reversion trade. Because we’re all kind of believe that this table costs 100, and you might say 99, I might say 101, but it’s basically 100. And then what happens post COVID, in some respect, it was like this shift to fundamental disagreement and this kind of dispersion of views that in itself can become price volatility, whereby you’re much more pessimistic, I might be more optimistic, we might still have the same kind of average value in terms of valuation, but the distribution of our views is so much wider. Parallel inflation, transit inflation, here to stay, not here to stay, higher for longer, whatever that might be, this disagreement in valuation, why does the distribution of potential outcomes, and in that context, any price movement seemingly contains more information to the extent that it shifts our future distribution of views, and therefore allows trend following to actually perform in some respect?

So, there is this potential categorization of market regimes between those of higher fundamental certainty and those of lower fundamental certainty and how they can relate to regimes, but also markets that can deliver more trend following behavior than not. So that was the first idea that came to mind with this question.

The second idea that came to mind with this question relates to some academic work done maybe 10 years ago, maybe more, that basically talks about a return predictability, even if you don’t have any serial correlation in the return pattern, and this comes if volatility is clustered. So, if volatility becomes higher and stays high, it is more likely that a big move is followed by another big move in the same direction, that creates skewness in the overall return distribution that in itself creates trend following.

So, you can perceive the price process as if it’s having a trend, but underlying that price process, there is no serial correlation. It is rather that the volatility itself is serially correlated in clusters. So that is also kind of related to this question.

Having said that, however, I do not necessarily believe that it’s higher volatility necessarily the reason why alternative markets can deliver more, I would say, more trend following behavior. I mean, yes, it can be one condition, but not necessarily the necessary condition in some respect. I mean, I guess reversing the argument, looking into the more conventional asset classes, it’s actually the lower volatile assets between them that have historically delivered stronger returns. And we’re talking about duration here.

So yes, volatility can be one factor for the overall strategy for the single markets, but there’s a few more kind of nuances in place. So, for example, alternative markets might have a very segregated marketplace between, if you talk about some of those alternative commodity markets, segregated market between producers, consumers, as well as speculators.

And I think those imbalances between the core users of the asset rather than the speculators on the other side might create more trendy behavior without that necessarily having to be associated with volatility. So yeah, that’s kind of some of my thoughts on that question. I think it has some value within a particular context.

Does that make sense? What do you think?

00:44:32 Niels

Well, it makes sense. But I think the beauty of this is that just like the strategy of trend following itself, which should be very hard, and you shouldn’t try to put it in a box. I think even this question is hard to answer without putting it in a box and I don’t think we should do that. The reason I say that is, I’m sure you can find examples. Let’s take cocoa, right? Cocoa was a nice, relatively smooth uptrend last year, then it went parabolic this year.

And so, there were plenty of opportunities last year. It didn’t have to go parabolic per se. Of course, there were, you know, for those who didn’t adjust the position, of course, has made some astronomical returns just hanging onto it with massive volatility, of course.

But that’s just one example. Another example is, of course, we had the bond trends in late 2021 and onwards. And there was not really a lot of panic in 2022. Equities and bonds were just moving down, not in a big panic, just in a nice downtrend. And we were there to capture them.

What I like about trend following is the fact that, you know, it does adapt to many different conditions. Sometimes it’s volatility and sometimes it’s not. What I’ve discussed on some of the other series, the global macro series with your colleague, for example, Christian Mueller-Glissmann and Cem, is that what I think is maybe more interesting is inflation, for example, as a factor.

And just say, when there is more volatility in inflation, that seems to have an impact on trending opportunities in many different markets. That seems to me to generate some opportunities compared to the very stable, low inflationary environment we had. But can we say that higher market volatility always gives better performance? I don’t think we can.

00:46:39 Nick

Not necessarily, right? Specifically, not in the transition.

00:446:42 Niels

Yeah. Yeah, exactly.

So anyways, thank you so much again, Peter, for your questions. They were very good. And that gave us an opportunity to, you know, dive into some interesting topics.

But we’ve saved the best for last, of course, Nick, because you’ve been busy. You’ve also wrote an article in the Financial Times. And therefore, since you’re here, I’m going to let you take us through that article and just enjoy your…

00:47:16 Nick

Summary, I guess.

00:47:17 Niels

Summary. I think that was the word I was looking for.

00:47:18 Nick

Thanks, Niels.

I mean, what’s the background? This was published by the Financial Times last week, and I have to thank Robin for hosting it on the Alphaville. The title is actually, I think it’s quite catchy in the sense that, you know, it basically talks about momentum, parties, and hangovers, and how is it possible that now I can connect parties and hangovers with, like, momentum? It talks about equity momentum.

