You're about to join Niels Kostrup 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. Welcome or welcome back to this week's edition of the Systematic Investor series with Nick Baltus and I, Nils Castro Larsen, where you each week we take the pulse of the global market through the lens of a rules based investor.
Nick, it's great to be back with you, as always. It's been a little while. Have you been. Yes, correct. It's been a little while. I'm very, very glad to be back. I think last time was this first week of August, very well timed. Just before my break. No, I've been, obviously, I enjoyed the summer.
We got back to Greece with family and then beginning of September when schools start again, here we are so kind of planning now for the year end, a bit of travel, some good projects at work, and managing a four month old alongside with my six year old daughter is in itself, I guess, the new way of living in September. Absolutely. Yeah. My travel season started as well, unfortunately, also started with me picking up some kind of bug on the way back on my latest trip.
So if I'm coughing a little bit along the way, we'll try to mute that as much as possible. But be warned, we obviously have quite a few topics lined up, thanks to you, and as always, they're great. But as we also do every time we speak, I'm really interested in just sort of what you are picking up on your radar at the moment, not related to the topics necessarily we're speaking about, but just sort of what you find interesting.
I've got some completely random stuff that I wanted to throw at you, but what are you sort of finding interesting at the moment? I mean, from a business standpoint, the one thing that typically keeps us on our toes, and now we also see another transition, is this kind of risk aversion that grows or falls as time goes by.
And I think for the last year and a half, maybe it has been a period whereby I, the vast majority of clients and investors would look into carry opportunities, maybe with some sort of short lived breaks. I would call March last year as a good example. But suffice it to say that post August we are seeing more and more and more of this defensive mindset coming back. There is much less appetite, at least in the past one month, for Kerry trades. So in this regard, it is something that keeps us busy.
And for almost like two years it wasn't really the key point. So that is something that has drawn my attention, I would say, quite substantially in the last couple of weeks, maybe. Yeah, I mean, it's quite relevant for our conversation, I guess, people interested in something that's a little bit more defensive. For sure. I found some random stuff I just wanted to mention that caught my attention.
The first one is kind of related to the fact that in the month of September, despite, you could say, certainly growing uncertainty, conflict, whatever, equity markets seems to be fairly immune at this point. And I did notice that the latest valuation for OpenAI because obviously that's the other big theme at the moment, doubled from the spring to now, 157 billion for a company that doesn't make much of a profit, as far as I'm aware. I didn't notice that.
That's more or less the same capitalization as the firm that you work for. And Uber and at and T is in that range as well. So I thought that was kind of interesting. The other thing that keeps us busy generally, is this idea of whether inflation has been conquered or not. Clearly, central banks have declared victory recently by starting to lower rates, including in the US.
But there's a couple of things that actually did show up this week as something that could spark a little bit of inflation, in my opinion. One is that I think last night, so overdose overnight, the dock workers that have been on strike for a few days now, they've agreed a 62% wage increase over six years. That's pretty decent. It might inspire other people to go on strike and try and get a better deal.
Now, clearly this is good news for a lot of people, especially those who love bananas, because apparently 75% of all bananas into the US comes through the east coast ports. So thats obviously the good news. Now, another strike that is somewhat in the news is of course, the strike at Boeing. I think last time I heard they had rejected a 30% pay rise over four years. So clearly these pay rises are meaningful.
And I just want to just put it out there as part of the conversation about inflation, part of the conversation about the dynamics and the changes we're seeing in the labor markets where historically, at least for many decades, employers were at the wheel determining terms. And now it seems that is changing.
Another thing that could be inflationary is the fact that this morning, just a couple of hours ago, we're recording on Friday, the European Union voted to impose tariffs as high as 45% on EV's from China. That's certainly also going to. Hopefully not hopefully, but it's going to put some price pressure, I guess, on these vehicles. And of course, we've seen some pretty decent move up in energy prices this week given what's going on in the Middle east.
So for me, those are kind of interesting things. They all point in same direction, but maybe for other people. The biggest news, Nick, that I picked up yesterday is the fact that apparently next week we're all going to find out who Satoshi Nakamoto is because HBO claim that that's going to come out in a new movie that's being released next week. So to all our bitcoin lovers out there, that's going to be interesting. They're looking forward to it. I am going to watch it if I can.
I don't know if I subscribe to HBO. I have no idea. But anyways, it'll be interesting to see what they come up with. I did forget one thing, by the way, that also is somewhat in the news is of course, this chinese bazooka that has been unleashed, which have certainly had an impact on the markets, especially chinese equity markets, in the last period of time. And I think this also propagated in some commodity markets as well.
We see things like iron ore following a year of falling prices, having a significant rebound in the last month. I would say commodities as a whole had a very strong September, but coming from very, very different regions. Whether that would be wildfires and drought for some of the ags, obviously we have the Nat gas moves and the energy moves in the broader complex here, China impact in some of those metals.
As a topic, I would say September was quite interesting for the commodity complex, both on the beta side as well as the implications for some of the systematic strategies we're running. It was not a great month, by the way, in this regard. For systematic strategies on that front, sure. Good stuff. Well, we just opened the month of October. We just opened the fourth quarter.
