Design or Luck: Why Trend Following Results Diverge ft. Katy Kaminski

[00:00:01] 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.

[00:00:23] Niels

Welcome and welcome back to this week’s edition of the Systematic Investor series with Katy Kaminski and I, Niels Kaastrup-Larsen, where each week we take the pulse of the global markets through the lens of a rules-based investor. Katy, it is wonderful to have you back this week. How are you doing? What things are going on where you are?

[00:00:41] Katy

Things are good. Falls in full force here, you know, kind of back to school. I’m happy about that. And you know, everybody’s focused on the Fed and that’s about it. I don’t know, not much else.

[00:00:50] Niels

That’s true. You know, in some ways the summer kind of seemed pretty quiet in terms of news that really moved the markets. But perhaps we’re just getting used to this narrative driven news flow. But you never know, the autumn may still present some surprises and maybe even some strong trends. So, we can always hope.

Now what’s great about today’s conversation is that we’ve got some new papers from our friends at Man Group, as well as Quantica, on some of maybe the less discussed areas (if I can put it that way) within the CTA and trend following space, but also a few other things that will come up. So, it’s really relevant to be talking to you today and tackling this.

And also, before we dive into all of that, I’m always curious to know what’s been on your radar, if anything, other than the markets.

[00:01:47] Katy

Well, I think the markets, I’m just waiting, I mean, and I got several calls yesterday, both from press and clients. Everybody wants to know what is going to happen in fixed income in the US but if you look at the trend signals, they haven’t moved much.

So, I’m kind of thinking that could be a potential catalyst. And it’s something that, you know, there are so many themes, like inflation, independence, like the Fed, like changing, like allegations. I mean, there’s a lot of drama but not a lot of movement yet. So that’s why I think that’s interesting from a trend perspective.

[00:02:20] Niels

Yeah, very true, very true. So, on my radar I put three things. One is that… and the first one is a little bit related to fixed income and bonds. It’s just the European political chaos we have at the moment. I mean, the UK and France, you know, it’s kind of like a casket-case really.

And some of the polls, even in the UK, now suggest that Nigel Farage’s party would get the most votes if there was an election right now. I mean, that’s just extraordinary to think that you have like a real contender between the two main parties.

And then of course France, another Prime Minister coming in. And some of the ways that they describe the financial situation of France, obviously, should lead to a lot of uncertainty about that. And of course, if there’s anything, bond yields are very good at reacting to uncertainty, especially about budgets and these kind of things. So, it’ll be interesting to watch as well as, of course, what happens in the US at the moment.

The next thing I had on my radar, which is a kind of a new one, and this is not an endorsement of anything like that. But last night I did watch the Apple event. And I was kind of blown away by one of the products and this is this new AirPods 3 where, essentially, if what we were told yesterday is true, you’re now going to be able to have a live conversation with someone that doesn’t speak your own language. And it’ll translate simultaneously, between the languages, if they’re both wearing these AirPods 3, which I find absolutely extraordinary. But I’m also a little bit concerned. What about all those people who have a job?

[00:04:02] Katy

How do you check if it’s right?

[00:04:04] Niels

That’s the other thing, what if it’s wrong?

[00:04:07] Katy

What if you’re talking to somebody and somebody can hack in and change your conversation. I don’t know. I was like, wow, that’s so cool. That is really amazing.

[00:04:14] Niels

I didn’t think about that, but I did think about all the translators that are hired by the European Parliament or the UN to sit there and translate all this stuff. I mean, they may not have many things to do if everybody’s just wearing these new AirPods.

[00:04:30] Katy

Maybe it’ll be like, you know, TV and stuff where your kids go, you actually like had to turn the channel by standing up? And they’ll be like, you did? And you dialed the phone like that? And it’ll be like, you had to translate the language? Why didn’t you just put on your AirPods?

[00:04:45] Niels

Yeah, I know. It’s a crazy new world for sure.

And then the final thing, something I picked up from our friend Andrew’s LinkedIn feed just now, because there is apparently a Q2 2025 Brand Awareness ranking. I’ve never heard about this before.

Apparently, there is a ranking of how well hedge funds do in terms of brand awareness, and they distinguish between how well they do among consultants and how well they do among asset owners (however we define that). And I will say, it’s nice to see that some of our friends are established on the list.

The among consultants, Man Group has the top rank. Crable actually comes second. That’s a little bit of a surprise to me. Winton comes fourth, which is great. Aspect is on the list, number 20. AQR is number 19, that’s a little bit surprising to me. I thought, awareness wise, with all the research they do they would be higher up.

And then there are, in terms of the European based managers, you have CFM is doing well, top 10. Systematica is top 10. And even Andrew’s firm, DBI, a 9th place which is, you know, a great, great job on that. I imagine it’s all his appearances on Top Traders Unplugged that has helped them with all this awareness, I’m sure.

Among asset owners, it’s a little bit different, actually. CFM is doing best in terms of the CTAs we know of. Winton also coming in fourth on this one. Man Group is a little bit lower at number six. But yeah, it’s a fun list. I’ve never seen it before. I’m not sure how valuable it is but you know there’s nothing bad in being known among investors or consultants, I guess.