So, I guess, for the benefit of the audience, we’re not focusing for this particular part on the typical gross asset trend following. Instead, we’re looking into equity momentum. I believe the terms momentum and trend have been used interchangeably in the industry.

Being a bit more strict, at least with the semantics, at least in my lingo, trend is about a univariate movement that a market that is appreciating continues, whereas momentum is more of a cross-sectional one whereby outperforming markets continue outperforming the underperforming ones, but they don’t necessarily have to have a positive return, per se, or a negative one. It’s a relative game where the winners keep on winning and the losers keep on losing.

So why parties, why hangovers, why this is relevant and why this is topical? Maybe you have seen, since the beginning of the year, equity momentum in the US, as well as globally, but I think in the US it has been a bit more pronounced, has had a very, very, very strong start. It has been by far the best quarter. And that’s not even a quarter out of the four in a calendar year. It’s any kind of three-month period.

It’s by far the largest performance that equity momentum has had at least for the last 20, 25 years in the US, maybe doubling the second best. And this is primarily driven by the AI hype, all the tech stocks and the NVIDIAs of the world and so on and so forth.

Now clearly, it’s a trade that has been very successful. I think part of that has allowed the indices obviously to appreciate substantially because of the amount of concentration they bring to the indices. And therefore, the index performance has also helped trend followers, kind of connecting all the dots between our prior discussion.

But here is now the big question, because when we look into momentum in equities, the more it performs, the more we want to buy the same names that have been performing and the more we’re buying them, the higher they go. And the higher they go, the more we want to buy them.

So, this kind of self-fulfilling prophecy, this positive feedback loop, you end up having portfolios that get too concentrated. Within the broader indices, the respective sectors that have been driving equity momentum have by themselves become dominant. So, the level of concentration has been rising.

If you look into hedge fund positioning, they have had, over the last couple of months, the largest or one of the largest exposures into the momentum factor. This can either be explicit by following those trends or implicit by a respective stock selection that might not necessarily be driven by momentum per se, but that might become a byproduct of a stock selection exercise.

So, the question now becomes, when is it going to end? And if it’s going to end, how is it going to end? Because historically, we have seen momentum crashing and the academic literature is quite rich, talking about momentum crashes. And if it were to crash, then obviously, what could be the consequences of that crash? By the way, it’s now already still, sorry, it’s actually now half of Q2. Wow, it’s actually half of Q2 almost, right? 10th of May today. And the momentum trade keeps on performing.

So that was what I attempted to address as in the need to time momentum or maybe share considerations and implications of timing momentum and basically try to conclude by the fact that it’s very, very hard to time that. So, there are a number of things that maybe an investor can do when it comes to experiencing this momentum rally and basically benefiting from the recent performance, but being quite concerned about the future.

Now, how do parties and hangovers come along? I tried to relate, if you like, a momentum trade to how a social gathering operates. You basically want to go to a party or a gathering in search for amusement in the same way as momentum traders pile on to momentum trades on purpose to get financial returns. But there’s a precondition for those trades as well as those parties to be successful, and that is that other people come.

So, there is an early stage crowding requirement for the trade to, I guess, take off. Now, the more it takes off, the greater the fun, respectively, the return we’re getting out of it. And there comes a time whereby you need to continue buying exactly what you have had, or maybe try to make more fun out of being in the same party.

And the catalysts here might be deploying leverage or basically getting a few drinks maybe, or having the music playing on the background, or maybe, I don’t know, putting the volume higher. So, these are just catalysts that can make not only human behavior, but also financial investment more successful in a momentum rally. The problem is when once you get to midnight or 1am or 2am, you know deterministically the following morning is actually coming.

So, in principle, you can just go and you’re like, look, I’m out. I’ve had my fun. I’m kind of stepping out and then see you tomorrow. The problem is that behaviorally, we have the so-called fear of missing out. It’s basically being teased out by our friends the day after. Look, you should have stayed longer. We had fun until early morning.

Obviously, that might sound a bit humorous, but when it comes to financial investments, when it comes to momentum trading, the fact that a lot of wealth is managed by an intermediary, being an asset manager, that themselves compete for capital and therefore they are very adverse to redemptions. There is not only empirical results, but also research documenting the fact that momentum traders acknowledge the cross risk, but it’s actually the fear of a redemption. In other words, the risk that is associated with a tactical decision to get out of a trade, specifically at a time that is kind of rallying upwards and upwards and upwards.

So, the fear of missing out, I think there’s a nice analogy here between how human behavior operates in a social gathering and how a momentum trade operates when it comes to the late hours of performance. By all means, momentum in itself is anchored to cognitive biases, so it’s also a behavioral pattern itself. But then again, we’re 2 a.m. or 3 a.m. or whatever the time might be, and we know that some sort of a crash is coming.