It's going to be a quarter where I guess the focus might be the US election in a few weeks, but we should still talk a little bit about what's going on at the moment. My trend barometer at the moment, it finished at 36 yesterday. So not a great reading, a little bit weak. And I think that's being reflected so far in performance in October, even though it's only a few days old. But as of Wednesday night, beta 50 is down 78 basis points in October, still up 3.34% for the year.
SoC gen CTA index down 92 basis points, up 1.56% for the year. Socjun trend down about 1% still. Well, I have it down as 1.17% for the year, but I think it might be up 1.17% for the year. I think it's a typo on my side. And the SG short term traders index down 63 basis points and down ten basis points year to date. Contrasting that with equities. MSCI world down one and a quarter roughly for the month of October, up 16% ish so far this year.
The S and P global developed sovereign bond index down eight basis points, up 2.34% for the year, and the S and P 500 down about 1% for the month of October, but still up a healthy 19.5% so far this year. I mean, you mentioned, just before we jump into the topics, you mentioned a little bit about kind of performance, what youre seeing. Obviously August and September has been interesting.
We can pack it with some of the things that you wanted to start out with, which are reflections on August, but it may also include September. So why dont we just start with that? I know youve been traveling, having some interesting conversations and its kind of nice to have a real time dialogue about what is being talked about. Yes, yes.
So as we're preparing the agenda for today, I thought that reflections following the August moves for the, I guess for the trend world is an interesting topic, not because we're going to go through and see, okay, what happened in August and so on and so forth, more so because I'm getting interesting discussions with regards to the impact of speed or the impact of equity defensiveness, or maybe lessons learned, if there are any lessons that we learned out of it and having gone through that, do we
have to change anything in the models? We've run some analysis and there are some interesting results that I wanted to more share with you, or maybe we cross check our views. I would say that August was very special in the sense that going through the downturn, and obviously with a big question mark whether that downturn would continue, we end up eventually realizing one of the fastest v shapes in equity markets.
Eventually August was a positive month for equities, even if during the month, for example, Nikkei had one of the largest negative returns in history.
By the way, one of the topics that has been discussed that maybe not as much in the context of trend following, we saw the keri trade unwind and some of the currencies that year to date have been delivering positive contribution, be it japanese yen or China or Indonesia currencies obviously turning the other way around and starting appreciating quite substantially in the middle of the month. Let me start from a couple of points.
I should start firstly by saying that the underperformance that came from the reversion in equity markets primarily, and then some of the currencies, is something that, at least in my experiences, did not surprise anyone in the sense that it does exactly what it says on the team. You have been piling on historical trends. Year to date, equities have been rallying. There's a great performance that came up until end of July, let's call it.
And then the reversion was literally the, I guess, the mirroring effect of being exposed to equities. Maybe what was a bit more, I guess, interesting year to date was that no other asset class was actually trading as much as equities. And therefore some dynamic risk scaling and maybe some conviction in favor of equities overweight the asset class. And therefore, after the fact, you had a bigger, in relative terms, turning point.
So I think that was one interesting conversation that we had with Kleiss. Is it actually worth allocating more to trendy asset classes or not? Can I? Yeah, please. That's the first point. That's the first point, exactly. So on that point, what I would say is, I agree with you that from a trending behavior, you could say equities has been a great sector. But here's where some of the other moving parts in a trend following model comes in.
Because what we did see starting in, say, q two, was of course, the reversal of the short fixed income trade. And for those who use correlations as part of their risk management, that would actually lead to the equity exposure being reversed or reduced over the summer, because suddenly you have two positively correlated asset classes being long. So that is one thing that I just want to throw in there, and that's absolutely fair. I'm actually looking at some stats.
You're absolutely right around maybe may, coming into June, we saw some relative reduction in the risk in equities, primarily coming from the rates move, which I have to say, however, I know getting closer to August and the reversion of those rates moves, you ended up reducing the rates exposure, and then just before the reversion in equities, you ended up kind of allocating a bit more to equities. I guess the important point goes as follows.
The reversion, had you been quick, would have been at your benefit if the downturn was contagious. So I think there are two important points assessing the downturn in itself, but then assessing the month as a whole and assessing the v shape. So before I go there, let me give you the second interesting data point, at least as far as I'm concerned.
Currencies, including the August reversion, have had so far, I believe in 2024, the worst year of attribution in a trend program in their history, or maybe the last 25 years. I don't think this is something that has been well discussed, at least in our calculations. Just to say again, so I don't confuse anyone, the FX attribution in a trend program year to date has been the most negative it has historically been across the asset classes for any other calendar year.
I don't think this is very well communicated or maybe well understood, but I think it's something worth pointing out, pointing out now to get to more interesting results, at least for me, the V shape gave rise to the question, should you be quicker or should you be slower? So we've run an analysis whereby we look into very short term signals to more longer term signal. Let's call it one or two months of, let's say some rolling window or a half life.
It doesn't really matter up to, let's say twelve months. What happened in August was that being faster or slower would get you better off than being somewhere in the middle. So let me explain that having an average speed, which is relatively, I guess, medium term, three to six months, will get you worse off than being faster than that or slower than that. And obviously the question becomes like, how is that possible? And this is possible because the V shape has two effects.
Either you're too quick to go short, then it reverts, and you're quick to go long. So you participate partly on the downfall as well as on the recovery, or you're too slow, and then you don't really react to it, because by the time it's down, maybe your gross exposure dropped by a bit, but then it recovers.