Sadly, I didn’t see our firms on the list, neither of them. But there we are. We’ll stay under the radar even if this was on my radar this week.

Anyways, let’s elegantly jump to more familiar waters, namely the trend following update. I mean it has been a decent start to September after a couple of good positive months for CTAs and trend followers.

Of course, equities, metals continue to do the heavy lifting I suspect, supported by a few other commodities like sugar, which has been a while since we’ve seen that, perhaps even some of the oil products are helping out. Livestock, on the other hand, is taking a little bit of a breather this month and one thing I mentioned last week in the in the podcast is, I’m not entirely sure if managers at the moment agree on whether to be long or short fixed income.

It’s funny you mentioned fixed income. When I see the daily changes on some of the mutual funds and ETFs that managers have, it really seems like fixed income plays quite a huge role, but they’re not going in the same direction, depending on whether bonds are up or bonds are down for the day. So, it’s kind of interesting. I’m always curious to hear your observations about the month, the time we’re in right now, and what you see.

[00:07:59] Katy

Yeah. So, starting with fixed income, the signals have been very mixed and kind of weak and teetering around between should we be short, should we be long? And we’ve seen a little bit more long signals in the US more recently. UK was definitely a short signal, and some of the European bonds.

[00:08:18] Niels

True.

[00:08:19] Katy

But what’s interesting to me is if, you know, someone was talking about, we were talking about SOFR yesterday. And if you look at SOFR, SOFR is at about 3.4%. And if you think about that, if the rates right now in the US are 4.25%, 4.50%, the market has already decided what’s happening. And what we’re probably going to see a trend on is whether or not they get that right. So, I think why you haven’t seen that sort of trend build-up is it’s been sort of trying to anticipate the Fed. And I think where it’s going to be interesting is, you know, if things change in a direction that’s much larger than what we’ve seen.

So, I think that’s what my take on it is, that the market is always trying to get ahead of it, figure it out. The market has already decided that we’re getting cuts. The biggest shocks are going to be if we get a lot more cuts or if we don’t get enough. So, I think those are the sort of interesting points for me.

That’s why this next cut is so interesting, because if we go for 50, that’s kind of like, well, that’s what we expected or what people wanted. But if we get nothing, that could be huge. So, I think, you know, that’s an interesting space, but the catalyst of recent jobs numbers has definitely kind of baked that in. So, we’ll have to see.

But I do agree with you. It’s been teetering around and just, you know, seems like something exciting will happen there, but I just don’t know when that’s going to happen.

[00:09:48] Niels

I completely agree. And let’s not forget what happened last time they cut, actually bond yields went up. So, I mean, you never know exactly.

[00:09:56] Katy

I know, because there are some shorts. I mean, people do have shorts and you have investors who, they’re thinking more in terms of trajectory. But if the yields have already gone down, you know, they could actually go up because somebody said, well, it’s not enough. So that’s why it’s kind of an interesting conundrum because futures predict the future rates, not today’s rates. So, yeah.

[00:10:22] Niels

Well anyways, later this month we will know exactly what the Fed decides and yeah, it’ll be, as you say, super interesting.

My own trend barometer, oddly enough (or I shouldn’t say oddly enough), it is a little bit on the weak side still, has been for a long time, but that is because it uses some shorter-term look back periods and we know that short-term models, this year, are probably the ones on a vol adjusted basis that are struggling the most. So, let’s stay with that.

So, the BTOP50 index this would be as of the 8th of September, since we’re recording one day early this week, BTOP50 is up 91 basis points, but down only now 1.93% for the year. So, that’s a good comeback. SocGen CTA index up about 1% for the month, down now only 5 1/2% for the year. SocGen Trend up 1.25%, down 6.5% for the year. And the Short-Term Traders Index, nice to see it’s having a good month, up 0.66% and down just shy of 6% so far this year.

Of course, compared to traditional markets, MSCI World up 90 basis points so far as of yesterday, up 15.1% this year. Having a great year actually. The US Aggregate Bond Index up 96 basis points, up almost 6% for the year. So not a bad year for bonds in that space. And then the S&P 500 Total Return up 86 basis points, up 11.74% so far this year. So, you know, not a bad year all around.

Now, we’re going to dive into these topics that I talked about and as we also mentioned. It’s great to have you on the show Katy, because to be able to take these deep dives with you, not just on your own papers but also when other papers come out, is really wonderful. So, thank you for doing that and for spending time getting into it.

Now, trend following has a reputation of being kind of the simplest strategy in the hedge fund world. You buy strength, you sell weakness, you cut losses, you let winners run. So how different can managers really be?

Well, the team over at Man AHL has looked into this topic in a recent paper called the Dynamics of Dispersion, not because of luck, but because of design choices; how fast you chase trends, which market you include, whether you tilt towards bonds or FX, or whether you sneak in a little bit of carry.

So, dispersion isn’t random. It is the fingerprint of each system. And as I mentioned, it is, you know, a design choice. So, let’s dive in. I’m going to let you guide through the paper. I’ll follow up with a few observations, maybe some questions. But let’s start out with describing the paper from the beginning.

