Algo Anxiety… Can AI Resolve This? ft. Mark Rzepczynski

[00:00:00] Intro]

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

[00:0(0:25] Niels

Welcome, and welcome back to this week’s edition of the Systematic Investor Series with Mark Rzepczynski and I, Niels Kaastrup-Larsen, where each week we take the pulse of the global market through the lens of a rules-based investor. Mark, it is great to be back with you this week. How are you doing?

Is autumn setting in where you are?

[00:00:41] Mark

Autumn is starting to set in. Trees are starting to just turn a little bit of color on some maple trees, so this is a great time of year to be in New England.

[00:00:51] Niels

I had a walk this morning around eight o’clock and you could really feel the crispness of the air, but it didn’t feel cold, and it was nice and sunny. So, I have to say that’s certainly one of my favorite times of the year.

We’re going to be talking about lots of different things as usual, but as always, I’d love to hear, since we last spoke, kind of what’s been on your radar that you might find really interesting at the moment, which has nothing to do with what we’re going to be talking about later, I imagine.

[00:01:21] Mark

Sure. I took a little vacation in August. So, I was out at the Philmont Scout Ranch in New Mexico, which is always a great area and it’s great to clear your head up in the mountains. One of the areas that I’ve been working on a little bit is thinking about private equity. A friend of mine, Randy Cohen, is a professor at Harvard Business School, and is working on private equity index replication, which is actually a really tough problem.

It’s interesting that we’ve had replication for a lot of hedge fund strategies, including managed futures, and now people are saying it’s like, can we do this for private equity and make something that this, I’ll say, illiquid, can we make it a liquid alternative? And it opens up a Pandora’s box of a lot of quant issues and a lot of replication issues, and it’s an interesting problem.

[00:02:20] Niels

Oh yeah, just the thought of that, I would say, is certainly interesting. It’s also a little bit scary in my mind, because I think, and not specifically about this, but I do think that there is this tendency that financial engineers will come in and take something that is definitely illiquid, and by way of matching, making it liquid, which of course it isn’t. But the perception is that you can do these things, and so we’ll see. It’ll be interesting.

[00:02:50] Mark

I was going to say, tongue-in-cheek for both good and bad, never underestimate the ability of a quant engineer to try to replicate something or try to see if we could make it, in the laboratory, a little bit better.

[00:03:05] Niels

Yes, of course, absolutely. You know, we were talking about this just before we pressed record, and I was saying to you that what kind of was on my mind this week was inspired by something I read just this morning. So, it hasn’t really been on my mind as such, but I found it interesting, and I wanted just to mention it to you and to our listeners, and that was this news that Waymo Robotaxis, apparently, are doing now 100,000 rides a week. That is without a driver.

I have to say, I was kind of shocked by the number. I didn’t expect that it was being used by that many people yet. Of course, this is still the early days. So, I have a feeling that it may go quickly.

At the same time, I think when I think about these robotaxis, it does align and follow to a topic you and I are going to be talking about later, and I notice in your notes that you actually mention these robotaxis as something that is relevant for the discussion, even though it’s about trend following. So, can’t wait to dig into that, but we’re not quite there, Mark. We need to go through a few more topics, which we always do, such as a trend following update.

You know what, so far, I would say this week, overall, has been fairly quiet on the Western front, as they say. I suspect, still, that most of the performance is coming from sectors like fixed income, precious metals.

There’ll be some select markets that are hurting trend followers at the moment, like sugar, which suddenly has gone on a little bit of a tear, so some strong gains against the longer term downtrend. Some of the grains as well have had a bit of an upwards correction, which wouldn’t have been good for trend follows. Coffee has been strong, and that, on the other hand, has been benefiting trend followers, I think, at least most, I suspect, are still long in that.

My own trend barometer, yesterday, finished at 45, so that’s pretty neutral, but looking at the numbers, it’s not looking too bad so far in September.

These numbers are as of Wednesday, but I suspect yesterday was a positive month as well, but as of Wednesday, BTOP50 was up 1.18% in September, up 3.92% for the year, SocGenCTA up about 1%, up 2.42% for the year, the Trend Index up 1.42%, up 2.72%, and the SocGen Short-Term Traders Index up 82 basis points, up 30 basis points so far this year.

Equities are also doing well this month, MSCI World up 77 basis points, up 16.42%, so far this year. S&P Global Developed Sovereign Bond Index up 1% so far in September, up 2.45% for the year, and the S&P 500 Total Return up 1.16%, so far, up almost 20% year-to-date.

Anything, any observations about trend following returns since we last spoke?

[00:06:18] Mark

Well, I think the one thing is that you sent me an article the other day about what trend following could do in the not-so-distant future. I think that the focus is going to be macro trends, macro trends, macro trends. That the macro environment, those trends will drive price trends.

So, we really are going to be focused in on the environment that we live in, whether it’s going to be a slow growth/no growth; what’s the direction and interest rates? That’s going to be the driver of all of the trend performance between now and the end of the year and early half of 2025.

[00:07:00] Niels

Yeah, I agree. I mean, it does feel like, because there’s so much going on, on the geopolitical side and on the geopolitical side, that you would see, at some point, some really sustained moves, although it’s not like it’s been a boring year so far, but it’s just been sort of dominated by a few other things. And of course, it’s also been hurt, to some extent, because of the reversals that led to the carry unwind.