You're still long, but there is a middle point, like a sweet spot or maybe a bitter spot, that as the market drops close to the trough, you start reverting the exposure to short, and then it rebounds and you don't even recover. Like I tend to use this example whereby you have the physics 101 we did back in school days. There's a frequency at which the soldiers are synchronized on the bridge and the bridge falls.
And I think this is a good example whereby we have this v shaped dynamic, and this was enough for a medium term speed of to be the worst of all the others. And frankly, when we started running the numbers, I was not expecting this result to come out. I thought that being faster would actually get you better off. But the reality is conditional upon the recovery game. Being much slower or much faster will get you in a better place. Now do we change the model? My answer would be absolutely not.
Like I'm sure we can find other examples whereby one would be better, the other one would be worse. Beyond just data mining, I don't see any other value from just seeing that this v shape that happens to be a typical false positive, which however was very aggressive.
I'm not sure what has been your experience and your conversations, but I think to me that was a very very interesting example of neither extreme will get you as badly as a middle point, which is typically what we would follow as a normal speed. Yeah, I think these are definitely talking points at the moment. And there is someone out there interested in the trend space. He doesn't know that. He's inspired me to look into these things.
It's just something that I picked up from conversations where they have convinced themselves that a certain speed is the speed. And of course I beg to differ on that point. In fact, the speed that you're mentioning as being the worst in this situation, it's the speed that they think is the absolute best. So it just shows you that you can't really conclude that. Now I can give you an example where longer term speeds were not great and it was actually Covid.
And what I mean by that was that that downturn was just enough like three or four weeks wherever we as longer term trend followers had positioned ourselves short equities and I cant remember long bonds and whatever, just because that was enough time for our models to react to that. And I completely agree with your finding that in August it was just so quick that our models, yes theyve reduced risk mainly because of volume expansion.
Not so much about the signal making a change for those few days. But of course what determined also orcas to some extent is how you had handled the yen, the Nikkei over the summer months because there were signs of a change in direction, or at least there were signs of those trends being challenged, lets put it that way. They were not as strong as they were earlier in the year. I agree completely with all of that.
Now it actually inspired me to get our research team to run some analysis using different fixed look back periods. So all the way from 20 days is the shortest one to 260 days as the longest one. Then. Okay. And then on top of that we ran how we do it. Actually in the model where we don't have a fixed, we have a cloud of different look back periods and the selection is actually dynamic.
And I mean to no surprise and this is very important what I say now when I talk about short term trend following and in this case we're just using 20 days, not only is it by far the worst look back period, but it doesn't mean that, I mean that short term trading can't be profitable, it's just short term trend following. I dont think its profitable in the long run.
This is a 20 year simulation that we did and obviously this is subject to all the disclaimers that you need to state when we talk about simulated returns of course, but there were some short term managers that probably would have done better than the results I look at. Then you see a big improvement if you triple that 60 days, definitely over. Using this raw trend signal improves a lot. It's actually quite similar to 130 days. Between 60 days and 130 days, not a lot happens after 20 years.
Obviously during that period there will be a difference for sure. But once you got, then you get longer. So you double that again to say 260 days. There is a big, big big improvement by a big factor actually in terms of the simulated returns that it produces. And then as we probably will see in some of the papers as well, once you start being more active dynamic in the selection of parameters, which is how we do it on our side, you can actually make even further improvements is our finding.
So I completely agree with you on these findings. And of course it's a choice what kind of trend following you want to do. I would say almost like paraphrasing or adding to your points that I think the choice of the speed, there's no sweet spot for me. The slower you are, the less reactivity, I guess, or lack of ability of capturing a turning point error you have.
So that's the typical type one error, let's call it, or like type two in this particular case, the quicker you are then you have more false positives. Now striking a balance between the two net of costs is, sorry, gross of costs is a choice that is a pure trade off accounting for costs. The faster you are, the bigger the I guess the impact can be. So there comes a point whereby becoming even faster will get the costs eating up much more of your ability to react.
And therefore you don't even have to go in that unless you have like a very strong cost control system. Now that's the first point. The second point I would put is that you're mentioning Covid and how being reactive is obviously kind of helpful in terms of turning the corner. I would even add to that by saying that looking into August on a month that the market was positive doesn't even appear.
If you do your typical scatter plot with a trend following smile and the convex and so on and so forth, there comes a point beyond which you did not have trend following respond to a downturn because the downturn was so short lived, so it didn't deliver. The convexity would have done so if that was fast and the move was contagious.
Now it's another discussion to be had as to whether the v shape is something we can learn from it, but it's not ultimately a downturn that you would have expected a short or medium or long attempt, trend following to deliver return. The last thing I would say maybe counter to your and my views, static position sizing would have helped because you wouldn't have reduced your equity exposure throughout that v shape. So it would look more like as a beta move more than anything.
I'm not vouching for it by any measure, but I'm just saying that there are some counters and we can always find examples that, you know, one single choice justifies, I guess, a dogma. But I would not suggest that this is a lesson that I know that I learned from it. I'm just putting it out there for the benefit of the conversation.
No, I agree with you on the equity exposure, but we also have to take into account that static position sizing and all the other in the yen and all of those things would have had a negative mining. Now data mining? Yeah, you're data mining a little bit, but that's fine. That's fine. Now the other thing that I was curious about as we did this analysis was, okay, but what about the protection capabilities that people want from these strategies? How is that impacted by a look back period as well?