[00:13:20] Katy

So, I was really excited about this paper because I’ve actually done some research very similar over the past, and we can talk about that later. But one of the key, nice aspects of the paper is that they look at four different dimensions of trend; so different things that you might decide to include in your program, and how does that vary over time, and how does that impact sort of dispersion?

And this is really important for investors because if you look at any given year, you can have quite a bit of differences between how one manager performs and another based on idiosyncratic and different themes in the markets. And so, I think investors struggle with that sometimes because markets are very idiosyncratic.

So, it can be a year where it was best just to tie your hands and be slow, or it could be a year where (and we’ll talk about Quantica’s paper later) that you happen to be in oil, and gas, and exotic contracts, and they went crazy. So that creates a lot of differences which can make things difficult for investors to understand sort of why is this manager up and this manager is not?

And I think, you know, this is kind of one of the topics that I’ve spent a lot of time researching because explaining that dispersion helps investors understand the strategies and the pros and the cons. So, what they do look at, and they use four key dimensions.

They look at speed. So, do you go faster or slower? They also look at allocation. Do you tilt to certain asset classes? And we can think of some simple examples like CTA in the ETF space this year, does not use equities. And that’s been positive for them, especially this year. And you’ll see that dispersion. Then there’s another, they also talk about whether or not they include alternative markets, so additional markets that are perhaps less liquid and maybe less easy to access. And finally, they add carry or no carry. And carry, as you know, has about a 30%, 40% correlation with trend. So, it’s another thing you might add into your basket to smooth performance over different horizons.

So, they take these four aspects and they explain how, given this variation, they can get a much wider range of possibilities of return. Just like the SG Trend index or SG CTA index has, over time, the top manager and the bottom manager can be quite far apart even in an environment where you think trend should be the same.

One of the things I really liked in this paper is they do focus on crisis alpha. I know that’s something we often talk about, and they look at which strategies perform the best during crisis periods versus others. And the thing that stuck out to me is something that I have long known and seen in the literature for this strategy is that being faster is actually really helpful for really bad environments. And the reason I point this out is that a lot of focus has been on sort of slower strategies like replication and, you know, long term kind of following the index, slow strategies. And those have actually worked much better in very, very recent periods.

But if you look through a history of the CTA index and how the industry’s changed, we did this really neat paper called CTA Style Evolution, a couple years ago, where we matched different styles and tried to replicate. And this is why replicating an index like this is hard. And what you see is that managers have changed over time.

Today you have a lot more products that I think are on the slow side. And so, I think if people are using those strategies for a crisis period, they may get a little bit less juice from those strategies based on this analysis. And that’s something, you know, I have believed in for quite some time.

[00:17:19] Niels

So just, just to point one thing out, actually also, as you describe the index and some of these challenges, I did notice that they put, on page three, that they noticed, for example, that in 2009 just one constituent stood out with a double digit positive territory and the rest was kind of more flat to negative which of course creates dispersion. But you have to know what it’s made up of.

And then they point out that in 2014, I think it was, there were two constituents that delivered more like a zero return while everyone else was doing really great. And it turned out that these two outliers, I think, were discretionary managers and they were removed from the index the year after.

So, I know a lot of people, when they see the numbers, use this as an argument for saying oh yes, that’s why you should go with replication because then you don’t have all this manager risk, etc., etc. Well, you just need to understand where the dispersion comes from. And maybe not all managers are equal, even in an index of relatively few funds. So, you need to really understand these numbers before you just say, oh, that’s a big dispersion.

[00:18:35] Katy

And you’re right. The challenge of these types of indices is they’re not directly investable. So, they represent a hypothetical version of, if you were a CTA investor from this year to this year, what you might have had (because those are the larger ones that are included). And so, they’re just a historical study of different manager returns.

Whereas today’s managers are not the same as 2003 managers. And that’s why I like that paper, because we actually tracked style change, style drift, within the industry. And, you know, there can be good reasons for that too, like innovation, like coming up with new strategies, using machine learning.

So, my view on the index is you can’t get the index today, so you don’t buy 2007’s CTA, you buy 2025’s CTA. So that’s why studies like this are so helpful. Because if you are thinking about crisis performance, you can think like, what aspects do I want in a manager so that I can be ready for that environment?

You need to think that way because when you invest in a manager today, you don’t get the CTA index behavior because they’re not the same managers as they were. So, it’s still a helpful index because it was something that someone could have invested in.

And that’s why, you know, sometimes you’ll see criticism for, you know, analysis like this. But this is the only way, the way that Mann did this is, you know, kind of create a bunch of different trend managers as hypotheticals and aggregate them over time, and examine sort of behavioral differences, in different environments, to compare sort of what could have happened without sort of cherry picking the best version, which is what people often do with a back test.

[00:20:26] Niels

Yeah. Another thing that I just want to ask you before we continue down your little story here is how much do you think, given just simple differences in volatility of the underlying managers, how much does that play in an index like this? I don’t remember if they’re all more or less the same vol or whether there are some real differences between them. But over the years, for sure, there would have been differences, I imagine.

