Embracing Uncertainty using Adaptive Models ft. Richard Brennan

[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 Richard Brennan and I, Niels Kaastrup-Larsen, where each week we take the pulse of the global market through the lens of a rules-based investor.

Rich, it is wonderful to be back with you this week. How are things down under? I guess you are, you’re getting close to peak summer temperature while we’re freezing up here, or at least it’s a bit cold.

[00:00:46] Richard

Yes, things are warming up down here, Niels. So yes, we’re getting ready. We’re getting our togs out and our towels ready to go to the beaches and things are getting warm.

I know it’s so cold over there, but yeah, this is the time as we approach Christmas. You know, our traditional Christmas meal is sort of a few cold salads and a bit of cold turkey meat. You know, we don’t sort of go down the traditional path at all. It’s all hot down here.

[00:01:14] Niels

Yeah, that makes sense. It’s odd, you know, in the last week, so about a week ago, I think we had the biggest one-day dump of snow that I can remember in the 20 years I’ve been here. And now, you know, a few days after, it’s like 14 degrees Celsius outside. It’s like, it’s really crazy.

But anyways, we’re not here to talk about the weather, but we are here to talk about something, initially, that is often used when thinking about the weather, which is the radar. So, my question is, of course, as you probably guessed since we last spoke, what’s been on your radar recently?

[00:01:50] Richard

I suppose the thing that’s popped up most recently is this collapse of the ruble and the suspension of the ruble on the trading market. So, you know how I always think in terms of complex adaptive systems. We’re thinking in terms of a war on one front, all of these economic sanctions being applied, and all of these variables sort of all interplaying together probably to cause this collapse of the ruble, but you know, it’s just increasing the level of uncertainty.

This geopolitical conflict I think is something that’s probably going to be… I don’t know about you, Niels, but the way I’m reading the foreign environment, at the moment, is that it’s just escalating. Everything seems to be escalating all over the place.

So, I’ve got my concerns that there’ll be levels of desperation in Russia and what that brings with it. But, you know, I’d be interested to see how the Russian central bank attempts to deal with this current issue they’re facing with the collapse of their dollars. So, what do you think?

[00:02:57] Niels

So, honestly, I don’t follow the ruble very closely. I was aware from the headlines in the news that it had its challenges, let’s put it that way, recently. I think interest rates are back up to what, 20%, 21% or so?

[00:03:10] Richard

That’s right, yep.

[00:03:12] Niels

Yeah. So, I mean the thing is, I don’t know what to say about it, but maybe you can educate me here. So, during all this time, I think the ruble was a currency that could be traded via futures. Is that right? At some point, was there a futures contract for ruble?

[00:03:30] Richard

There used to be a futures contract for rubles.

[00:03:32] Niels

There used to be.

[00:03:34] Richard

But I have a feeling that was suspended. I’m not sure, but yeah, I haven’t traded it myself, personally.

[00:03:42] Niels

I wasn’t sure either, but I guess my point, what I was trying to think about when you mentioned it, was does it really matter, meaning the foreign trade?

Certainly, from a Western point of view, I understand it matters for Russia, but from the Western point of view, if countries are living up to all the sanctions, there really shouldn’t be much trade going on between them and Russia. Now, of course, in reality, there probably is, via certain back channels.

But also, I imagine a lot of the energy that they have been selling to countries like China, India may well have been denominated in other currencies. And if so, a weaker ruble actually gives them more rubles, if you know what I mean.

[00:04:30] Richard

Yeah, I suppose a weaker currency always sort of improves the export side of the equation. But the import side of the equation, if they’re dependent on a lot of those imports, that’s going to lead to significant increases in inflation. And, you know, if their interest rates are already at 22% and where do they go from there? Do they go higher? Yeah, it’s difficult.

But I think this was all stimulated by the blockage of the Gazprombank. I think this was something that about two years ago they threatened to do, but they never did it. But I think the US came out with sanctions recently that hit that Gazprombank. And I think that’s actually, probably what’s caused a catalyst for the decline in the ruble. But yeah, it’s going to Certainly present policy challenges for Russia, how they deal with it.

But, you know, does this mean that they have to take their eyes off Ukraine? You know, I’m potentially hoping it might, but who knows?

[00:05:28] Niels

I’m certainly not the best person to voice an opinion about sort of what goes on behind the curtain, so to speak. But it doesn’t seem to me like the ruble, in itself, is going to have much effect on whether or not they want to continue what they’re doing right now. But at the end of the day, I think everybody’s hoping for that something might change when the US Administration changes.

And what I did notice, again, this is not about global macro, but what I did notice one thing was the Danish foreign minister was interviewed like a week ago and ask very much about, so, on which term should potentially Ukraine accept, you know, a ceasefire, peace? And he very much said, well, you know, obviously we’re supporting their efforts, but it must be on their terms. And like one or, one or two days later, you see in the news that Zelensky comes out and says, well, we know now that we’re never going to get Crim back (the island of Crim). So, it’s interesting how the narrative is changing right now. That’s all I can say. So, I think something is changing and this could be just another leverage point that the west is trying.

[00:06:44] Richard

Well, there was always a bargaining chip. I think it was a Kursk region, or whatever, that region at Ukraine was able to infiltrate into Russia. They were sort of using that as a bargaining chip. So, if it comes to a position of an agreement reached between the two, I think giving back Kursk for the giving back of Crimea was the intention.

But I think that’s why Russia’s probably putting a lot of its focus, with the North Korean troops as well, onto that region to try and reclaim that back before any deal is made by, say, Trump coming into the equation. So, it’ll be interesting to see.

[00:07:20] Niels

Indeed. All right, well, my radar point has nothing to do with that conflict. It has to do with a slightly different conflict, and it’s the conflict between traditional investments and alternative investments.

Because I was just thinking, maybe also based a little bit on some of the meetings I’ve had this week, and that is, in a year where you have such a strong equity sector, really, what is that going to do for the demand of alternatives as we head into 2025?