Just to make a long story short, what I also found was that the longer you go, you actually get more performance during those periods where there is equity crisis. Now, of course, I have to add to this that this is based on crises that are meaningful and long. I dont mean about a week or two week crisis, so to speak. I think what the period has highlighted, and hopefully for investors, is that there are obviously different ways to deal with these things.
Also, if youre long only or if youre an equity investor, but it's just the price we have to pay for delivering the long term compound growth. It is the fact that we will go through these periods. Some of them will be too quick for us to hurt, and some of them will be just long enough for us to be caught. And that's just how it is. That's just the way of the world.
If you start trying to overthink your models and try and remove negative periods, because oh, I don't really like that you end up in a bad spot in the long run, is my experience at least. I completely agree and I think I've been clear. We've discussed it several times. If we really after like a daily move or a two week or two day move or a weekly move, we are not following trends anyway.
By design, this is precisely the space that maybe requires some contractual protection, which by the way, it's even more expensive. I'm not saying it's nothing the right approach, then there are ways that we can combine option based strategies, buying some optionality with trend following. I completely agree with you. It is in those long term sustained drawdowns that you get the outperformance. It's not in a short lived downturn that eventually ends up reshaping.
And I think testament to hopefully the amount of content and industry reports and education that people like you are actually offering to asset owners post August. We've seen none of our, I guess, business activities or if you like, discussions leading to any particular challenge or potential thoughts of unwinding a trend following exposure. It's absolutely clear. It says exactly what it says on the team. If the trend is continuing, it does the job.
These are the scenarios that you would have to be exposed to for the long term benefits, right? Being defensive as well as long term sharp. Well, this is the thing. And I would actually go further that, okay, we had an event in orchestra, which was really a non event, but the events from the last couple of years, in particular things like 2022, I think it has opened up the discussions.
I certainly feel that there are more people who are now serious and who want to dive deeper into these type of strategies and are willing to put action behind their research, which is obviously great in my opinion. Now I am interested in if you have other things you want to talk about from your recent travels, I know there were a few other points, or you want to jump into the next bigger topic we wanted to talk about. Look, I think segue to the next topic.
Maybe it's this a very basic empirical finding that goes alongside what you said. I just remembered it because you said 2022, and I think it's a nice step to the next topic we're going to discuss. What are the three worst caladayas of s and p since 2000? Can you guess? I'm not putting on the spot, but I'm pretty sure you're going to find them. No, I mean, I would say 2001, maybe, or two members. Two.com, correct. It was two. Okay. Yeah. So we ended zero, eight, and then 20.
Yeah. And then 2022. Correct. So.com, gFC, and the inflation crisis. Yeah. What have been the best two duration years? Duration years, what do you mean? As in fixed income. Buying bonds are. Buying bonds would have been the best two? Oh, I can't remember. But it wasn't. 2022 was not one of them, so. Exactly. Right. So.com and GFC, best years for bonds. Okay. Right. 2022, the worst year for bonds. What are the three best years for CTA's?
Ah. 2022 would have been one of them, 2014 would have been another. I would have thought in 2008. Exactly. And depending on variation, actually 2020, 2002 is one of the three, and 2014 is actually the fourth. So you're absolutely right. It's like those three plus one. So the point I think we're going to get to, which is more about equity bonds and so on and so forth, at least to me, it's an interesting path towards equities. Do the job, fixed income has done the job.
But what do we do now going forward, with all those dynamics in 2022, trend following and all that lot? I think. I think it's. I think that's now the time to move in that direction. Now. Mann wrote a paper. Edward Hoyle at Mann wrote a paper. I thought it was a, I mean, it was a very good reminder, actually, for me, when I read the headline. The headline is, although the name of the paper is risky business, why the smart money forecast risk, not returns.
And it goes very nicely with what I certainly often have said to people I talk to, probably on the podcast as well, and that is CTA's. Much to people's surprise. We are first and foremost risk managers, and we have no idea what returns we're going to get from the markets, but we have some idea of what risks we take, and this is exactly what they're getting at. Do you want to dive into this a little bit?
So I guess the gist of this work relates to whether we can forecast returns, whether we can forecast risk, and what does that mean? And I'm talking about the short term, and what does that mean for us? Allocation, as in combinations of equity and bonds and even maybe like commodities down the line. So they start by effectively looking into realization of returns and volatilities for the stock market, for the last hundred years.
And by no surprise, they do find that there are those times whereby risk is elevated, as in volatility, but then subsequent to volatility spikes, volatility remains high, whereas there is no association between today's realization of volatility and tomorrow's returns. This seems to go against academic principles that basically suggest if you experience risky moments, you should anticipate positive returns going forward. It's not necessarily against it.
I think what is missing here is the context of the horizon. I think they just put it nicely here. They say, look, yes, when risk is elevated and I'm in a down market scenario, it is more likely than nothing that in the next, I don't know, a year or two or five, my equity return would be positive, above and beyond cash to, if you like, to satisfy the premise for expected returns rising in a down market, but that doesn't happen next month.
So today's risk has no forecasting ability for tomorrow's return, but it's actually a good predictor of tomorrow's volume. So tomorrow, today's risk is related to tomorrow's risk, but today's risk is unrelated tomorrow's return. And taking that as a given, then they end up saying, look, if you have like a 60 40 portfolio, well, maybe you should think more about how risk is distributed, because in a 60 40 portfolio, it ends up being 90% equity risk and 10% bond risk.