[00:20:55] Katy

Yeah. And volatility does matter. I mean, it makes sense. You can think about (oh, I love to think of this analysis, and we can talk about it later), imagine each of these managers as a random variable. They each have a volatility and mathematically you can think about what the range of differences is going to be. It’s going to be affected by the volatility and it’s going to be affected by their underlying correlations.

And so, if you imagine strategies are changing, and correlations are changing, and volatilities are changing or different, those just naturally mathematically derive some differences in returns. Over long-time horizons, you can actually estimate those and provide theoretical foundations for that.

[00:21:38] Niels

Yeah, cool. All right, well I don’t want to stop you, so, continue down the path of this paper please.

[00:21:44] Katy

Yeah, so, there were two things that I thought were interesting in the paper. First, you know, focusing on showing that faster trend. You also saw that you that faster trend was more crisis alpha creative. They also demonstrate some outperformance of alternative markets.

I think that, you know, depending on the time horizon, there is some question about that. Quantica really sort of related that and did a really good job talking about that a little later (so we can talk about that as well). But, you know, there are definitely correlation benefits and some interesting attributes of alternative markets that have been desirable, and you can see that in this analysis as well. So, that, I thought, was also interesting.

[00:22:30] Niels

I noticed that again our friend Andrew Beer, he made a little comment on LinkedIn the other day, when this paper came out from Man. He wrote something like, my only cripple is that the model numbers seem off, well, by a lot. Over 25 years the average Sharpe ratio seems to be around 0.75. The awkward reality is that the actual Sharpe ratio of the SG Trend index is 0.36’. Is he onto something there or is it just part of this difficulty in replicating the index?

[00:23:04] Katy

So, if you think about when you use a basket of trend strategies, you know, a lot of them may be longer-term and, as I said before, some of the managers were discretionary. You pointed out that some of them were shorter-term. So, the index is not something that’s easy to replicate in a point in time.

So, my view is that I’m not surprised that Sharpe ratios are different because it’s a backtest. So, in that case, if I was thinking about replicating the index, you could do it in lots of different ways. And Alex Grazerman and I did this a little bit in our book and we can talk about that later, but, you know, you can do a basket of different window links and average them over a period of time. But we also know today that longer-term windows have done better during this history.

So, if you have more longer-term windows than shorter-term windows, and you’re a little different from the index, it is not surprising to me that the backtest Sharpe is going to be different. So, that’s why I focus more on the relative differences, and also the conditional environment differences, as something interesting to look at as opposed to trying to sort of match the actual Sharpe ratio, because I cannot replicate each of these CTAs point in time over 25 years.

[00:24:26] Niels

That makes perfect sense. And now, you brought up Alex’s name, and of course you have the leading book on trend following, I think. You wrote a lot about return dispersion, I think, in the book. Can you take us maybe back to some of what Man is doing now, and compare that to the work you did as well?

[00:24:49] Katy

Yes. Oh my gosh, like, we were obsessed with return dispersion, Alex and I, and we spent a lot of time thinking about it, and we have this one chapter and I love it. It’s chapter 11. So, declaring chapter 11. Anytime a client or someone talks to me, I’m like, you’ve got to think about chapter 11 – reading chapter 11.

It is an entire chapter of our book that’s just dedicated to analysis of return dispersion. And what’s fun about this is, this is what led us to (and I’ll talk about it a little later) some of our CTA benchmarking analysis and style analysis.

But what we do in this particular book, in chapter 11 (and this is fun to see that Man has actually done some of the similar things in their paper), we start off by looking at three different aspects of a strategy style. Ours was ‘equity bias or not’. You know, I think they added ‘carry or not’. We also added sort of allocation weights – was it equal risk or market capacity oriented – so, do you focus on having equal risk or do you like kind of focus on just the most liquid markets? And then we also talk about speed.

So, each of these dimensions we examine, and then we talk about how equal risk outperformed historically in our analysis. Again, a similar analysis of that approach in the Man article. So, that was kind of fun to see that they kind of use some similar methodologies as we did. Although they added, also, alternative markets, which was fun to see.

We then moved and looked at how do you position size and the different types of signals, which is something that Man didn’t discuss. So, do you use channel breakouts, or moving averages? And we even had these sort of random signals studies too, just to understand sort of how much of that is driven by the sort of allocation and how much of it is driven by signals themselves. So, we kind of went a little deeper into some signal analysis.

[00:27:01] Niels

Do you remember what you found? Because obviously position sizing is something that comes up often in my conversations.

[00:27:10] Katy

So, we weren’t looking at trying to performance. We were looking at sort of how much dispersion you get, because that wasn’t the goal of this particular chapter. The goal of this chapter was to demonstrate how tilting asset classes, changing position sizing approaches, how that naturally leads to dispersion and how much dispersion? So, I think our point was more, you know, look at these different aspects that all are kind of similar but a little different, that create return dispersion over history.