Now, we know, and people who listen to us know that we, of course, think it should make no difference because it’s a core allocation and it’ probably the best time now to make sure you have these alternatives, specifically trend following, in your portfolio before things get crazy in the traditional space.

But it’s not what’s happening. What’s happening, as far as I can tell, is that you have this kind of lack, or beginning lack of demand. It’s just not… Because we can’t compete. I mean, it’s hard to compete with a 25% up year for most strategies. And that’s pretty much where the S&P is sitting. So, investors of course being a little bit, you know, disillusioned by their alternative investments. So that’s what has been on my thoughts.

[00:08:40] Richard

I think it’s inevitable, Niels, that people always look at the, the near-term, the short-term performance, and they’ll be driven towards that short-term performance; return chasing as I call it. But as you and I know, it’s how you manage uncertainty that I think is the most critical thing for investment. And I think, at least with our models, we’re doing that.

So, we’re offering this protection against uncertainty to a degree which, you know, as far as I can see it, as I’m reading into the tea leaves of what’s happening around the world, I’d prefer to be trading our models irrespective of the returns being delivered by equities at the moment. I like the safety that our models gives us in preparing us for potential uncertainty down the track.

So, I’d always make that trade-off as opposed to the chasing of returns. But investors inevitably look at what is recent performance. You look at Morningstar reports, and all of these reports, they never look at 10-year returns, 20-year returns or whatever. They’re always looking at 1 year, 2 year, 3 year performance. So. the whole industry is just focused on this short-term must have, must have, must have. Which goes against the principles of what a whole process is trying to deliver.

You know, talking about the next 2,000 trades being important, how to manage uncertainty. Long-term performance is probably the best measure of how you can manage risk and uncertainty across multiple regimes.

All of these things sort of go against the principles, the promotional side that this investment industry is focused on, which is short-term returns, performance of the S&P 500 for the last year sort of thing. So, I don’t think we can ever get away from that.

[00:10:25] Niels

You know, we’re coming out of a big event, the US election. Markets are obviously interpreting the outcome very positively from a business, pro-business type agenda. So, equity markets, of course, are having a great time, specifically in the US.

Then you hear about all of the narrative around why, why is this happening? And you see, of course, that there was certainly another pickup when they announced a new first choice for Treasury Secretary. I guess he’s still a hedge fund manager, Scott Bessent, and all of that stuff. And so again, narrative controls where the markets are going right now.

But then I was thinking what if we start asking people we meet with, investors that we meet with, we ask them a very simple question. Would you rather know why the markets are going up or would you rather know when?

And of course, I think, for most people, at least if they want to make money, it’s probably better to know when they’re going to go up. And it’s kind of what we do. Right? It’s kind of what we do. We certainly don’t worry about the why. I’m not saying we can know the when in advance, but we know the when, when we see it, when the data tells us it’s now.

[00:11:49] Richard

Yep.

[00:11:50] Niels

And so, that was something that popped up when I was sort of overwhelmed by all this narrative in the news.

[00:12:00] Richard

We use this phrase now in, in our sort of marketing efforts, Niels.

[00:12:05] Niels

I knew you were already thinking about that. I feel that we need to talk about that after we finish recording. Anyways, let’s turn to this month. So far it has been a fine month, and I talked about this last week, when you have a big event and we know it’s coming like an election or like when we had Brexit, we never know if we, as trend followers, are going to be on the right side or the wrong side of that outcome. We just don’t know.

This time it looks like we’ve been on the right side of that outcome, but it’s been very concentrated, as far as I can tell. It’s equities, and it’s North American equities. Canada, US, seem to be the ones (at least the things that I track) that are doing well. The S&P 500, I think it’s made 52 or thereabouts, new all-time highs this year. So, it is clearly very beneficial for trend followers.

But there’s another area that’s also been very beneficial for diversified trend followers and that’s been the softs. I mean they’ve just been on fire. I think this month cocoa and coffee are up respectively. Well, they’re actually up about 34% each this month.

So clearly, it’s also been moving in that direction. If we look at the whole year, we could probably add one more sector potentially, that would be the grains. So, even though it’s been a challenging year and it’s been a weird year, strong in the beginning, very weak over the summer, maybe now, in last couple of months of the year, might be picking up again.

So, it has given us all the challenges that we’re used to, but it also has really emphasized a lot of the things we’ve been talking about for the past 10 years on the podcast: you know, the benefit of not only being rules-based but also being fully diversified.

So, that’s certainly some of the things that I’ve noticed this month. And, as I said so far, there’s like a few hours left of this trading of the month of November. So, you know, who knows where it’s going to end, but it looks like it’s going to end okay this month.

What’s been your experience? And feel free to go out a little bit further than just November.

[00:14:20] Richard

Look, if I was to list the outliers we’ve had this year, I’d probably include gold in there as well. So, some of the things, you know, bitcoin is another one, coffee, orange juice of course, that’s had a marvelous result this year, so, this year there’ve been outliers around in this year, but I think the performance dispersion…

[00:14:43] Niels

Can interrupt you? And sorry to call you out on this, but I think it’s interesting that you found great opportunities. Do you know what the year-to-date performance is on gold?

[00:14:59] Richard

Yeah, I’m looking at…

[00:15:00] Niels

Oh, sorry, I was looking at month to date. Sorry, I’m interrupting you for the wrong reason. That’s because you’re absolutely right.

Yes, I thought, that’s funny you mentioned gold because it hasn’t moved that much this year. But I was looking at month to date. You’re right, it’s up strong this year, up 21%.

[00:15:18] Richard

One of my focuses as well is on gold in Australian dollars. So, over here the two have had this marvelous marriage together. So, I mentioned to you that I’ve got a farm and I’ve got some gold deposits on it, and it’s interesting, now there’s a lot of activity with some of the companies coming in doing exploration over the farm. As the gold price is lifting over here there’s quite a bit of attention now being placed on gold resources in Australia. So, it’s interesting to see, but you.