And in down markets, not only is that becoming more risky because volatility is rising, you're not even compensated for it. So the portfolio, as a mix of equities and bonds a, becomes more risky, whether it's the volatility that is going up or the two can actually start correlating or breaking their correlation. I think 2022 was a prime example that both of them correlated highly, both of them falling.
And the point they bring forward, which I think has been well discussed in the industry for years, is that, first of all, you need to account for risk, and that is like volume. So at the minimum scale, your exposure as a function of realized volatility, because you don't have a good predictor of returns. I mean, you said it yourself, when we do trend following, in a way, that's what we do, we scale down when the market is falling.
And if you take it to the other level, you end up shorting the market. But even by just reducing the exposure in a volume spike, you somehow implicitly play a trend. Yeah, this is, of course, one of the big debates we've had over the many years on the podcast. Because obviously with technology as it was back in the seventies and the eighties, that's not how trend followers started out. They had the static position size and some still do.
Of course, what is somewhat interesting actually on that point is that I think it's fair to say that probably most of the bigger managers today, they would certainly embrace dynamic position sizing, I think. I feel certain when I say that. But I also know, of course, of people who don't do it. And it's not that I could say specifically that their returns are vastly different. I don't know about the sharp.
Maybe that it will show up in the sharp a bit, but it is sometimes difficult to tell the difference once you zoom out a little bit between the two. Now, on a month by month basis, okay, maybe you could detect who's doing what, but long term, and maybe that goes to the robustness of the underlying strategy itself and the signal generation itself. In the long run, maybe the difference is not vast. And sometimes, and this opens up so many doors, maybe I should start many directions now.
No, go on, go on, go on. I'll tell you my view. No, but it's a little bit the same idea, this argument about should you trade 500 markets or 50 markets? I mean, it'll be different from time to time. Will one be vastly better than the other? Overtime, I don't know. I'm on the sideline on that. I don't think it will be. And I certainly don't buy into the point that the trading 500 markets is better than 50 markets.
But now I've opened two contentious topics in one sentence, so I need to really be careful now. No, look, I'll say the following. If volatility clusters, which they show that it does, and we have so much evidence that today's volatility is a good for tomorrow volatility. And there is no association between volatility and expected returns in the short term.
Or, and we have experienced for some of the markets this negative association between volatility spikes and subsequent returns, in contrast to what financial economics would suggest, then you're better off scaling the exposure as a function of all because you scale down prior to negative realizations and you scale up prior to positive realizations.
The reason why I feel over the longer term you can get to similar profiles or performance statistics with more static versus dynamic positioning to your point, is because volatility, beyond the fact that it's clustering, it's also mean revering. So over the longer term you have some exposure that you go in and out, in and out, in and out on a dynamic basis, but over the longer term tends to average out to some sort of long term volume.
However, and I think that's the most important point, that as far as I'm concerned, is not the long term performance but the local one, specifically around the turning points, specifically around the times that it matters the most. Like actually you reminded me, now I'm going off tangent for a second. Seven, eight years back, we wrote a nice paper with a good friend and this was about predicting equity market returns.
And there's this vast literature that says, oh, you can use, I don't know, dividend price or annex price or GDP growth. There's a variety of predictors for equity returns. And there were some models in the literature that was suggesting that hey, you need to constrain your model, because if you constrain your model to only predict positive numbers, otherwise you should scrap your prediction. If it's negative, then you outperform in the longer term. And our point was the following.
Equity returns over the longer term are positive by design, purely because they constitute compensation for you taking on growth risk. Now if you constrain a model, however good or bad, that is to only keep the positive returns over the longer term, you would favor better because they happen to be more positive than not.
Like, no, it's almost as if I'm telling you to predict the coin tones, a coin toss, and you tell me look, it's 50 50, but if it's like 70 30, the probabilities by design, if I constrain my model to only give me positive numbers, I'll be close to the 70 rather than the 30. So it might look that those models look better over the longer term.
But my point is, what is the case in a down market scenario that you think you have a positive expected return, you're piling on equities and then you're getting hit by your prediction. Are you actually remaining solvent?
So that was the point we were making, that maybe over the longer term, assessing prediction models and suggesting that a particular model is better than another might have statistically better performance because it captures the 70% of the times that this was positive but not the 30 that it was negative, but at the time it was negative. Leaving that through and managing money on behalf of that model, you might not even live up to the longer term to suggest that your model was a good one.
And I think that's what happens here as well. And there are those choices that matter in the micro scale, possibly not in the macro scale, but it's the micro scale that our money ultimately is managed. And that's, I think, why it's very important. Anyway, I can digress all the way. No, that's fine. Now, we talked about volatility and forecasting that and using this in your models. I think they also bring up correlation forecasting as well. What are your thoughts on that? Actually, when we build.
There's a nuance here. When we built a portfolio of two markets, correlation doesn't really matter. It only matters for leverage. If I have $100, whether I do 60 40 or inverse volume one, inverse volume the other, and I sum them up and I divide by the sum, so I still have $100, then correlation doesn't really matter if they become more or less correlated. It would simply be that I would need to. I mean, it would have no impact whatsoever on the relative holdings between the two.