And maybe I should mention, actually, that Alan and I sat down with another host on the show, Kevin Coldiron, who is, of course, one of the authors of the book, The Rise of Carry. That was written five years ago, so we sat down last week to record an episode that’s going to come out on Wednesday to see where the thinking is about carry now, five years on, and after what we saw happening in Japan. It’s a fascinating conversation.

So hopefully, everyone will tune in and see what Kevin feels they’ve learned on this topic in the years that has passed since they wrote the book.

[00:08:09] Mark

And that’s been a really important topic overall, because we do know that a lot of CTAs have mixed trend following with carry. So, because they’re uncorrelated, how do you blend the two of those together to come up with a better strategy?

Then you have to think about, well, what are the periods of risk on, risk off, because that’s what drives that low correlation or negative correlation when we have a risk on environment. It’s going to be very good for carry. When we’re in risk off, that should be very good for trend following.

And then finally, I think the big issue is that, for something that’s relatively simple, there’s a tremendous amount of uncertainty surrounding carry because we never really know how large it is.

So, people have talked of hundreds of billions. Some people have talked about trillions. Some have talked about, okay, is it active carry that are very short-term? Is it the, we’ll call it, mega carry between long-term interest rates or long-term performance across countries? So, I think there’s a lot of misnomers about carry.

[00:09:23] Niels

Indeed. And, actually, you could say that it’s almost something that feeds through society as a whole, not just financial markets. You think about sort of the principle of carry and where, in a sense, and also certainly it ties to trend following, because most investors prefer nothing happens really or nothing happening. And at the same time, they can earn a little bit of money every day. So, we know that’s the challenge that we, as trend followers, have in compelling them to look at something that actually makes money when markets are not static or stable.

What you said about carry also nicely feeds into our first topic today, which is about the Fed. You could say that the Fed and other central banks have really been responsible for the rise of the carry regime by the way they have done their monetary policies for the last few decades and really are at the center of carry.

But anyways, you wanted to talk about the Fed and the expectations and the anxieties and so on and so forth. So, let’s dive into that. Let’s rip off the band-aids.

[00:10:41] Mark

I would say that with the Fed making their announcement on a 50 basis point cut on Wednesday, and here we are at the end of the week, it’s virtually impossible for us not to say something about the Fed. I think that was really amazing, when you take a step back and after you sort of get done reading all of the apologists for why 50 basis points make sense, what was the objective of the Fed for the last more than the last decade? It’s forward guidance. Can we give a clear indication of what our policies will be, what we will do, so that you can be able to then plan accordingly? And if businesses and investors can plan better because they have an idea of what the direction of Fed policy will be, well, then you’re not going to get distortions and you’re not going to get upheavals in the markets.

The day before the Wednesday announcement, it was almost a 50/50 chance between there being a 50 basis point and a 25 basis point cut, which is extremely odd. Usually, the few days before the Fed announcement, we’re about 90%, 95% certain what the Fed is going to do. So, you had, actually, maximum uncertainty even the day before, day of.

Second is that what happened is that you had a lot of the economists, not the market practitioners, who are saying that the cut was only going to be 25 basis points. So, you have the, we’ll say, established economists who were at 25. The market itself was evenly distributed between 25 and 50.

And so, you see that the actual announcement comes up. You say that 50 basis point is a big move and usually is an indication that we’re having some kind of problem: that we’re starting to have to have a major reversal in policy. And lo and behold, there wasn’t really a strong reaction on Wednesday. There was a positive reaction on Thursday in the stock markets around the globe.

But then you look at the dollar which did not set a new low. You look at the 10-year and 30-year bond prices, they actually fell. So here you had a 50 basis point cut and longer term bond investors have said, “We actually think that I want to try to sell bonds. I don’t want to hold bonds”, which is the exact opposite of what you might expect.

And then you look at actual policy, and I sort of half joke, and I’ll say this tongue-in-cheek, whatever happened to the average inflation target? And so, it seems as though that we’re still not at 2%, we’re above, but we sort of said that it’s time to switch. Let’s worry about the real economy.

But then if you look at the GDP now forecast, which is from the Atlanta Fed. So, it’s a Nowcast, and they compared it against Blue Chip forecasts. They’re saying that growth should be, in the next quarter, up around 3%, which is much higher than the Blue Chip forecast. So that Nowcast, taking their most recent data is saying, we’re actually going to have stronger growth than what we expected going forward. Yet the Fed cut 50 basis points.

And finally, yes, we do know that there’s a weakening in the labor market. There’s been a lot of major revision in employment numbers because of the way they count that. But it doesn’t seem as though that we’re in, we’ll call it a difficult period or we’re on the precipice of falling into recession.

So, the whole action of the Fed seems odd. Now, I think it’s important to say like, oh, that’s interesting. And it provides a little bit of area for discussion. But what do systematic managers, how do we sort of perceive or how do we think about this? And that’s the important part for, I think, our podcast. A systematic manager, and I’ll also say quants and trend followers, they’re always interested in what we call post event behavior.

We don’t care about what the Fed may do as a forecast, generally, what we’re only interested in is once you take an action, or there is an event, what happens afterwards? And how can we exploit that? And, in some sense, we can go back and then look at past 50 basis point events and say, here’s what we think should happen. We can say here’s the count. But unfortunately, we don’t have a lot of numbers for that count. It’s not a big sample size.

But you know, the reason why you want to be a trend follower, is that you say, I don’t have a forecast of whether the Fed is going to be at 25 or 50. What I do know is that I can sort of follow what happens post event. And if there’s an adjustment in portfolios driven by investors, I’ll take advantage of that adjustment period post event.