[00:15:45] Niels

You need to hire a protection team soon, Rich.

[00:15:46] Richard

Well, I’m thinking, I’m thinking that, Niels. Yes, I might have to do a bit of intense fossicking to try and take the deposit myself before they come in. So, yeah.

But you know, some of the outliers, bitcoin, gold for me in Australian dollars, you know, coffee was marvelous, orange juice, marvelous. And look, the particular strategies we deploy, which are particularly focused on these outliers, have done well this year. But there has been a lot of performance dispersion in the trend following camp. And I’m assuming it’s not that there has been an absence of these outliers. So, I’m assuming it might be the investment universe, or the particular strategies deployed or whatever that’s accounting for this dispersion. But I’d be interested to see the end results at the end of this year, and an assessment of what were the chief reasons for this level of dispersion amongst our group of trend followers.

[00:16:45] Niels

Yeah, no, I agree. All right, let me quickly run through the numbers as of Wednesday evening. And yesterday of course was pretty flat because it was Thanksgiving.

BTOP50 up 2.22 for the month, up 3.34 for the year. SGCTA up 1.50 ish for the month, up 85 basis points for the year. SocGen Trend up about 3% in November, up about 1% for the year. And the Short-Term Traders Index down .25 this month and down about 60 basis points this year.

If we look at the traditional, obviously they’re flying. MSCI World up almost 4% this month, up almost 20% so far this year. In the fixed income space, the S&P US Aggregate Bond index up 63 basis points in November, up almost 3% for the year. And the S&P Total Return, 500 Total Return, up a whopping 5.25 this month, up 26.87% this year.

Very much in line, as I discussed with Cem last week, very much in line with what you’ve seen in terms of election year returns, especially during populist periods. So, something he called very early this year, by the way.

All right, well let’s move on to some topics that, let’s put it this way, they are very diverse in nature this week. So, the first one is actually something we didn’t get to the last time we spoke, as far as I can tell, and it is down your alley.

So, I’m going to be paying close attention, but it has to do with how we manage uncertainty in “chaotic systems”. So, I’m excited to hear.

[00:18:29] Richard

Okay, so look, it probably would have been timely if I was able to do it at our last time we got together, which was about a month ago. That was right on the heels of Hurricane Helena and Milton. Those two hurricanes in the Gulf of Mexico that caused that havoc in the US. But what this is, this particular topic I’d like to focus on is the change in meteorological methods that have occurred since about the 1990s with the advent of what we call ensemble forecasting.

So, ensemble forecasting has now replaced what we call single path deterministic models of forecasting, and that occurred in about the1990s and was particularly used for things like hurricane and cyclone forecasting. So, if you can imagine, when you watch the news these days about hurricanes and cyclones, they usually have this trajectory map which says the likely trajectory of a cyclone. And they’ll have all of these different possible paths that the cyclone. And there’s a different probabilistic outcome for each of these particular paths.

That’s used to basically not only give people advance warning, but also for policymakers to have a level of preparedness dependent on the probabilities for where those trajectories arrive at the land mass. It tells them, with a degree of probability, what is the likelihood of a hurricane or cyclone reaching this point here, or this point here, or this point here, giving a cone of probabilities.

This is as opposed to the traditional way that meteorological forecasting was done prior to the 1990s, was what we call a single path deterministic outcome, where they’d basically use historic data and a range of historical data. They’d basically say, all right, given the nature of this particular event, let’s look at the historical record, look at similar instances, and use that as a basis to plot a single path trajectory with what we think is the most likely trajectory of a hurricane.

So, I think it was in the 1990s when this ensemble forecasting came into play. It was prompted by, I think, in 1987, what we call the Michael Fish moment. He was a British meteorologist working for the BBC, if you remember, and there was a viewer who came onto the channel saying they’d heard that there was a hurricane headed their way to South England. And Michael Fish sort of, you know, breezed it away. He said, if you’re watching, he said, then don’t worry. He was very confident using these single path deterministic models that they were doing. But lo and behold, the hurricane did arrive. It caused, I think, 18 deaths in Southern England and extensive damage. And there was a big hoo ha of the time, and I think that’s probably prompted the adoption of these ensemble models.

So, using ensemble models is a much more efficate way of determining probabilistic scenarios for chaotic events. So basically, we’re dealing with nonlinear systems with these chaotic models.

So, if you could imagine normal weather every day to day, weather is very normal in behavior. We know what happens with a degree of certainty based on historic frequencies of events. But in certain periods of time, weather can become chaotic. And this is particularly associated with these tropical lows or these hurricane developments of hurricanes or cyclones, which cause these chaotic anomalies, which are very unpredictable in nature. And they’re very similar to the tail properties we experience in the financial markets.

Most of the time markets are fairly predictable. They’re regime based. There’s fairly sort of regular regimes; we know summer, winter, spring, autumn. We know the particular weather patterns with a degree of certainty there. But there are times where our weather displays this chaotic property. It’s the same way as the financial markets, at times, displays these chaotic properties.

So, ensemble models, they address these challenges in nonlinear chaotic systems by deploying multiple simulations with slightly varied initial conditions and parameters, which allow for a probabilistic interpretation of the most likely trajectories. So, under chaotic conditions, the combinatorial effect of different variables can have vastly different effects on hurricane trajectories.

And the way to think about this is to think of these many variables like being magnets of different strength located along a future path. So, your location of an iron filing going through that future path, along that trajectory, will be influenced by these magnets in various combinatorial ways. Sometimes the path will be predictable, but at other times the path becomes extremely unpredictable and divergent in nature.

So, to better represent possible future trajectories under these sort of path dependent landscapes (a bit like these magnets being placed along the way that alter the trajectories), we apply many different simulations of future trajectories by slightly varying the initial conditions and then seeing where each possible future trajectory leads us.