They will just come from the inverse volume dynamic. In a risk parity world, correlation only matters for the top level leverage if we have a volume target. But if we have a volume target, we don't have $100. We need to level up or level down.
So the point they're making on that is that if you have a volt target, then it's absolutely critical to keep in mind the fact that when equities and bonds suffer like they did in 2022, not only does the risk go up because the volatility has increased, but also because the correlation goes more positive. And you overshoot, I guess, the volume you thought you had.
So the point they're making is that simply by scaling your exposures as a function of volume, it's also important to take into account correlations for better distribution of risk. In a way, I'm a proponent of this approach, as long as correlation is not too noisy and ends up creating more costs than not. But the point of the paper is to start from a 60 40 equity allocation.
A 60 40 portfolio, in notional terms, would have much more, is coming from equities, would have a realized volatility that is very time varying. And this is precisely the consequence of individual volts as well as correlation. Then they say, how about you have a volume target, which in itself allows you to scale the exposures inversely to volume, but also have a top level leverage adjustment to hit that volume. So that reduces your volume. The portfolio is better behaved and more balanced.
And then they say, how about you also become active or more frequent in assessing those volatility ratios? Because that eventually performs even better precisely because it's now active. And taking into account the contemporaneous dynamics around volume. So these are the three kind of transitions that go through nominal allocation, long term risk allocations, they call it static and then active risk allocation as a function of the realizations of volts and correlations.
That's basically the whole story. It's not too different to building risk parity portfolios in a way, the way at least I read the papers, it's about risk parity portfolios and being risk prudent, which is something you and I, I think I agree very much. So that's kind of a nice segue, because there's a third paper, or maybe it's the second paper, I can't remember what paper we're on now by longtail alpha, that talks about risk parity and trend following together.
So maybe you can talk a little bit about their findings. Yes. So I guess the starting point is on this one is the end point of the previous one that says, okay, suppose that I have a risk priority portfolio between equities and bonds. Okay, fine. So better risk balance between the two if I account for leverage, as in correlation between the two.
I also have a volt target that is kind of well behaved through time, but I still hold two long exposures into asset classes, that there are times that both of them fail, specifically when the equity one correlation goes positive. So beyond the fact that the volatility can overshoot, if that correlation spikes to positive beyond my projections, I'm also exposed to the fact that both of them can sell off.
So by design, if I hold a and b and a goes down and b goes down at whatever volume or correlation, I'm just going to lose money. So the point is, what can we do in that portfolio? And it's no surprise that the historical negative correlation between equities and bonds allow those portfolios, whether it's 64 or expired, to perform very well. But then the 2022 case comes around and they make the point that, hey, trend following can help you.
And they go into this journey of adding trend following, like a basic trend following example. You can even think about the, you know, the stocks and trend index they use, I believe another index, I can go into the details. There's a Credit Suisse index that we're using anyway, it doesn't really matter. It's a trend following strategy that you can add on top of equities and bonds, and lo and behold, it kind of works, it is helping.
And then they go one step further that say, okay, how about we replace bonds with trend following? And that, to me, brings me back to maybe a couple of years that we're discussing about bond replacement with trend following as opposed to portfolio completion with trend following. What they find no surprises is that adding trend following over the longer term in the place of bonds leads to a relatively worse portfolio, or maybe, let's say similar portfolio to having stocks and bonds.
But we know the story, right? Historically, over the last 2030 years, following fixed income trends and taking negative exposure. If, for example, there was like a short term underperformance in fixed income, you wouldn't be paid for it because the reality is just a long bonds exposure was good enough. Of course, the story changed in 2022, and I think that's where the transition in the paper goes and says, look, why do you want to go short? Positively carrying assets as in bonds.
So then they introduce that flavor and say, maybe we should do a trend follower that doesn't go short assets with positive carry and doesn't go long, assets with negative carry. Did I say that correctly? Let me just rephrase. Do not go long appreciating markets that have negative carry. Do not go short depreciating markets that have positive carry that comes from bonds, right?
No, do not short bonds if you're experiencing an upward term structure because you have to pay the carry and therefore do not trust your, I guess, your trend. And of course, what that brings no surprise is that in that they call it carry optimized trend program, you are not shorting fixed income historically as much as you would with an unconstrained trend. And therefore you don't pay the premium of the negative carry prior to 2022 while still benefiting in 2022.
So that's the enhancement above and beyond just adding or substituting bonds alongside stocks still in the risk parity context, right? I mean, concluding they end up also adding commodities into the mix. So if you add commodities alongside stocks and bonds, historically, yes, you have some diversification benefits, but sharpe ratio drops. This is no surprise because maybe for a decade or two the negative roll yield in those commodities would come at a cost.
But incorporating the curry flavor in the trend following camp allows you also to avoid those commodity positions. So putting everything together, they finish off by saying, if I build disparity portfolios between stocks and bonds, whether I have commodities there or not, and they do bring some benefit and some value, having a carry optimized, trend following strategy is adding much more than just, than just beat asset classes.
But again, it kind of reminds me, maybe the first time we spoke about like maybe the first time we spoke period, was at the time that we had 2022.
And I was making the point that maybe one of the more successful inflation hedges is to go short bonds, you go short equities and you go long commodities, and you find yourself paying the equity risk premium, you find yourself paying the term premium, and you find yourself being exposed to the negative royal yield and trend following at the minimum can allow you to be a bit more dynamic around those. Around those dynamics.