That crash can be driven by two things, in a very simplistic but also generic sense. It can be something very endogenous, like funding liquidity is not there anymore, and credit lines basically say, look, you’ve grossed out your portfolio as much as you could, no way we can actually go beyond that, which is kind of equivalent to basically saying, drinks are running out. So, in some respect, you can start observing how gross exposures play out in the market and maybe get a sense as to whether a turning point is coming.

Again, it’s very, very hard to call the turning point, but maybe these are some data points somebody can follow to effectively avoid a fire sale or a short squeeze dynamic where everyone is running for the exit, and I guess the door is small. It’s ultimately a liquidity discussion at the end of the day. How quickly can we get out?

Another reason can be very exogenous. You can basically talk about a neighbor calling the police, which is maybe relevant to having a very idiosyncratic shock in the market. That might be, I don’t know, that the AI hype suddenly fades because somehow there’s some negative connotation with the use of AI. I don’t know. I’m just coming up with a random excuse that could, in turn, tear the trade down. Obviously in this regard, it’s even harder to foresee that possibility.

Whichever the reason might be, a downturn, or if you like, some, I wouldn’t necessarily call it a crash. I don’t want to be predicting crashes, but certainly it’s part of the cycle of momentum trades that some underperformance can come. It can even be value investors coming in, basically buying the chips only expensive.

It’s all that I’m basically trying to put together and concluding by arguing, if you cannot fight the trend, at least try the best you can to take advantage of it, but doing it in a more risk prudent way. Somebody can argue that by vol scaling you are exposing to equity momentum. You kind of reduce your participation when it’s volatility increases, which a nice illustration or analogy is by turning it into water as the night progresses. So, try to be less exposed to it while still maintaining some exposure to it, so moderating it.

It can be about being more selective. There is some nice research recently that came out that shows that those losing stocks that are far away from their recent highs are the ones that typically rebound more than the rest. So, avoiding being allocated to those losers might be a way to help, in this particular case, or maybe completely hedge out momentum exposure, whichever your momentum exposure is, whatever your risk model says, try to find ways to get out of it.

That’s basically what this article is all about and effectively says to enjoy the ride and try to avoid the hangover the day after.

00:47:48 Niels

Absolutely. That was a great summary. Obviously, people should go and read the article, themselves, in the Financial Times.

As you were talking, I have never heard this analogy with a party and a hangover and all of that. As you were talking, I was trying to think on my feeders here. What kind of guest would a trend follower be, and maybe there’s a reason why we never get invited because you could say that, well, if you’re arranging a party, depending whether the party takes off straight away or whether it’s a really slow start, you don’t know when the trend follower will actually arrive, whether it’s at 6PM or 8PM.

If the party gets completely crazy, and lots of people have too much fun, you know we’re going to leave, at least if you have dynamic position sizing. Once the thing gets crazy we’re going to be scaling out, and leave the party. So, maybe we’re not the best guests to be invited. We’re a bit too unpredictable, to turn it back to Anthony Todd’s conversation, Unpredictable Alpha.

00:58:49 Nick

Look, if anything cocoa exposure has been dialed down before the downturn came just because the spike up was recorded as a high vol spike that made the signal look weaker even if it was still going up. In some respects that’s exactly what you are saying. We had down scaled.

00:59:06 Niels

Yeah, exactly, exactly what I’m saying.

Alright, well I think we’ve certainly covered a lot of ground. I really appreciate all the preparations and insights that you shared, Nick, as I’m sure our audience does as well. So, as usual, I’m going to mention that if you appreciate what Nick has just informed you about, as well as all the other wonderful co-hosts on the podcast, why don’t you show your appreciation by going to iTunes, Spotify, Amazon and leave a rating and review, and share the episodes. It is the best way to show that you get some value from these conversations.

Next week I’m joined by Mark, so, another trend follower coming. Usually Mark brings a lot of different topics along as well which is always wonderful and surprising. So, if you have any questions for Mark, send them to info@toptradersunplugged.com and I’ll do my best to get you some answers.

From Nick and me, thanks ever so much for listening. We look forward to being back with you next week, and, as usual, in the meantime, take care of yourself and take care of each other.

01:00:22 Ending

Thanks for listening to the Systematic Investor podcast series. If you enjoyed this series, go on over to iTunes and leave an honest rating and review and be sure to listen to all the other episodes from Top Traders Unplugged. If you have questions about systematic investing, send us an email with the word “question” in the subject line to info@toptradersunplugged.com, and we’ll try to get it on the show.

Remember, all the discussion that we have about investment performance is about the past and past performance does not guarantee or even infer anything about future performance. Also understand that there’s a significant risk of financial loss with all investment strategies and you need to request and understand the specific risks from the investment manager about their products before you make investment decisions.

Thanks for spending some of your valuable time with us and we’ll see you on the next episode of the Systematic Investor.