We then use what we call a frequentist probabilistic approach, which simply sums up the different possible paths and determines the likelihood of certain outcomes. So, each model in itself provides a single prediction. And then we simply count the proportion of how many models in that ensemble reflect that prediction. So, the probabilities reflect the proportion of models in the ensemble that agree on a particular outcome, if you know what I mean.

So, we can get convergent situations. So, under convergent modelling with ensembles, we find that that many of the ensembles align together. That, therefore, means it’s the most likely outcome. And when you have these convergent trajectories with these hurricane paths and cyclone paths, it allows for focused resource allocation in advance. Basically, because we have these convergent ensembles, we know in advance, with a high degree of likelihood where the hurricane’s going to land. And we can do preparatory things in advance because we know it’s going to focus on that particular area.

But when we get these ensemble models showing extreme divergence, this means that the trajectories are very uncertain in nature, and we get very widely varied trajectories, this therefore displays a very uncertain outcome. However, from a planning perspective, this is also very important because it ensures that we have broader preparations, resources being distributed wider, more widespread readiness.

So, these ensemble models have really changed the landscape for policymakers and preparedness, etc. simply because we’re looking at the small deviations that occur in a trajectory that can have significant ramifications in a chaotic environment.

[00:25:55] Niels

How do we translate that into our world of trend following? Because, of course, we don’t sit down every day and look at all the trajectories per se, and we certainly don’t have any view on the future as such. So, how do we translate that into trend following?

[00:26:16] Richard

So, if you could imagine, if these ensemble models are so effective in looking at chaotic weather systems, you can imagine that we could possibly deploy them in our trend following landscape through ensemble models.

And this is an area that I’m particularly focused on, is that as opposed to attacking a particular trend with a single model, when we develop a trend following model, the constraints of that model basically dictate how price has got to move for us to be able to remain in trend using that particular system. So, the entry parameter, the stop, the trailing stop, et cetera, they basically create the constraints of the system we use to define our trend. And when price falls outside those constraints, the trend ends for that particular model. But when they lie within those constraints, it’s still a valid trend or a valid trajectory.

So, if you could imagine, by applying a diverse ensemble of trend following models in the fat-tailed regions, these chaotic regions, it allows for vastly different possible trajectories by applying a diverse range. If we apply a single model, we’re requiring a degree of precision in that trajectory to stay in the trend. When we’re applying an ensemble of models, it allows this freedom of movement and it allows the fact that we might not be successful in all of our models, but the balance of our models might be successful.

So, it’s more a way of being less prescriptive about applying a particular model to a specified trending price series and less prescriptive and more probabilistic in applying an ensemble of models, short-term, medium-term, long-term models, etc. Because in advance we don’t know what the possible trajectory is going to be. But if we have this diverse array of models, we’ve got a greater likelihood of capturing more of that essence of that behavior, giving freedom to move as opposed to a more prescriptive form of system.

[00:28:25] Niels

I’m wondering, Rich, whether you have any sense of the benefit, or at least the benefit in your experience, that you get from adding more different systems rather than, say, having a few different methodologies and then having lots of different parameters such as look back period. Of course, it reminds us a little bit of the discussion about how many markets to trade.

There’s probably also a point in time where you don’t get much benefit from adding more different types of models. And maybe also there could be maybe a little bit of a lock-up or draw with which type of model you pick depending on what environment we are in. So, how do we approach that from a perspective?

Because also I think it’s very… I often say to investors when I meet with them, I say that (and correct me if you have a different opinion), well, maybe there’s only like five or six different ways of doing trend following, really. I mean, there’s obviously a few more, but predominantly most of it could probably be covered by five or six different ways of doing trend following.

The different methodologies don’t really make a huge difference in terms of where people would get into a trade. Meaning a lot of the big trends, you could almost see them with a naked eye. So, you know, depending on speed a little bit, of course, but a lot of the models will get in at the same time.

Where I think things are different is probably where people get out, and also how they manage the risk during the trade. Again, we’re looking at two equal portfolios of markets just to make it simple. So, I’d love to maybe bring that into play as well, how you think about that.

[00:30:20] Richard

So, what I have done is I’ve tested a comparison between applying say 10 unique trend following models. When I say unique trend following models
I’m saying I might be using a class of 8 different trend following models, but the 10 models that I’d be using might have parameter variations using an existing model. So, I’d be comparing the use of 10 trading models per market, to the use of 20 trading models per market.

And what I found was that it was regime dependent. In other words, without the presence of outliers, the lower level of diversification was beneficial. It exceeded the performance of the higher level of diversification of trend following systems. But with the presence of outliers, the performance of the 20-level diversification was better than the 10-level.

Now, a couple of reasons I’m thinking that this is occurring is because I think that the use of these diverse models is more beneficial in what I call chaotic regimes. So, where there’s a presence of outliers is, to me, saying that the regime has a greater influence.

I’m actually referring to these outliers as basically being transition events between regimes. So rather than when I’m referring to a regime, I’m talking about something that can be sort of fairly predicted in advance: this regime that’s been occurring for the last three or four years or whatever, and suddenly we get this abrupt transition, we get this massive outlier event, and then over a period of time that outlier will settle down and a new regime will occur for a period of time. So, sort of like a transition between regimes.

So, the way I’m understanding the performance of 20 models versus 10 models… And this is why in the presence of outliers, the 20 models do better. And I’ll tell you why, it’s because we’re actually concentrating further into that outlier. If you could imagine.

I’m only applying 10 models as opposed to 20 models. When I’m applying 20 models, that means that if the outlier persists, then at the end of that period of time I’ve got 20 models deployed on that. So, I’m getting this nonlinear effect of the outlier and also the nonlinear effect of the 20 systems creating this massive windfall. And so that significantly changes my P&L. But it’s only with the presence of outliers.

In the absence of those outliers, the greater levels of diversification, I find, impede performance. And that’s because of the level of whipsaws, the underperformance, et cetera, with the greater number of models.