And it's no surprise that in this paper they say, hey, I have stocks and bonds, maybe commodities at the time, that I have stagflation and so on and so forth. I'm benefiting from that. That's the whole story, right? I mean, nice paper, easy to go through. Is it something groundbreaking? I think we know the stuff, right? No, no, exactly. And again, I very quickly glanced through the paper this morning, and I agree. With the carry screen, by the way. In a way, yeah.
Well, yes and no. There's a purity part. There's a purity part. Well, this is the thing, right? This is the thing. Now we have, the mic is yours. Now, we talked about it a little bit earlier today that, sure, you can go back and I don't even remember how long they went. I'll tell you, 98. 98, let's call it like 25 years. I mean, of course, 98 is kind of exactly when this carry regime started, right?
Clearly you could have found a benefit of excluding or finding ways to limit the short positions or short exposure to fixed income. But here's my point. It's just that this is exactly what I mentioned earlier on, trying to be too clever and adding stuff to a trend following model to avoid some of these periods that you don't like. I just think it's a dangerous game because we don't know what the future holds.
And clearly the first 20 years of this century is probably going to be very different when we look back in 20 years. You and I are doing this podcast in year 2044, hopefully, and we're going to look back and say, oh, Nick, do you remember when in 2024 we talked about this study and they had found a way to reduce the negative drag from going short bonds? But boy, in the next 20 years, you wouldn't want to do that because look at where rates are at 25%, whatever.
So all I'm just saying is I like the purity of trend following, not being too clever, not trying to forecast, not trying to overfit. And yes, you pay a price for it, but it makes it more robust in an uncertain, unknowable future in my view, humble view. I remember the days maybe ten years ago, a bit more. 13 maybe when the first few years post GFC, post 2000 920 ten, some of the underperformance started coming through.
There were a number of doubts as to whether trend follow would work if rates go up right. I'm sure you remember those days. Roy Niederhofer was on the show because he wrote a paper in 2014 making that point that longer term trend followers would not be able to make money when reintroduced rates started to go up. We brought him back a couple of years ago.
Alan and I, we talked about this particular paper because I remember it caused a lot of challenges for us in the longer term camp because people believe that narrative. And what Roy said on the podcast, it's somewhere in the episode he said, well, he didn't expect the yields could invert. The funny part, I do remember that we don't know what the future holds.
And in this case, shorting Bonds was one of the most profitable trades for a long time after having been long, bonds being one of the most profitable trades, you know, maybe as long as. You and I, we just discussed it, right? Taking those shorts back in time, you know, was not, was not compensated for. Anyways, let's bring it all home. We've still got another ten minutes or so. Let's bring it home with a paper that is very familiar to yourself because you're one of the co authors of it.
I thought of discussing that part primarily because, by the way, it's on sizing, primarily because I'm getting more and more and more questions these days because I think there's a lot of industry maturity post 2022 as to what is the value out of trend following. And obviously the question then becomes, how much should I allocate to it? The sizing question, once somebody starts looking into it, becomes, I guess, of existential nature because it's very hard to come up with a number.
And we wrote something five years ago, it was November 2019. And that's precisely how much should somebody allocate to trend following? And interestingly, there was another paper from one of the large cdas that came out a couple of months ago on the same topic. And even the approach that they take is very similar to what we had five years ago in that conventional ways of determining sizing do not work. So let me go into the specifics.
Being agnostic about what is your asset allocation and what is the alternative you want to add to it. I guess the modern portfolio theory would suggest that you put those two numbers. In an optimizer, you have some expected returns for asset A, expected returns for asset B, some volatility, some correlations. You build some efficient frontier, and then you have like, no, a portfolio that is maximizing the sharp ratio. That's the tangent portfolio, for example.
Now if we take any SAA, like no equities bonds, call it 60 40, call it risk parity, we've discussed it at length in this past hour. Just have your SAA and then you put next to it. And for people who don't know what. SAA means, strategic allocation. Yeah, so strategic as allocation or policy portfolios for the asset owner community is a guideline as to how their core allocations should play out.
And then typically, depending on the organization, you have an overlay, you might have any liquids bucket. There's obviously the, and that I guess is the alternative complex that goes alongside the strategic as allocation. Of course there are tactical components. That's the TAA that we typically talk about, or I DAA, like dynamic as allocation. In case you need more acronyms, I can continue.
But anyway, long story short, if I have an astral location, which is a blend between fixed income, between equity with some regional tilts depending on the mandate, what more can I get from a trend follower? Assuming that we know that we can get this convexity, this protection in extreme downturns and so on and so forth, how much do you allocate? That's the question. So what is the expected return of trend following to put into a mean variance optimizer?
Bluntly, if you ask me, I'm going to tell you zero, because I have no clue and it shouldn't work in the first place. I think Andrew made a very nice point recently. Said people that don't feel comfortable about trend following is because it goes against their premises and their fundamental understanding of the market, that past returns should not predict future return. But the fact that it does cannot be ignored.
Let's say we use historical Sharpe ratio of trend following doesn't really matter, just pick a number. That number historically as a sharp ratio is more likely than, not even higher than an equity bond portfolio. Let's call it equity bonds is like no, 0.6. We can say, okay, a CTA or a trend follower could be like 0.70.80. .9 the numbers don't really matter. What matters at the end is that any invariance analysis would tell you to do 70% trend following, 30% the remaining of your portfolio.