And then we need to look at how that post event plays out across markets. So, what happens to the dollars? What happens to oil? What happens to stock indices? What happens to fixed income? So, I want to be diversified so that I could sort of say that if there are secondary effects across a wide range of markets, I could still be able to exploit those.

[00:16:33] Niels

Yeah, I mean, just for maybe for newer listeners, we obviously don’t want to make it sound like that we actually go in and make some analysis specifically on what moved the market on the day after, the two days after. But I think what you’re saying is that our systems will automatically take this into account because they’re sampling a new data set that includes the day, i.e. the price move, post the event. Now, it will take some time, for sure, for that to manifest in any change in signal. But it’s all part of the process, right?

As you say, we don’t need to worry or speculate about what the Fed will or will not do. But we’re just going to react to whatever the markets, which are, again, the real voting machine, how the market interprets that news.

[00:17:28] Mark

And I think that a lot of times, one thing I try to be careful of, and I was having this discussion the other day, is that I have to separate when I look at a market event into three things. There’s what I think they should have done. So that’s my personal opinion. I could sort of say they’re either wrong or they’re right.

I then have to think about, well, what is actually the perception of the market on whether they think this activity, or this policy change, is wrong or right? But then the third is to say, well, what is my reaction to whatever the event, abstracting from what I think the market forecasts versus what I might think? And just say, what is the market going to tell me in prices and how do I then exploit that?

And either that I can exploit it because there’s going to be continuation of a trend, there’s going to be a reversal of trend, or there’s going to be some kind of mean reversion or divergence.

[00:18:36] Niels

You know what, linking to the fact that the Fed caught and cut and cut substantially, in my mind at least, and also the fact that other central banks have changed their rates recently, most of them to the downside, BOJ, of course, to the upside, I couldn’t help noticing a comment which actually ties back to the speech that two of the central bank chiefs did at Jackson Hole. But we know whatever happened this week was probably in the cards, even when Jackson Hole was being held in August. So, I’d like to just read you a couple of paragraphs from David Dredge, a good friend of the podcast and guest on the podcast in the past.

He does a monthly report. He posts it for free on LinkedIn and it’s definitely recommended to read it. He writes about the speech from Powell at the Jackson Hole.

He summarizes this way. He says, “The gist is, we can all rest assured that they have this price stability (aka inflation) problem, that they never saw coming and certainly had no role in causing, under control and they will shortly be turning their focus to the other leg of the dual mandate, employment… It is curious how their own extreme policies like ZIRP, QE, FAIT, which were explicitly targeted at generating inflation, are never attributed as having a role in the subsequent inflation. And yet their latest rounds of policies are giving all the credit for doing away with something that they had duped as transitory, prior to deciding they better do something about it.

He then goes on to take a little bit of a swipe at the governor of the Bank of England, Bailey. He says, “Governor Bailey’s AI prompter took the instructions, per blaming the failure to achieve their mandate on exogenous events to a new level. Just a quick check on the word count for Covid, pandemic, Ukraine, Russia, shocks, coming in at no less than 50 instances! Just to be clear, it wasn’t their fault! But now, their policies have restored order and we can all rest easy, everything is back on path, and they are in total control.”

And he quotes him, actually Bailey, where Bailey has said in the speech, “Inflation expectations appear to be better anchored, which I put down in good part to the presence of independent central banks with clear mandates and nominal anchors, usually in the form of inflation targets.”

I thought that was brilliant because it’s funny how they forget what role they played in all of this, I have to say.

[00:21:28] Mark

When you think about it, Nassim Taleb always talked, in many of his books, about that you have to have skin in the game. And when you think about it, the central bankers often don’t have skin in the game. They can’t really lose their job because they’re appointed for long periods of time, and they can always talk their way out of different issues, but they don’t have money at risk.

I think the big difference between traders and Fed officials is that we constantly have skin in the game. And our skin in the game is that if we don’t perform, well, then people are going to vote with their feet, take money out of funds, go find somebody else to manage, and then basically ostracize you into yesterday’s news, because they say that we’re not going to follow whatever your views are.

[00:22:25] Niels

You know what, Mark? I’ve not looked this up, but I wonder how many times has the central bank cut by 50 basis points at the same time as the equity markets being at new all-time highs? I think that must be close to a first, if not a first.

[00:22:39] Mark

And that’s really a telling point because, in some sense, if you have the stock market at all-time highs, and we’ll sort of say that the stock market forms forward expectations about the market (albeit sometimes when they change, it could change fairly rapidly). But we’re not sort of saying that the stock market is telling us that we’ve got a problem in the latter half of this year or next year. We’re sort of saying that financial markets, if anything, the volatility has moved down.

We had the carry problem in August, but volatility is not high, uncertainty is not high. So, it seems to be that 50 basis points would be sort of rash. Now, you could sort of say what was the policy that got us to above 5% in the first place, but this seems a little rash.

[00:23:41] Niels

Good stuff. Let’s move on to another topic. I know in the past that we’ve talked about a framework you called VUCA.

And now you came up with another framework, which I’m not entirely sure how to pronounce, capital K, small u, and then a capital U. I have no idea. So, can you help me here?

[00:24:00] Mark

Well, this is the K-u-U framework. I didn’t come up with this. There’s a group out of the Wharton School that came up with this probably about a decade ago.

But I think it’s a variation on my theme of VUCA, but it’s another way to also sort of frame different problems. And when you think of KuU, there’s what is known, what is unknown, and then what is unknowable. And when you think about a lot of what we do, you should always sort of frame this as what do I actually know?