One thing that did stick out was we studied things like, you know, market volatility environments too. Do you have more dispersion than 2008 versus others? And we did find that 2008, given the size of the moves, regardless of which types of methodologies you use, you had a bigger range just because you had much bigger moves in that year. And this is an old analysis, but, you know, it kind of makes sense. You saw the same in 2022 as well. Everyone was up, but the range of up varied a lot.

[00:28:13] Niels

Okay. And I think there were some other things you looked into in your paper, as far as I recall.

[00:28:17] Katy

Yeah, so, I have two other aspects that were kind of fun, and I’ll talk about those two. One of them is kind of nerdy. So, I kind of like that one.

So, we also tried to think about it from the perspective of an investor, in this chapter, where we looked at sort of how much return dispersion you had as you add more managers together. And this was important because it’s hypothetical. You know, my original background was as a CTA allocator many, many years ago. And so, I was very concerned about this question. So, as you add more managers together, how much return dispersion sort of dissipates?

And what you see is that by adding at least two, maybe three together, gets you pretty far in terms of adjusting some of that return dispersion. We didn’t go into the specifics of which manager. We kind of used a random sample analysis. We randomly sampled them and then aggregated that over time.

But this is helpful because the number that kind of sticks out is 3% to 4%. So, the return dispersion actually went down quite a bit. But it’s not drastically different from return dispersion you might find in a replication approach as well.

So, with replication approach, we wrote a paper on this last year, or earlier this year, on sort of replicating, and there’s usually 500 to 600 basis points of slippage expected. So, that’s normal for an index that’s volatile, that isn’t directly investable.

And so, it kind of shows there are a lot of ways where investors can mix different things to try and reduce some of that return dispersion. But guess what? You always have it. You can’t just buy the index in the space. So, I think this analysis helps investors understand, even if you have three managers, you could still have some return dispersion that is for the whole aggregate portfolio versus the index. And that’s important if you’re an investor with two or three CTAs and you think, oh, I’m 400 basis points or 500 basis points behind the index, that can happen by chance with two random variables that have those correlations.

[00:30:25] Niels

Yeah, well, we’ve certainly seen tracking errors between even those who try to replicate a benchmark. So, for sure.

Now, I don’t know if this is in the paper, so maybe it’s just more your insights that I’m looking for. Of these four kinds of areas that drive dispersion, are there any one of those four you would say are more important, sort of the main driver of dispersion?

[00:30:59] Katy

So, we kind of showed that they can vary over time. So, in an individual year, one or the other could be very different. So, asset allocation, for example, if it’s a year like 2019 where the only trend was fixed income, you know, that could be the driver of that year’s main dispersion.

We did see some differences in terms of risk allocation over long-time horizons, but that was a relative performance choice. I think any of these particular ones will kind of stick out in a given year depending on what’s happening. And I think that’s why it’s hard because you can’t just pick one, like, oh well, if I think about this one aspect – like alternative markets. So, that’s why the Quantica paper is useful. Well, if I just have alternative markets, I don’t have any return dispersion. It’s better. That’s not true.

So, it really depends on, you know, each market environment is different, and over time, each of these factors can contribute to that. And that’s why they look at so many, because each one of them have a different environment where, hey, it wasn’t good to be fast this year; or, man, if I had a tilt.

I mean, I think the example in the ETF space is even bigger where, you know, if you had avoided equities this year, like, that was fantastic. So, any of them can do it. And that’s why we study so many.

[00:32:24] Niels

Yeah. And plus, it makes it actually pretty difficult for investors, as you say, to decide. But very helpful to know, also, that you get a lot of that dispersion disappearing if you just allocate to two or three managers, which kind of makes sense.

We’ll come to the Quantica paper shortly because I actually do find it to be a really great paper, and the first paper that I’ve seen that dives deep into this discussion we’ve had on and off about alternative markets, and so on, and so forth. But there are other papers out there, many of them you’ve been writing about. Let’s talk a little bit about some of the other things that you’ve worked on and that you found that you think are relevant for this particular conversation.

[00:33:12] Katy

So, what’s really interesting is this obsession with return dispersion caused us to do some very interesting research. It’s over 10 years ago that we started looking at this. But basically we, when we wrote our book, Alex and I, we realized, like, we’re very irritated that there’s no benchmark. So, we started sort of explaining how you could design a benchmark, and then we created style factors similar to equity.

So, if you’re thinking about equity world, it’s so great because you have the benchmark, and then you have your style factors, and then you do your regression, and ooh, you have too much small cap. And you just know what drives relative performance much easier. So, our goal was at least to tackle that type of thing.

And what we did was design, and I’ve built some other factors, and we track them at Alpha Simplex too, because when we look at a given year, like you asked before, it’s very idiosyncratic. One year it’s speed that matters, another year, smaller markets, correlated markets.

And so, by tracking those factors, you can disentangle the aggregate performance of the benchmark versus these factors. So, you can kind of try and explain some of the relative dispersion across different managers by looking at their style factor loadings. So that was, you know, a very interesting area. It’s something I continue to look at and use as a tool when I’m examining performance over different horizons and looking at different managers trying to understand, thematically, why they have differences.