And if you consider our process where maybe 30% to 40% are winners, that means that in these environments which are free of outliers that we harvest or we benefit from, the vast majority of those trades are going to be losers. And so, the less systems we’re getting involved in, the less participation in that, the better it is for us.

So, that’s how I’m not only seeing that in the system diversification, I’m also seeing that often our debates over how many levels of diversification. Because, you know, I clearly see in fairly predictable regimes, lower levels of diversification are better than higher levels of diversification. But when we get into fairly chaotic regimes, the higher levels of diversification are better. You get more idiosyncratic outliers in all different areas of the distribution in these more chaotic areas.

So, you know, I think this is why it’s probably a debate we’re never actually going to be able to settle, Niels, because I do think it is influenced by the regime you’re in. And I think certain models will do better in certain environments, certain models will do better in other environments, if you know what I mean.

[00:34:10] Niels

Yeah, it is a little bit counterintuitive because if I understand you correctly, what you’re saying is that actually having more models can give you more concentration, in a sense.

And it’s funny because I remember Rob brought up an article by one of his former colleagues at Man who had written that particular post about markets, that people think that by trading more markets you actually get more diversification. But actually, his view was that it could actually be the opposite.

You know what’s interesting also, since I’ve been traveling this week, I spent a little bit of time listening to podcasts, and I heard a recent interview with Stan Druckenmiller, who’s always interesting to listen to. And one of the things that he said was sort of one of the key reasons for his success and that was this willingness to take concentrated bets. I mean, that’s just how he explained one of the key reasons for his success.

Yet at the same time, I think a lot of investors get scared when they see that our portfolios might converge to a single theme, maybe expressed in a few different sectors, but still. And it is that willingness to embrace that volatility, to embrace that concentration, that also pays off in the long run in terms of delivering above average returns, I think.

[00:35:43] Richard

There was an expression I heard the other day, Niels, where it takes guts to be a greedy pig. It’s not a very nice expression, but that’s how I feel with these ensemble models. When you’re progressively concentrating your Risk in one outlier with more and more models getting sort of invested in it. But the good thing is it’s not a hindsight decision. The signals are being generated, it’s saying for your models to get invested in that trend, and as the trend matures there are more and more signals, so it still is incredibly rules based.

But I think this is our way to support Buffett’s view that you do need concentrated risk on your big winners. And I think this is how we do it in our trend following landscape. Even though we talk about diversification, to me, diversification of markets, as opposed to a position in all of those markets, it’s more a watch list where I’m waiting for these outliers to emerge.

When they do emerge, I might invest a small allocation into that and then as it progresses and endures more, and more, and more. So, it’s using diversification to get wide coverage, but not necessarily participating in all of them and only participating when your signals are activating.

[00:37:02] Niels

Yeah, that makes sense. All right, well the next topic is maybe a little bit controversial if I may say so, and so I can’t wait to hear your thoughts on this. We’re going to be talking about kind of the hidden risks in financial markets. And so, let’s dig into it.

[00:37:25] Richard

Okay, Niels, so what this is about, it’s about during periods of apparent market stability, risks often accumulate unnoticed beneath the surface of complex adaptive systems, eventually erupting in what we call these disruptive events. So, traditional risk management models such as value at risk, Sharpe ratios, even position sizing algorithms such as, what is it, the Kelly criterion, et cetera, they’re using the past or the history to basically give a risk allocation to the process going forward. And I feel that that can foster a false sense of security which fails to capture the full scope of vulnerabilities in a portfolio.

So, to effectively manage risk, I think we need to move beyond these backward-looking historical measures and adopt frameworks that emphasize resilience and adaptability. So, true risk is notoriously difficult to quantify, and it lies not in what has already occurred, but in the range of possibilities that could unfold.

So, real world risk is less about predictability and more about the unpredictable. And I think risk needs to be seen as vulnerability, not probability.

So most standard industry quantitative measures that seek to define risk, they look at it as variance in returns based on historical data. So, what this does by only focusing on historical variations about an expected return, it leaves portfolios exposed to unforeseen scenarios; what could happen rather than what has happened.

So, you know, if we take the 2008 financial crisis as an example, the traditional models assume the stability of these mortgage-backed securities and these overlooked the systemic vulnerabilities caused by their widespread adoption and the interconnected derivatives that were associated with them. And it wasn’t the lightning strike itself; the subprime defaults, but it was a contagion or the systematic correlation that brought the system to its knees.

So, a more comprehensive approach to risk would account for both measurable historically derived factors and the unquantifiable uncertainties that defy traditional modelling. So, these uncertainties represent the unknowns that can derive significant market disruption.

So just let’s have a look at a couple of examples. Let’s look at an avalanche scenario. So, we see snow slowly collecting on a, a slope, a mountain slope. The history of that slope is that, you know, as the snow is connected, we get a lot of data that suggests it’s very stable. But underneath that sort of slow building concentration of snowflake, snowflake, snowflake, we bring the system closer towards a threshold – a point of critical collapse. This is very much like the financial markets.

You know, even though we don’t necessarily see the risk in our historical volatility of our models, we see changing leverage, changing asset concentrations in particular classes. These are all changing the underlying foundations that hold the glue together of the entire financial structure. And this builds these risks.

It’s a bit like the stresses that build up underneath the surface of the earth before we get a big sort of seismic shock, and a big sort of earthquake event. We don’t see these things occurring, they’re very imperceptible, but they’re there. It’s these changing sort of variables that we treat. Because we can’t see it historically in our models, we don’t account for it.

So, these vulnerabilities, they often emerge from the interconnectedness of systems where hidden leverage, excessive concentration and unforeseen feedback loops amplify risks, leading to outsized and sometimes catastrophic impact. So, recognizing and addressing these dynamics is essential for giving truly resilient investment strategies.

So, there is, I believe, an antidote to this fragility found in systems and this is convexity. So, investors who typically optimize for linear average outcomes, they don’t account for the nonlinear nature of these markets.