And so, what I know is what I can count. Those would be objective measures. So, say if I count in the past this number of events, I could figure out the probabilities going on.

Then there’s what is unknown. So, what can’t I handicap? So those are the things that I should avoid in some sense, because I don’t have a way to objectively measure.

And then we’ll sort of say there’s the unknowable, which you say like, well, that’s why I have to diversify because if there’s something unknowable. Well, those are the issues that I have to sort of try to make sure I avoid.

And we’ll say that when we look forward, okay, we’ve got an election coming up in the United States. The event is known, but the outcome is unknown. But it’s not unknowable because we will have resolution at some point. So, what we should see, and I think what you’ll sort of see is right around that election date, you’re going to see heightened volatility because that’s when uncertainty is going to be resolved.

We’ll say that revisions in unemployment that we saw were something that was unknowable. We know that revisions occurred. We didn’t expect that it was going to be a revision of a million jobs. But now we have to sort of say, how do we incorporate that?

So, what you’d say is that what is knowable, it could be said, what are the trends? What is unknown is what will the markets do in the future? And what is unknowable is just the fact that sometimes you have to follow prices because there is no way to handicap the future of these unknowable events.

[00:26:27] Niels

Yeah, no, absolutely. We’ve learned something new, Mark, a new framework to add to the library. Can’t wait for the next one.

[00:26:34] Mark

And I think that this is really important even when you talk about the idea of passive investing, index investing, active investing, it sort of says, well, if you think about what is known; can you pick managers that can be able to have better knowledge, which is, you know, are they ignorant or not, do they have strong knowledge about an issue, or what is the amount that’s unknown or unknowable?

If you fall into the camp that there’s a lot more unknowable information and there’s a lot more unknown about the world, then you probably want to be more of a passive investor and you want to be more of an indexer.

[00:27:22] Niels

Or if I can interrupt here, for me, when I heard you talk about it, it actually is not so much about whether it’s passive or active as such, although it plays a role, it’s actually the diversification that it comes down to in my mind. That the one thing that kind of helps, on all fronts, with this is just being diversified. And I mean truly diversified, not a portfolio of a hundred different stocks, but a portfolio like a trend following portfolio of a hundred different markets, essentially.

[00:27:54] Mark

Right. And that’s when you use this KuU framework, what it really tells us is that, okay, we want to increase our positions or tilt on what we think is knowable, and that could be a trend. It could be that you have an information advantage.

But because we say that there are three parts of this framework, and we say the known, we know we’ll take a little bit extra risk yet. But that there’s the unknown, unknowable, and we’ll say that’s the reason why you have to stay diversified. And diversification means across asset classes. It also means going long and short because, you know, just being long only may not be enough because what we say that, for example, the correlation between stocks and bonds is positive. Well, then if you’re long only, that’s not going to help you.

So, when you think about it, systematic managers, trend followers, who focus on both long and short and maximizing the set of markets or asset classes that they diversify over, are being able to best deal with what is unknown and unknowable.

Yeah.

[00:29:15] Niels

Which actually takes us very nicely into your third topic. I think you’ve planned this. Which I think is a really, really important topic.

And you wrote in your headline, Algo Anxiety and Algo Anxiety. It still exists with AI. And so, talk us through these things.

I have some comments and analogies, as we talked about early on in this conversation. So, let’s dive into all of this. This is important stuff.

[00:29:49] Mark

A lot of this was driven by an article that was posted in the FT this week. And they talked about that there’s algo anxiety with AI. And I’ll talk about how they actually measure that in just a second.

But what jumped out at me was something that I wrote about in my blog, 10 years ago, where the thought that there is a lot of algo avoidance and anxiety. And the basic premise of avoidance was that when they tested different people, and they found out, for example, even on weather forecasts, if they’re a human is making a forecast, they say, okay, it makes sense. Let’s check on whether they’re a good forecaster or not. So, we can measure their success of these human forecasts.

Then we have algo driven forecasts. And what you found was that people actually avoided sort of an algo driven forecast, even though it might do better than the human based forecast. So, now let’s move 10 years forward.

What they found is that they looked at AI. And so, they sort of said, well, let’s look at what is the reaction to some investors to some new bit of information. And then we say that new bit of information was driven by a human forecaster. Then we’ll say it was driven by some chat GPT or AI forecast. And then they’ll say, what was the combination? And let’s look at the response time and the size of the response.

What they found is that for those forecasts that were AI generated, investors were a little bit slower at responding to that new bit of information than if it was generated by a human. So, it’s almost as though that I hear what you’re telling me, as an algo, I hear what AI is telling me, but I’m a little hesitant to actually jump in and react as quickly as if somebody told me that same information or made that forecast.

[00:32:12] Niels

You know, it’s funny because I certainly have felt this anxiety for trend following on the basis that people felt that there was kind of no control and it was a computer and so on and so forth, which of course couldn’t be further from the truth, really. The computer is really only there to implement the strategy that’s developed by humans. But I feel, often with investors, that there is kind of an underlying assumption that something sooner or later will go wrong and it’s all going to be the system’s fault, so to speak. So, something happens in markets and the system is caught on the wrong side and it’s being blamed.

But then, when I think about this being a human, and I have no way of proving this, but I kind of feel that if a discretionary manager came and after having lost a lot of money from an event that would be rare, but certainly probable, but rare, then the excuse is often from the investors to maybe justify that it was okay for you to lose that money because there was no way you could have known that. Like if there was some way that the computer should have known it.