[00:34:48] Niels

You know, it’s funny, as you’re talking, and I don’t remember the year, but I do remember, I think it’s one of the first live talks I’ve seen you do. It was an event in London at one of these fancy hotels. And I even forget the guy who ran that series of events.

[00:35:10] Katy

Battle of the Quants, I remember.

[00:35:13] Niels

Battle of the Quants, that’s exactly what it was.

[00:35:14] Katy

It was you and me in London. I remember that.

[00:35:15] Niels

Exactly, and I mean, this is a long time ago.

[00:35:20] Katy

It was like 15 years ago, I think.

[00:35:22] Niels

Yes, I agree. But it was interesting in the way you kind of were able to describe different managers and what they were tilted towards in terms of different factors.

[00:35:36] Katy

That’s awesome. You were at the CTA Style Factors Analysis, the first paper on this.

[00:35:41] Niels

Yeah, absolutely. I had to learn about it, of course. Yeah, absolutely. What other papers have you been, or have you delved into other papers?

[00:35:51] Katy

For return dispersion?

[00:35:53] Niels

Yeah, or similar things.

[00:35:54] Katy

I have one that’s really not as well known, but it is so nerdy and so cool. So, I want to talk about that one.

[00:36:00] Niels

Oh yeah, by all means.

[00:36:01] Katy

It’s called, Quantifying Turbulence in CTAs. And what this paper does (and this is going to get a little nerdy and I hope that’s okay), but basically, we use something called the Mahalanobis distance. So, you look at sort of the distance between different, random variables and you try to measure how much turbulence – so, movement across a group of things.

It’s really neat. It’s used a lot in the research by Mark Kritzman, who writes a lot of stuff for the FAJ and stuff like that.

But basically, you use the Mahalanobis distance to talk about how much movement there is. And what’s cool about it is, it’s a point in time measurement of the movement. Because usually, when we talk about correlation, we have to use a window of data. So, you’re kind of messing up…

So, if something big happens on one day, you need to like a year of information and you’re kind of like averaging everything when you look at correlation. What this does is it looks at point in time differences and it actually measures an instantaneous effect of movement across a group of things. So, it’s super neat, and it’s kind of you looking at it in a distance metric. Here’s where it gets even nerdier and I’m going to explain it. I’m sorry.

[00:37:21] Niels

Okay. Yeah.

[00:37:23] Katy

By taking this turbulence metric, you can decompose the movement across managers between something they call a magnitude surprise and a correlation surprise. So, a magnitude surprise is sort of like, boom, a big shift that is not a change in correlation.

[00:37:45] Niels

Right.

[00:37:45] Katy

And correlation surprise is where your big shifts across the managers is actually a shift in their underlying behavior. So, magnitude is, boom, something happens and everything reacts. Correlation is things are changing, and returns are moving, and trends are different across the manager, so they actually don’t look as similar as they normally do.

So, this paper is really cool, and why I like it is, you can actually do point-in-time analysis on a day. And so, we did this analysis for CTAs and if you look at magnitude surprises, there’s a history of them and we plot them. There are a few of them that are really big and purely magnitude, so you can guess which ones they are: Brexit, you know, SVB, Black Friday, like Liberation Day. It’s not in the data set, but I’m pretty sure it’s a magnitude surprise.

Yeah, so that’s cool because we also showed that CTAs do not do as well on magnitude surprise days. Which you’re not surprised by, I think, but that correlation surprise was more sort of an inflection point. So, periods where, you know, trends are changing, and what you do may create differences in performance as there’s sort of rotations in those trend signals and they kind of move around.

And so that is a good way to kind of think about, and it might be a good metric to understand what’s driving different behavior across your returns and understand that return dispersion. Is this driven by just a shock or is it driven by sort of actual changes in their returns? So that’s super nerdy. It told you. But it’s really cool. The pictures are really neat.

[00:39:33] Niels

I never heard about this kind of analysis. I’m sure there are a few of our listeners, as well…

[00:39:37] Katy

It’s also fun to say the Mahalanobis distance. It really sounds nerdy.

[00:39:42] Niels

Sounds very…

[00:39:44] Katy

Geeky, right?

[00:39:44] Niels

Exactly.

Now, before we dive into the last paper, the Quantica paper, let me ask you this. So, we get all these kinds of tools now, analysis tools, and people can get really detailed in trying to “explain” differences between managers and between models, etc., etc., for the purpose of people who want to invest in trend following. And again, picking one manager is probably not the right way. Picking ten is probably unnecessary as well.

But my concern, a little bit, and I know it’s kind of a weird thing to say that I’m concerned that we’re almost giving too many tools, there’s a risk of investors who are not in the engine room making assumptions from their own analysis that may not be true. They don’t know what’s going on in terms of potential changes in the manager, and so on, and so forth.

Is there another kind of simpler way to think about these things where people don’t go too overboard in terms of the deep dive? Do you know what I mean?