So, when they’re using standard deviations, Sharpe ratios, et cetera, they’re not taking into account these building, what I call these warehouse risks, that are actually in the system, but we can’t see any evidence of them historically.

We saw that in the track record of some of these fantastic funds that have collapsed, we didn’t see any evidence in the record. Everything historically, using these historical measures, saying there was limited risk there, but what was building there were these unforeseen warehouse risk buildups. So, resilience (this is a term I think is more important), this lies in constructing portfolios with positive convexity – so, strategy is designed to benefit disproportionately from favorable conditions while limiting losses during downturns.

So, I’d just like to throw an analogy in here. It’s using a racetrack. So, without brakes, the safest approach to drive cautiously around that racetrack was to reduce speed. Even on straightaways, you’d reduce your speed so that when you came to the inevitable corner, you were able to take that corner.

But if you’re using robust braking systems, a driver can confidently accelerate capitalizing on those straight sections while safely navigating curves. So, what this convexity requires is a solid braking system and the ability to accelerate when conditions are favorable.

So, this creates this disproportionate approach, this asymmetrical approach to managing risk. It’s saying we’ve got to mitigate our adverse risk with braking systems. But with these strong braking systems, it allows us to go super-fast, concentrate our risk, really leverage, really take the opportunity when the conditions are favorable, provided we’ve got this dual barbell approach being applied to our investment strategy. (I’ll talk about what the barbell approach is shortly.)

So, portfolios equipped with effective downside protection or the brakes, they can therefore pursue these higher upside opportunities without exposing investors to catastrophic losses. So, this is what I’m talking about.

If we’re applying positive convexity in our portfolios, it allows us to create greater uplifting power without sacrificing our adverse volatility. We’re always mitigating our adverse volatility, but it’s giving us lifting power by embracing this idea of positive convexity.

So, the reason why traditional models typically fail under this stable market regime, so they’re under this illlusion that stability holds no risks. So, these periods of prolonged calm give rise to hidden risk. And I’ll give you a couple of reasons why this occurs.

So firstly, they’re using backwards looking metrics, so measures like variance at risk. These underweight the possibility of unprecedented events because they’re relying on historical data. For instance, before 2020 few models incorporated the potential for a global pandemic halting economic activity overnight. But we saw that unfold. These historically backward-looking models didn’t see it in the historical record, and they didn’t account for it.

Another reason for this illusion of stability is what we call volatility compression. So, these stable periods encourage risk taking behaviors as investors extrapolate calm into the future. Because it’s been calm in the past, they assume it’s going to be calm into the future. And it causes them to apply increased leverage, correlations tighten, systemic fragility grows. So, in a way this is likened to building a sand pile, you know, these grains of sand being added, which is building up this systemic risk where a single disruption can trigger a market avalanche.

The third thing is we get this mispriced insurance. So, stability reduces the perceived need for protection, driving down the cost of insurance like instruments, such as options. And ironically, this is when insurance is most valuable, as systemic risks are quietly building underneath this surface.

So, this is why positive convexity adopts what we call this barbell approach. So, research shows us that in the long-term, in long-term compounding, extreme events, both positive and negative, dominate outcomes or portfolio outcomes. So, there was this 40-year study done at the S&P 500. The 10 best performing months accounted for 30% of total compounded growth, while the 10 worst months accounted for 40% drag. The middle 460 months contributed little to overall performance.

So, this insight sort of underscores the importance of what I call hiring a goalkeeper for protection. This is mitigating losses during excessive downturns and fielding more strikers, as a baseball analogy. If you field more strikers, taking the opportunities, you’re capitalizing on a significant upside opportunity. So, you’re looking at this disproportionate application mitigating risk,

So, minimizing your adverse risk exposure is one side of the barbell, but leaving yourself open for unlimited opportunities is the other side of the barbell. So, this barbell strategy, this balances this conservative risk management with this aggressive pursuit of opportunity. And particularly, it’s very effective in managing these hidden risks.

So, I just thought this is important. So, one of the key things here is traditional risk models, they often prioritize precision over preparedness. So, they focus, for instance, on volatility reduction rather than ensuring resilience. This focus, I think, needs to shift towards understanding vulnerabilities and designing portfolios that thrive under uncertainty.

So, you know, when we talk about volatility targeting and all of these methods that are designed to reduce the unfavorable volatility based on historical measures, I think that’s a bit short sighted, and I think that there are better ways to address this by this convex approach to portfolio management. I don’t think we should be taking opportunities where we can, hitting the bat wide, and mitigating our risk. So that was the next topic I thought I’d like to bring to the attention.

[00:48:47] Niels

Yeah, I mean it’s super interesting and, and I mean, if we just narrow it down to our world. So, I think of two regimes or two different frameworks for managing risk. I think about the one you mentioned, the VAR portfolio approach. And then you could say, well, there’s also the market by market using stop loss approach.

I take your point about VAR, I mean, sure. But I am curious as to why probably, and I’m saying this with some degree of uncertainty, why larger managers would have come to the conclusion that probably VAR, in whatever shape or form they use it, makes more sense than having stops.

So, I’m looking at, and obviously through our conversations over the years with all these managers, that is how I see that they have evolved. So, I’m curious as to why that might be if we have all this hidden risk in a VAR model,

[00:50:02] Richard

I’ll go from the sublime to the ridiculous. Okay, so, why is VAR used as such an important model? I think this is because it can be quantified. So, I think the difficulty of risk is our difficulty in quantifying it. And I think this was a measure that said, hey, we know what we’re doing, we can quantify risk.

So, here’s a measure, VAR, which gives us an ability to reduce our volatility of our overall portfolio. And I think that gave investors a bit of satisfaction that, as risk managers, we knew what we’d doing and probably that’s why there is a large acceptance of it.

Also, you know, I think when we’re talking to multi-managers or people outside our profession of trend following, in trend following we typically say volatility is not risk. However, with other market strategies, I don’t think that they’re sort of confident about that. They use these volatility measures as their best way to mitigate uncertainty.