I feel there is this little tendency that the humans, we get off more lightly because clearly, we could not have known that the BOJ would hike rates and carry trade would be unwound. Well, of course we knew they’re going to hike rates.

So, the computer is obviously not in the know. I understand that, at least the most trend follows, we don’t allow our computers to read the newspapers and all of that stuff, right? But anyway, that’s kind of how I feel the difference. But it does take us back to this discussion about driverless cars.

How would we feel, even if someone could show us a statistic that these driverless robotaxis have had fewer accidents than taxis that have humans in it, would we still be hesitant to jump into a driverless taxi? I know I would be, for sure. I even heard at some point, Mark, that they were talking about that maybe AI at some point, we don’t even need pilots on the planes anymore because it’s on autopilot for the most part anyways.

And I know for sure that I wouldn’t want to go on a plane without pilots either. So, I don’t know, what do you think about this?

[00:35:06] Mark

So, my daughter lives in Beijing and so she’s taking the driverless taxis. She said that the first time it was a little off-putting, but I think she’s used to it. I haven’t asked how many times that she’s done it.

But it’s amazing that, as humans, we’re willing to forgive other humans. We’re unforgiving with respect to systems and models. So, if let’s say a manager, as you talked about, he fails. Well, okay, let’s hear the reason. He can give some excuses. And we say, okay, don’t do that again.

A model can’t give excuses. It will just say, this is my forecast. Here are the chances of me being successful. Here’s my percentage correct or my success rate. And then it is what it is. It won’t give an excuse. And we love to hear excuses. We love to engage in forgiveness. And we don’t trust models.

So, I think that just from all of my experiences is that people, you’ll go into a meeting, and they say, well, do you follow the model? And you say, yes, you follow the model. And then do you tinker with the model? No, we don’t really tinker with the model. But then at the same time, you say, but if you look like you’re driving and you’re going to hit a brick wall, are you going to change something? And, generally, you’re going to have to say, well, yes, we have a mechanism that someone could have an override if necessary. If there’s this one-off event, we can be able to stop from making a mistake.

Now, surprisingly, especially with trend-following, when you think that there’s this one-off situation then, in reality, following the models, generally, would be better off than actually turning the model off or reversing what the model told you to do.

[00:37:12] Niels

Yeah. I mean, there were actually, and this is public knowledge, there were quite a few managers who stepped in during COVID, during that month of March of 2020, to override the models. They say, as far as I understand, that this was done for risk management purposes, I don’t know if it was worse or better to just have left the models to do the job. Of course, at DUNN we don’t do things like that. We do let the models run. But it doesn’t mean we don’t monitor things, but we have to kind of allow the models to do their job.

I mean, we’ve talked many, many times on the podcast about backtesting. And I think we may talk a little bit about this as well later on today. But the thing is, a backtest is worth very little if you tinker with your model and override it, then you can’t really use a backtest, because how would you have handled certain situations in the past? So, it is interesting.

[00:38:17] Mark

And I think that this goes back to, in general, and the reason why we have these podcasts to begin with, is that I will probably say that the number one flaw with systematic and trend following managers, in general, is that managers are not very good about explaining what they do. And so, the corollary to the study that was mentioned before, they said that if people better understand AI, then they had more confidence.

Surprisingly, they also found that women have more confidence in AI than the human. So, they broke it down into a little bit more detail. But the argument would be is that we probably see the holdback for systematic and trend following is that managers still don’t do a very good job of explaining what they do.

Part of it is to say, well, it’s not easy sometimes. And we’ll sort of say people don’t want to give away all of their secrets. And I’ll also say, on the other hand, there’s the sender of the message, which could do better.

But then there’s the receiver of the message. The investors are not very good about understanding what managers do. I don’t think that they often ask good questions about why a model should do well or not well. And I think fundamentally, I sort of say the key focus is always when will your model do well? And when will it do poorly? And why?

And so, the answer, for example, can’t be that if you’re a trend follower, that you say, well, I do well when there’s trends because that’s tautological. So, you have to be able to say, what are the environmental factors that are necessary or sufficient to be able to drive positive return performance?

That doesn’t mean it happens all the time. But you have to be able to articulate what are these periods when you’re going to do well or not well. And I think that that still requires a lot of fundamental research to be able to say, what are exogenous factors, outside of just price, that might lend itself to better trend following behavior?

[00:40:39] Niels

I get that, Mark. But I will say, and I don’t mean to step on anyone’s toes when I say that, I mean, the concept of trend following, I think is generally well understood, pretty easy to explain. I do feel sometimes, and I kind of understand why they need to go there, but I do feel that sometimes people want to know more than that really is necessary.

If you know what I mean, I mean, it’s almost like, okay, but are you using a 20-day or a 30-day volatility flow, for example? Of course, you can answer the question, but how on earth will that actually help that investor understand anything about the future performance? It won’t. So, I feel sometimes that some of the questions we do get, they may look good in a due diligence report that, oh yeah, but I asked this question, I really went into detail. Look at all the information I got. But how are you going to translate that into something meaningful?

As soon as there’s negative performance, you’re going to get all the usual questions anyway. But I do think that there is a bit of a disconnect in terms of what investors think they need to know and what is actually really relevant for them, so to speak. Maybe that’s also what you alluded to.

But just going back to one thing, just so I don’t forget it, we talked about humans versus algos, right? One thing I do think is a real advantage of the systematic approach (I don’t like the word algo, by the way. It sounds fancy, but we just use rules, right? It’s a rules-based approach) is, of course, the fact that one of the biases that we as humans have (and there are a few of them that obviously are important), one of them is anchoring, meaning that people often expect something that has happened in the past.