[00:40:51] Katy

Yeah, I mean, I think my view is, you start simple. When I talk to investors in this space, I say pick two or three, at least two or three managers that you like that you think are a little different, and you can do some simple analysis of their returns. But the real truth is, I do believe CTA’s are one of the few strategies that has nice accretive properties compared to equities.

So, I try to think about the big picture, is that there are few tools out there, in the alternative space, that don’t just have hidden equity beta. And so, you know, having something different is good. So, start simple, don’t overanalyze, don’t get analysis paralysis.

[00:41:36] Niels

Right.

[00:41:37] Katy

And, you know, kind of combine things that are different. And that goes for things like even replication. So, maybe you can have some of that, and you can have a manager that’s going to be faster because you think, I like the crisis alpha and that one’s slow, so let’s combine them. Or you have another manager that’s diversified. But the point is to really sort of not over… Because I think if you get too complex, you have too high expectations of what your results will be and that, as we know, you can never predict. So, every year is different and every style factor that you thought you picked, there’s no magic factor, they’ll change. So, it’s about diversification, which is, you know, kind of the simple answer for me. That’s what I would say.

[00:42:30] Niels

Okay, well, let’s dive into the last paper, the Quantica paper. What I really liked about this paper and all the papers they do, is it’s pretty detailed. And they tackle an issue that I think we’ve talked about over the years on the podcast. And it deals specifically about kind of this battle between managers who stay with very liquid, developed market portfolios and those who, let’s call it 10 years ago, started to move into alternative markets.

Some of them, maybe, claimed that they were trending better so there would be better opportunities. Certainly, for a while, performance was very competitive, without a doubt. Now, we’ve had a couple of years where it’s kind of been the reverse.

So, it’s really an interesting topic and I like the way they are kind of very methodical in terms of the way they try and see it from different angles. So, why don’t you, again, put your teacher hat on here and tell us a little bit about what they were trying to do, and some of their findings, and whether you agree or not agree.

[00:43:48] Katy

I really liked how they set it up. So, they set it up as having sort of a traditional market portfolio with 50 markets, and an alternative market portfolio with 120 across a wide range of sectors, some of which traditionals don’t invest in; things like credit, gas and power, more exotic ags and you know, livestock contracts. So, kind of that like, you know, hodgepodge of like stuff that’s a little bit less easy to access for sort of futures markets.

And what’s fun about this paper is, you know, it goes back to the Man paper. What’s the essential difference here?

One of them has different sectors and different sector compositions. And during different periods in time those different sectors may contribute poorly, or not, to trend following. And what they do show, which I think will really help a lot of clients out there that have both of these things, I think there was a sense that, you know, alt markets, there is a case that correlation, being lower, has benefits over longer time horizons. But given the extreme outperformance in the recent period, it could have maybe gotten people’s expectations a little high.

And then, when your expectations are high and it doesn’t work for the last three years, people start questioning the thesis. And what this paper did, which is so nice, is it just kind of said, hey, you know, there has been an extreme outperformance in some certain sectors. And it happens to be during this time horizon alt markets were in those sectors, for example, gas and power, credit, some things that, that worked really well.

And I really resonate with this narrative because I remember, many years ago, there was big discussions about metals should not be traded in trend following. I’ve looked at the backtest, they’re terrible. And you can’t think that way as a trend follower. You have to think like every asset is an opportunity for trend. I just don’t know which one is going to work.

And so, I think if you imagine any particular asset that maybe never trended sometimes can start to trend. So, like palladium or platinum used to not be trendy. Now they trended a lot recently. Gold has been a huge trend recently, nickel. All these things were things that you would have had questions about 15 years ago, like whoever cares about copper? But now, we all care about copper.

So, they kind of highlighted this particular theme that actually showed up to be a very strong performer in alt markets that’s not in traditional markets. And I think that was helpful because it can kind of explained to a client, who may be thinking about alt markets, is it broken or not (kind of question)?

It’s like, well this was a very favorable environment for these assets. And recently those assets have been very range-bound after the big shock of Covid. So, I think that’s very interesting and it was a very well done analysis because it explained that asset class tilt and why that was important, in relative performance, over the recent periods, which I think will be helpful for those that really think about alt markets.

[00:47:04] Niels

You mentioned that there was this one particular sector, I think it was like 40% or 50% of the outperformance came from gas and power. And, of course, we know that. At least, also, from memory, I think some of it was also related to the Ukraine situation where these markets really took off, and so on, and so forth.

[00:47:22] Katy

Well, there’s an example, they may not have trended at all. And then suddenly you have the situation in the world where that asset trends a lot. And you know, you can’t replicate that. It’s going to be, what’s that asset class going to be now? I don’t know yet. And that’s what trend waits until it finds what’s the next trend.

[00:47:41] Niels

Yeah, I guess my question was how much can we extract from the fact that we’ve only got 10 years worth of data where there really are some alternative market managers around. So, the period we are comparing is quite short in many ways.

Since you are the quant of the two of us, I mean how much does that play into you, in the back of your mind, saying, yeah, I mean it’s great for this period of time, but do I really think that this will be any kind of guide for the future?

[00:48:20] Katy

Well, this is why I like the paper so much because they actually address that issue, right? They said let’s look at Sharpe per market on each individual trend.