I think from our trend following perspective, we recognize that these events that we’ve never seen in the historic record can occur. And you often say on these podcasts, prepare for things we’ve never seen in the past. I think our history only provides us with a very small sample data of what possibly could happen. I think there is a far greater potential, in the future, that risks that could emerge that we’ve never seen in that historical record.

But there is another side of me that says the business of investment has probably made these models of VAR, et cetera, very widespread. And that’s because by flattening the volatility, suppressing volatility, it makes it a very appealing product for investors.

And, I personally think that might have been one of the dominant reasons for its acceptance as a model, especially for the large fund managers that were attracting large AUM. The investors were confident that they knew what they were doing. They liked the fact that they didn’t offer these volatile returns. It gave them confidence that the managers knew what they were doing. But I half suspect managers knew better than that, and they probably knew they had exposed portfolios that were vulnerable to situations.

But you know, I’d like to sort of interview the Nobel Prize winners who were appointed to LTCM and say, did they have any knowledge of the potential of markets to deliver that unfavorable result that hit them before it was too late?

[00:52:48] Niels

Yeah. So, I have a variation of that one where again, I don’t know that many funds anymore target volatility. I know it’s a nice term that some of our friends like to do, or talk about. I think they target risk. I actually don’t think they target volatility per se.

If you look at the big managers evolving volatility, rolling 12 months volatility, etc, I don’t think it’s stable. So, I don’t think they target vol. I think they target risk.

Now, I think you are onto something about the business. And here’s just my thought on that. I think where VAR might lend itself quite well to the larger funds is the fact that you only need to do a small adjustment trade every day. Because if you have large AUM, and you suddenly need to get out on a stop, or three stops, whatever, you may have a much bigger market footprint impact than just doing small adjustment trades every day.

So, you know, that’s just how I’m thinking out loud about these things. The truth is probably somewhere in between, but it is interesting nevertheless and it’s super important. I actually think that’s probably where we always get asked when we go out as a manager, so, how are you different? And as I said in the beginning of this conversation, I don’t know that we are so different in identifying trends. We can do it in different ways, but we’re probably getting in more or less at the same place.

I think where we are different and I think what makes our returns show up in a different way is probably how we manage risk, how we get out. I think those are the more differentiating factors. That’s just kind of my thoughts.

[00:54:42] Richard

I agree. This problem, this sort of hindsight, use of backtests, use of historical data to project the future result, I think it’s not unique to our industry. I find this, the development in Brisbane, here in Australia, we get prone to these massive flooding events once every 50 years or something like this.

But lo and behold, council over 10, 20 year period when things haven’t flooded, they’re very open to sort of open up land in flood prone areas for development. And lo and behold, because they haven’t seen it in their short 20 year backtest or whatever, they think it’s very safe. Everyone invests in that property. And lo and behold, you know, it’s as regular as clockwork. We get this massive infrastructure damage associated with these flood events.

It’s prevalent. You know, the insurance industry is continually underestimating these tail events, mispricing these tail events.

When dealing with risk, I think it’s because people like to think that risk can be quantified, and I don’t think it can because I think there are two aspects of risk. There’s that which can be quantified and there’s that which I feel falls outside quantification, which is what I refer to as uncertainty. You cannot predict, you cannot quantify uncertainty. But it’s there, it’s always there, especially in these complex adaptive systems.

And it turns up at the most inopportune times, usually when things are appearing incredibly stable. And you know, and I’m just continually amazed at how the methods that we use to assess risk just seem to be archaic. They’re not recognizing this latent risk, you know. So yeah, it just frustrates me a lot of the time.

[00:56:34.] Niels

Now we have two topics left, but we only have time for one. Do you want to stay in the financial world, or do you want to go completely off topic, so to speak, of our usual topic?

Unless you can turn it back into some kind of thing that really does relate back to our industry. So adaptive market hypothesis or the rising infertility crisis, those are the two things you mentioned to me.

[00:57:04] Richard

I might go for the rising infertility crisis because that’s something we probably haven’t heard. I think a lot of us have heard about the adaptive market hypothesis. That was a great hypothesis. So, you know, let’s venture onto this emerging trend.

So, it was only about two decades ago, I remember I was continually hearing across the social media and the platforms that overpopulation was the biggest concern for this planet. And now when you’re looking to social media, you find that underpopulation is now the biggest crisis facing this planet. And this is because people don’t understand how these systems operate. They have cycles of prosperity, and they have these counterbalancing cycles.

There are these continual cycles and people only tend to think in what they can remember. They’ve considered Post World War II, the population has been building ever since and it’s going to continue on forever.

But we know, just when we apply the simple compounding, what is it, I think it’s called the compounding factor of 70, when you have an annual growth rate of say 3%, to look at the doubling time of that annual growth rate, you go 70 divided by three and it gives you a doubling time.

You realize, when you apply these doubling times to these very conservative average annual growth rates, you see that nothing can have a sustainable average annual growth rate on this finite planet. There must be these cycles where we get periods of boom and periods of bust. It’s inevitable.

So, the biggest concern, at the moment, is this growing concern with this level of infertility. This is worldwide trends. So, over the past several decades, global fertility rates have experienced a dramatic decline, falling from over five children per woman, in about the 1960s, to around 2.3 today. This trend marked by countries like South Korea. So South Korea has got fertility rates now as low as 0.78. And over here in Australia, it’s at 1.5. And this signals a sort of a seismic shift in population dynamics.

So, we were once concerned about these fears of overpopulation. Societies now grapple with this profound economic and social implications of declining birth rates. So, there are a number of potential reasons for this declining fertility across the world. And I think that the World Health Organization said it’s going to be the third biggest disease of the planet coming up. By 2100 97% of countries around the world are going to face severe population shortages.