I think the strength of a rules-based approach, a systematic discipline rules-based approach where we don’t have any expectations, is the fact that we will go with a market move or a trend for as long as it will take us, even if it’s never done this before. Like, I guess a simple example would just be when interest rates came down to 2% a few years ago, or maybe a long time ago now, people were saying, well, surely 10-year notes are not going to go lower than that.

And lo and behold, they went a lot lower. And in Europe, we even had negative interest rates, something that people had not, I think, expected. So that’s where I think the differences lie.

The advantage of a discretionary trade, of course, is that they can anticipate, which a model can’t do. And if they are very good at that, that has value, of course. And actually, the combination might be a good thing. Who knows?

[00:43:51] Mark

Systematic managers have to do a better job of explaining what they do. At the same time, and I’ll agree with your point, is that the bar that’s set by investors is much higher for systematic managers than the discretionary manager, because there will be, as I said, yes, you will have these discussions where people say, well, what exactly is the method of trend? What’s the look-back period? What have you done this way? How have you tested this? What if you looked at this?

You don’t do that to a discretionary manager where you say, OK, well, tell me exactly what is the process you go to pick your choice of investments every time. And how do you do that? What did you look at next? And what were the inputs? So, I think the bar is high.

So, I think that this is a more nuanced discussion. You’d say, it’s not the specifics, but why do you do what you do? So, for example, if you’re an intermediate trend follower, what are you trying to capture as opposed to short-term or very long-term? What are you trying to capture through diversification? What are the factors that might lead you to have better performance versus poor performance?

And I think that this is, we’ll call it, the up analysis or, as I call it, there’s a difference between performance attribution and contribution. The contribution is saying, well, I made money, this amount of money in the cotton market. I made this money in crude oil.

What we’re talking about is the attribution. So why did you make money in the crude oil market? Why did you make money in the bond market? And what were the conditions that allowed you to do that? And then here’s the model that allowed me to exploit those conditions.

[00:45:52] Niels

There’s another challenge, Mark, in all of this. And that is that even if we do explain well, what we do, and we probably would also explain why we believe that. Usually that would be based on evidence, on facts, on track records, some longer than others.

The other thing I think as managers you are faced with, and maybe this is normal, but it certainly is a real challenge. And that is that those people that we’re explaining it to may have a particular belief or a bias that just doesn’t agree with what we present, even if we have the evidence.

So, some people might think that, well, your trend speeds should be three months. That’s the best. Well, based on what evidence would you say that? Or you have to trade 500 markets. That must be better than trading 100 markets. So, I think it’s a challenge. I think managers, yeah, I think we have a lot of challenges in getting our message across.

Well, we’ve got so many topics. I’m almost going to suggest that you pick a few. We’ve got something about the market efficiency that Alan and I touched on a little bit last week. We have the three Cs. We have Memory of a Bear Market. We have Overreactions and Market Beliefs, I guess. And we have also something actually where I wanted to comment a little bit, which was something about the SG Index and picking managers. Anyway, tell me where you want to go. I’m open.

[00:47:41] Mark

Well, there are three things that I think are most timely right now. And one is this idea of the three Cs. Then I want to talk a little bit about market efficiency, and then we’ll finally end up with market memory, because these are all tied together.

The three Cs that I’m focusing on are: What is market connectedness? What are the correlations? And what’s the potential for a contagion?

And the work that I have been doing with some good friends up the road from me, up in Vienna, is that we’re looking at network design, and thinking of the market as a complex network. By a complex network, you can think of each market as a node on a network, and then it’s connected with other markets.

And so that’s the connectedness. And this could either be through the fact that they’re clustered together. It could be that they’re connectedness because of commonness across the traders. There could be connectedness based on similar events. And the connectedness then leads to correlations.

And then the correlations, because those change over time, are fairly dynamic, can lead to contagion. And what you have to think about when you build portfolios is, always, we spend all of our time thinking about the divergence, or we’ll call it diversification, when in reality we should also think about how are markets connected? How are they clustered? What is the network of markets? And how will they behave at times?

Because the behavior of one market will spill over to another, and that’s an opportunity for exploitation. A perfect example is that we had the Fed announcement. So, we say, OK, we know that’s going to have an impact on the stock market. We expect that it’s going to have an impact on the bond market.

But you also have to think about, well, what is going to be the impact on the dollar? How does the dollar then feed back onto other stock indices? How is the dollar then going to be able to drive or may have a causal relationship with crude oil? How is the dollar also going to have a spillover for commodities?

And so, we always think of the idea that we want to try to avoid connectedness through diversification. But I think that there’s also ways, and I think the really good managers also think about how they can exploit connectedness because one market reaction could then lead to a spillover, after a short delay, to another market.

I think that if you can be able to think about the cross correlations, the cross trends, the relationships across markets where a move in fixed income will spill over into something else, or the move in equity will spill over to something else. That the direction of causality may cause spillovers in other markets. That this is really important if you want to go to the next level of playing diversification, but offensively as opposed to just defensively.

[00:51:10] Niels

But that is some kind of factor analysis, would that be correct?

[00:51:13] Mark

Yeah. Now, there have been people who have looked at factor analysis. There’s also work that has been done, it’s a little bit older but it’s a little bit obscure, on variance decomposition where we could sort of say, how much does market shock, in one market, lead to market reaction in other markets, and how much do they then move back and forth?