Instead of thinking like I know which trends or this period was better, they kind of explained relative adjustments for that period. And then they say, you know, if you think about trend more from a theoretical perspective, you have assets in your trend portfolio. The P&L streams have a certain Sharpe ratio, the Sharpe ratio of those P&L streams, and they plotted it as well, what’s the correlation over time of a trend?

And Alex and I did this even in our trend books. So, you think about each asset as a unit of trend, right? And the P&L streams have a specific Sharpe, you need to have an assumption for that. Then when you build a portfolio of those assets, the correlations matter.

And so, their argument, and this is a mathematical argument, is that on aggregate the more lowly correlated P&L streams that you can add, the little bit higher your Sharpe is. And I think there is definitely some mathematical credence to that argument.

I think where they didn’t clarify this a ton but they do address it, is not all of these assets are exactly the same because they’re not as tradable. For example, trading exotic power contracts may have some more costs associated with it. So that reduces is what seems to be a higher Sharpe, closer to the traditional Sharpe.

But their argument is that they’re similar. So, they don’t argue that alt markets trend better per se, they argue that they trend, but maybe if they trend better, they cost some too. So, the point is they do provide some diversification.

So that’s a mathematical argument that when you combine a lot of things that are different, you can sometimes increase your overall risk adjusted performance. And so, you know, I think that was very well done because yeah, it’s hard, like you said, it’s really hard to statistically prove those.

But there are mathematical foundations to like diversification improves risk-adjusted Sharpe. The question is, you need to figure out what the Sharpe inputs are and the correlation inputs are. So that’s hard.

[00:50:43] Niels

Absolutely. And there’s one other thing. I think they do mention it from memory. I think they do mention it, but I do think it’s relevant and that is of course that because of this relative outperformance that the alternative market managers did for a few years, a lot of money went into those, actually.

And I wonder if the last couple of years’ worth of underperformance, to some extent, I’m not saying all of it, but to some extent is also influenced by the fact that now we actually have maybe too much money tracing these less liquid markets, adding even more so to the transaction costs, etc. Now, I don’t have the answer, but I think it’s something people should be thinking about.

[00:51:25] Katy

I think that would be a great analysis. I would love to see that because crowding, and I’ve always been so excited by crowding, is an interesting factor. It’s a two-edged sword, right? Because on one side trend following still works in some of the most crowded markets we know. But you know, you start to wonder a little bit when you get the things of a little bit more capacity constraints.

And so, I think what one could argue is maybe when you look at a backtest, when less people were in those, you get a more normal Sharpe for those positions. So, it’s just really about diversification and less about like these are better or worse. And that’s kind of the read I get from their paper.

But I think a more detailed analysis of crowding and changes in those markets, even a case study of one market, it could be super interesting. So, I’d love to see someone, you know, actually do that analysis and work so that we could have a better sense. Because that question came in my head as well.

[00:52:24] Niels

Yeah, okay, cool. Anyways, the paper is the Q3 2025 paper from Quantica that you can find on their website. So, definitely go and check it out for sure.

So, we’ve talked a lot about papers. Is there anything that you’ve been thinking about writing about maybe for our next conversation, Katy, anything that has piqued your interest lately that you think, yeah, well, maybe I should do a paper on this.

[00:52:49] Katy

Yeah, I’ve been thinking a lot about macro, and systematic macro, and some interesting connections between the two strategies, both systematic macro and trend. So, I’m hoping to have a paper. We’ll see. I’m working on some stuff with some colleagues, but I’m very interested in macro. That’s something that I’ve been kind of fascinated by and it’s done a lot better during Liberation Day. So, it’s kind of some of the themes like how much of economic trends versus you know, technical trends. So that’s kind of a neat area of thought and research right now.

[00:53:23] Niels

Yeah, I think that’s very relevant and very interesting. Absolutely.

Anything else you want to sort of close out with today before we wrap up our conversation, Katy?

[00:53:34] Katy

No, I think, you know, return dispersion is such a fun topic. Hopefully one day I’ll come back with another paper on style factors and we can talk about that as well. We talked about turbulence. I mean we really got the gamut. We had return dispersion in our topics on return dispersion, so that’s cool.

[00:53:52] Niels

Yeah, definitely a deep dive for sure. Well, as I said, I can’t thank you enough. I really appreciate all the work that you put into to these conversations.

And of course, if the audience listening to this feels the same, why don’t you go to your favorite podcast platform and leave a very nice rating and review for Katy. It really does help more people find the show and be able to listen to these nuggets from her.

Next week I’ll be joined by Rob Carver, so of course another fan favorite. So, if you have any questions for him, feel free to suggest some topics or some direct questions and I’ll try and do my best to make sure we get them discussed. As always, the email address is info@toptradersunplugged.com so that will be for next week.

For today, Katy and I would like to say thank you for listening and we look forward to be back with you next week. And until next time, as usual, take care of yourself and take care of each other.

[00:54:53] Ending

Thanks for listening to the Systematic Investor podcast series. If you enjoy 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.

And 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.