And when we look at the shift in demographics with the older population, the baby boomers getting older, the younger people replacing them, you see how this is going to cause a lot of problems. But you know, when we’re addressing this problem, we first need to understand its drivers. And this stems from a variety of different economic, environmental, and cultural factors.

So, we’ve got economic pressures. So, you know, things like rising housing costs, education expenses, healthcare burdens, these all discourage larger families. So, South Korea, for example, has staggering house prices, this has driven its fertility rate to record lows as couples are now sort of prioritizing financial stability over having children.

We’ve got environmental challenges. Microplastics and pollution are not just environmental crises, they’re public health crises. So, the microplastics, I think they’re causing sort of endocrine disrupting chemicals which interfere with hormone production. This lowers fertility rates globally. And all of us are getting these microplastics from fast foods to the cosmetics. You know, these defoliant cosmetics a lot of women use, they’re embedded with them, microplastics, I never knew that before.

But you know, all of these abrasive cleansers and things have microplastics in them to assist the abrasion. And you know, the old chopping board, the plastic chopping board that a lot of us used to use to prepare our salads. You know, this is throwing microplastics into our body and this potentially is leading to reductions, for men, for their sperm efficacy, etc.

We’ve also got the obesity epidemic. This has been a significant impact on fertility rates. So, it causes these hormonal imbalances, increasing pregnancy related complications. This trend, this narrows a window for family planning in many societies.

We’ve got cultural shifts. So societal values now are emphasizing career advancement, personal fulfillment, you know, delayed family planning. So, in Japan, for example, cultural norms around professional success contribute to consistent low fertility rates of about 1.4.

So, all of these interconnected factors are highlighting the systemic nature of this fertility decline, which is a response to the cumulative pressures that create these feedback loops and naturally curbing these population growths in a self-regulating manner.

So, we often saw the experiments we used to do on lab rats and lab mice where we allowed them to overpopulate in a confined area. And we’d find that there’d be a period of explosive growth, and then they’d stabilize, and then a period of reduction. We’re finding the same thing happening with this sort of changing level of population dynamics.

Now this is something, we should accept that this is a normal, cyclical change that’s going to be prevalent for eons to come, if we continue to exist. So, we’ve really got to plan for this. We’ve got to look at the underlying causes. There are certain causes we can mitigate.

Like we can have strong health programs designed to reduce the levels of obesity, and we could have big programs to reduce the level of microplastics, and all of these things. So, there are lots of things we can do to change this, but I don’t think we should think of it as a crisis.

I think we should accept the fact this is going to occur. We’re going to get these periods of boom and bust. Just as we get into financial markets, we’re going to get these feedback loops occurring. These feedback loops are these amazing things that basically self-regulate the sizes of economies and they occur in very weird ways. So, we get these environmental feedback loops.

So, the microplastic pollution that’s diminishing this fertility, this indirectly reduces population growth and lessens environmental strain. With economic adaption, we’ve got rising costs of living, encourage smaller families, alleviating some pressures on resource and infrastructure.

So, there are all of these sort of feedback loops occurring, you know, beneficial and adverse in all of these different ways. So, these feedback loops suggest that declining fertility is not merely a problem to be solved, but a need to adjust to these systemic imbalances.

So, these adaptions come with significant challenges including, you know, economic stagnation, ageing populations, social instability. So, the modern challenge we face with this is that we need to look at this in terms of, you know, complex adaptive systems. Once again, look at the impacts. How, as a society, do we adapt to these changes? We don’t need to necessarily think we can stop these changes, but we certainly need to be able to adapt to these changes.

So, I just think it’s a fascinating discussion on a new emerging trend. It’s a fresh trend, to me, because I was getting a bit tired of the old overpopulation debate and now we’re in the debate of the lower fertility debate.

[01:05:03] Niels

Sure, I think you’re right in a sense that these are some of these very large cycles that we don’t really notice that much unless we live in a period where we see both, both transitions. I don’t know how long these cycles are, but I completely agree with them.

I’m not going to take a fight with the global warming here. I’m just saying that there are some people who believe that also global warming is a cycle related to sunspots. Whether it’s right or wrong, I don’t know. I’m just saying that it is interesting because it does change a lot of things for us along the way. And, of course, in the very end that is also going to impact financial markets.

[01:05:48] Richard

Yeah, I think what it teaches us though is that rather than adopt static solutions, we’ve always got to adopt adaptive solutions, flexible solutions, recognizing that the situation changes. And I think this is where, you know, most of our planning tends to be stuck in a static land as opposed to adapt land.

[01:06:07] Niels

Yeah, I think that’s a great point. It’s also a good point to end our conversation because it is something we don’t talk enough about. I think, actually, when we talk about trend following systematic trading, we don’t talk so much about the adaptive part of it, even though it’s there. And you obviously make the point about adaptive systems. So it is, yeah, it’s great and different.

So, thanks for making the efforts, Rich, as usual. Next week I will be joined by Graham Robertson from Man Group. He’s back. So, if you have some questions maybe to some of the blog posts he’s written recently or any other questions, feel free to send them to info@toptradersunplugged.com.

I will soon be joined not only by Rich, but by all of the co-hosts for our annual recording which will take place next week. So, although we have each submitted a couple of topics, I am, of course, open to a dramatic ‘not seen by any of us’ topic if someone has a really good one that we should tackle, since we’re all together. I think we’re also going to revisit our outrageous predictions from last year.

All I can say is mine, unfortunately, or maybe fortunately in some ways, did not come through. I only got 50% right, but we’ll see what the new predictions will be for 2025.

Anyways, enough from us. We appreciate your attention. We appreciate all the support we get from you and if you really want to show your appreciation, of course, as always we encourage you to go to iTunes, Spotify or any of the other big podcast platforms and leave a rating and review. Or just send us an email saying what you think about what we do every week. From Rich and me, thanks ever so much for listening in.

We look forward to being back with you next week. And until next time, take care of yourself and take care of each other.

[01:08:18] 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.