So, if you have a shock in stocks, it should tell us something, or we should be able to decompose a shock in stocks on the impact on other markets. And if we can know the direction of causality from one set of markets to another, that that’s really going to help us in how we can build portfolios, or where there is going to be risk, or where there could be potential opportunities.

[00:52:07] Niels

I mean, I can see it within a sector, right? Of course, you can have a common factor that certainly can have an effect. I think as soon as you go outside the sector itself, I think there are relatively few sectors where you can have high confidence in any kind of reaction pattern, frankly.

I mean, of course, you would think that within the financial sectors that there should be some. But as soon as you move into some of the commodities, maybe even with gold, maybe oil does have a stronger link. I’m not sure.

I get your point. I think it is relevant, but on the other hand, I also sometimes think that we can make things too complicated and think that we can kind of have a really strong sense and confidence in how things will play out. But I think the strength of trend following is the fact that we don’t really rely very heavily on knowing how things will play out. Does that make sense?

[00:53:17] Mark

I think that you’re showing a little algo anxiety there, Niels.

Now, I do think that the challenge is that you don’t want to make this overly complex, but at the same time, we do know that markets trade in clusters.

So, we know that there’s a base metal cluster, and that’s probably going to be very much driven by the business cycle. Precious metals is a different cluster than base metals. But then we also find that some of the precious metals also move between the precious metal cluster and the base metal cluster.

The two of them sometimes can go back and forth and understanding how that relationship might exist or how it may play out is important. Another cluster would be that we know there’s a cluster in energy markets. But at the same time, we know that shocks in the energy market will have a spillover effect into the real economy and other sectors. So how do we look at this as a more complex network?

[00:54:29] Niels

Yeah. And for those who don’t want to get down to the 3C analysis, they can always stick with the 4Cs, which are cut, clarity, color, and carrot, according to my wife.

[00:54:41] Mark

There you go. That’s it.

Now, the reason why this is, is that this is all tied to efficiency. I know Alan talked a little bit about Cliff Asness’s view. And I think that I’ll probably take a little bit more nuanced view than he talked about, that is that social media is driving markets to be less efficient and that there’s more gamesmanship. I’d probably take the view that you have to think of efficiency, that it’s fairly dynamic.

There will be periods of time when markets are more efficient, and there are periods are where it’s less efficient. Some markets are more efficient than other markets. And I think that an area that has been understudied, is that Andy Lowe came up with the idea of adaptive market hypothesis.

So, he said that this is a different way of thinking about market efficiency. He wrote a book on it. I think it was very good, but a lot of people haven’t gravitated towards that.

But in some sense, if you’re a trend follower, you really believe in an adaptive market hypothesis, that markets efficiency evolves over time. And because it’s an evolutionary process, then there could be trends in markets, and different markets will vary on their efficiency based on liquidity, how difficult it is to value, whether there are more hedgers or speculators in the market, how many people are actually following the market, what is the economic environment in the business cycle at a given time. And so, I think that we need to sort of think of this as much more dynamic.

And so, I can’t really sort of say that markets are more or less efficient, at any given point in time, other than the fact that this is a dynamic concept.

[00:56:35] Niels

Yeah. Actually, I would say when I interviewed Andrew Lowe a few years ago about this, you know, he basically just said, yeah, I mean, markets are efficient, but not all the time. So, there will be these periods of time where you see these big moves that we call trends, and that’s completely to be expected.

[00:56:56] Mark

And it is surprising that there are good ideas that sometimes, if you want to say that there’s memes in research ideas, some memes take hold of people, others don’t, and this one I think has been underutilized.

[00:57:13] Niels

Yeah, true, true, true, true. Well, next, Mark, you still have a few minutes.

[00:57:18] Mark

Well, the one thing I guess is that it was driven by a comic I saw last month in the summer. And I think it was an old New Yorker comic. So, they have an older gentleman in a suit, you know, leaning back, smoking a cigar. He looked like an old time Wall Streeter. And then they had a bunch of young traders all around him. And they’re saying, can you tell us again what it was like to be in a bear market?

I think that the one thing that scares me is that we have this 50 basis point cut. And then we have sort of like this view that, well, you know, things could be a little bit unsettling. So we have to get ahead of it.

For a lot of people, if you started your Wall Street career at 25, worked for a few years, went to MBA school, came out and said, right now you could be 40 years old and never seen a large bear market. So, you’d have to go back to 2008, you know, in that period.

And so, what happens is that you have a collective memory of people who don’t understand what happens if markets go down in equities. And I think that a perfect example of that was that there are people, after the carry debacle in August, they’re saying the Fed should cut 75 basis points immediately. Okay.

Now, a 20% correction in the stock market, which would be a bear market, you would have a devastation on a lot of people’s wealth. In many measures that wouldn’t really sort of eliminate all the overvaluation in stocks by some measures. And in some sense, it would just take you back about a year and a half worth of returns. So, it wouldn’t really have that large of a devastating impact.

Yet, the Fed and everybody is sort of focused in on that we have to always go higher, that they don’t really have a memory or they don’t accept that there’s a possibility that we could have a large down move in stocks. And this reaction problem is a behavioral problem, but it’s something that systematic managers are better able to address because a lot of times they’ll backtest models over a very long history. Their reaction is to say, well, even if I haven’t seen a bear market, I know what to do if market prices go down.

I’m going to be a seller because that’s what a model is telling you. So, I think that’s really important is that there’s going to be a time when equities are going to start to fall and people say, I don’t know what to do, yet the systematic manager should be able to have a plan in place because his model allow for going both long and short in the marketplace.

[01:00:25] Niels

Absolutely true. The other thing that is important, I think, about this, and that is when these market moves happen, and often what we see when a trend is coming towards the end of it, even though we don’t know exactly where that’s going to be, but often there is a pickup of volatility around these big turning points. And that, in itself, actually often triggers trend follows to reduce their risk.

So, of course, you can have a sharp reversal, there’s nothing we can do about that. But often you do see this. So, I think you’re right that there are risks out there, just from the simple fact that there is very little experience by many institutional investors who may be controlling large amounts of money.

And then, of course, on top of that, you have the whole discussion about just flows from passive investments and what that’s going to do once the markets go down for a sustained period of time, not the usual two or three weeks that we’ve been used to the last few years where every time there’s a crisis, Central Bank seems to be able to fix it in a short period of time. But yeah, I think that is a relevant thing.

Before we leave it today, though, Mark, I really would like to talk a little bit about one of your other topics which is the SG Index and picking managers. You had been thinking about this, I always think about this. So, what are your thoughts?

[01:02:01] Mark

Well, Transtrend did a little paper, and I think it was done a little bit in jest, as we as we talked about, that there’s a 25th anniversary of the SG Index and that, oh, well, just by the index. And yet, you know, in some sense, index replication of just holding a bunch of managers doesn’t sit well with me in the idea that you could be able to pick other managers that you can blend together that might do a better job. And interestingly, if you look at the construction of the SG Index, they look at what is the top managers in terms of AUM, and then they adjust this once a year.

And so, you could have big changes based on, “popularity of indices”. I think that in reality you don’t need the whole complement to what they do with that index. In reality you can pick up a diversified portfolio of CTAs that should be able to outdo the SG Index.

[01:03:17] Niels

Yeah, I mean, I read, actually, over the weekend, a blog post, not specifically related to what we do, but it was related to something very similar. And that was from Michael Green, who of course has written a lot about this problem about passive investing and how that distorts markets and all that, and probably we’re still yet to see the effect of that. But the title resonated with me, because I think the title was, Why Bother Being Better?

And this goes back to this idea that if you’re in our industry, why not buy the index? But of course, we as managers, we’d like to think that we can outperform these indices. And as Harold from Transtrend, I think, I can’t remember the precise wording, but he’s also basically suggesting that investors should try and find managers who strive to be different from the index. And ideally, of course, be better than the index.

And obviously, we have Andrew on the show on a regular basis, and he’s a great proponent of the replication of indices. He’s been instrumental in now creating this new SocGen CTA, or investable CTA index, even though it’s a different part of SocGen, as I understand, that’s producing it. Of course, it’s a giant backtest of a model. People should be aware of that. But what I feel about the topic is that I get the sense that we’re trying to promote something that if you buy the index, there is no risk. It’s the index.

But buying a CTA index is not the same as buying an index of stocks. And the other thing that I don’t think is being discussed enough is that, and I’m not saying this is being promoted anymore as an idea, but I certainly remember initially that buying a replicator index is also there to solve the problem of manager risk, because if you just pick one or two managers, they may not be beating the index. Sure.

So, there’s a manager risk, but it almost makes it sound like there is no manager risk in buying a replicator of the index. I think there’s a huge manager risk. And often these replicators run very small firms, much smaller than the underlying managers. Plus, their models could be flawed. They could break. Who knows?

So those are some of the issues that I struggle with. Let me put it that way. And I feel that we need to be more open and clear about the pros and the cons about these things.

I think managers, I think we’re pretty open about the risks by picking a manager. But I’d like to see more of that openness on the ETF replicator side to be perfectly blunt about it.

[01:06:37] Mark

Well, replicators try to sort of say, I’m going to give you the median return for a given strategy. And that’s fine if you say like, I have no skill at picking managers. I want to have a low-cost alternative, and I’m satisfied with the median.

It could be useful at the same time, if the replication is supposed to get the median, that’s different than an index, where the stock index is… What we find is that most managers underperform their benchmarks. So, a median replication strategy is not a benchmark replication.

So, the benchmark of buying the S&P 500 is saying that I’m just going to hold this package. And I know that this benchmark is usually underperformed relative by the set of peers.

The replication is different because what they’re saying is that I’m not a benchmark. I’m going to try to give you the median return of a bundle of managers because I have enough information to sort of say I can be able to figure out the factors that drive the average manager. So, it is a different product, and it has a different set of investor base.

The investor base is the people that say, I don’t want to have the downside risk of being wrong. I’m willing to accept the median, and I want it as a low-cost alternative.

[01:08:25] Niels

Yeah. Well, I’m sure we’re going to be talking about these things in the future. But for now, Mark, I really do appreciate all the hard work and preparation you did for today’s conversation.

And if, just like me, you appreciate this, please head over to your podcast player of choice and leave a rating and review to show that you appreciate the time that’s being put into creating these conversations.

Next week, Alan will be joined by Cem as I’m offline. So, make sure you go and get your questions ready. Cem’s conversations are always great. And with Alan, I’m sure it’ll be super interesting. You can send the questions as usual to info@toptradersunplugged.com, and I’ll make sure that Alan receives them.

That’s it for today from Mark and me. Thanks so much for listening. We look forward to being back with you next week. Until next time, and by the way, remember to listen to Kevin Coldiron, Alan and me next week. Until next time, take care of yourself and take care of each other.

[01:09:34] 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 you 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.