So we're overlaying an alternative allocation with the strategic allocation it doesn't really make any sense. So conventional ways do not work. Then you can say, you know what, scrap the expected returns and simply focus on a minimum volatility portfolio. How can I minimize my volatility? Well, I can solve for this combination that the downside risk, for example, is minimized by the addition of x units of trend following.
And I'm talking about on a funded basis, we can solve it for unfinished allegations. It does really change the result, even that. And I can quote some of our analysis gives you a 52%. And I'm using here data from 1999 until 2019. As we said, you know, this was work that we did five years ago. So that now basically says you do 50% trend following, another 50% equities and bonds, it's too, too high, right? It's too high as a number to be consumed by an institution. Right?
So then what we suggested back then is that there are two driving forces here, and these are the ones that should be balanced in determining that size. And that's where this basically paper goes. Basically says if you're holding a policy portfolio, a mix of equities and bonds, this is your benchmark. And benchmark has a very important psychological impact in how investors operate. And I mean, not just investors, all of us have a benchmark to beat in our personal lives, in our professional lives.
If you beat the benchmark, you're better. If you, if you miss the benchmark, you're actually underperforming. So anything you do away from it is an active risk decision that unless it's properly rewarded for it, you'd rather not have it in the first place. But any active decision consciously comes with risk. That's the typical active risk that we talk about in single stocks, for instance. Right? So how big should that size be of allocating some dollars into another alternative?
That's the question, right, that we have in here. So the point that we're making here goes as follows. There is no way you can maximize your returns. There is no way you can minimize your downside risk, because both approaches would give you unrealistic numbers.
What you really have to assess is a scenario analysis and value add of various increments of trend following allocation, be it 2%, 5%, 10%, 15%, but then calculate a tracking error, quote, unquote, as to how much you now deviate from your benchmark, because that's now your risk. The more you deviate from it, the better your tail characteristic would look. The better your sharp ratio would look, the better your downside performance will be. The more likely it is that you're going to maintain.
You remain solvent in a way, in a down market, but that comes with a business or a career risk. So where this kind of story concludes is that it's a balance between those two. What are those characteristics you're after in terms of longevity of an investment? And I think the industry paper talks about a one year horizon versus a ten year horizon changes dramatically your perception about CTA's. Maybe in a year, the value add is not as great as it is in a ten year.
So the horizon is very important. The downside, the risk mitigation is very important. Maybe the long term sharpe ratio is the consequence of those two. But any decision becomes an active decision. Where do you draw the line? It's very different to me and you. That's the typical risk aversion in a different way, I guess, in a different concept. Right. And I'm not saying that you're not allocating into a trend follower is a risky decision.
All I'm saying is that it has to be seen as a strategic decision. But sizing, it should not be too small because the impact would be tiny. But the more it becomes, the more conscious an institution should become about a tracking error to what ultimately everyone is assessed against.
And I think it's very hard to go against what we are assessed by maybe, and I think we've discussed it here a couple of times, maybe the strategic ass allocation should change and contain maybe five or 10% or 20% of trend following. And that now becomes a different story, because if that is the benchmark, then we have more room to play with it. And I think this is where the industry is at the moment.
Whether a 60 40 or a risk party combination of asset classes is the right benchmark, and the policy portfolio to be held or managed futures becomes part of that ecosystem and is accounted for as a core component in SAAE, as in strategic, as the location for the lack of a better acronym. So that's the whole story about sizing. I think it's a very, very, very important topic for us owners.
I think it's crucial because it is what's going to move the needle at some point if we do see a change in the way these quote unquote, benchmarks are constructed? Maybe my final question, because you wrote this paper five years ago, as you said, maybe it's too early, but do you see any signs that the strategic asset allocation, quote unquote, the benchmark is slightly changing to maybe allow for a higher allocation, to things like trend following.
So you mean introducing trend following in the strategic allegation or allowing for bigger sizes? Or maybe in a sense saying, well, maybe the 60 40 isn't the best today. Maybe it was for 20 years, right?
But maybe today, four years after interest rates found a low in a new environment, in a de globalizing world, whatever the arguments may be, in an inflationary environment we haven't seen for a while, maybe 60 40 as the starting point is really not where we should be having this conversation from. I mean that because that is, as you say, that will allow more room for, quote unquote, the positive tracking error. Hopefully that things like trend following could introduce.
I would answer by saying that I don't see the reverse. Like I do not see, I do not see any specific, I guess, gravitational power holding now asset owners into the 60 40 premise. There are challenges that are coming to the surface month after month, year after year. These are not the same conversations I was having five years ago.
The fact that this report became again relevant for a good amount of our clients is precisely the consequence of those questions now being asked much more than in the past. I completely agree. I see the same. And that's why I said, look a, we need to rank it again and discuss it and be having one of the large CTA's producing pretty much the same report in some variation of the analysis two months ago. I think it's testament that it is a topic that is heavily discussed.
It's not that, hey, let's talk about that because I think it's interesting. I think it is a topic that most of my current discussions are going about. If I can throw in a little bit of a topic that we can think about, maybe for next time we talk, even though it's probably not something we have much knowledge about, but I feel this conversation is changing simply from the fact that bonds had a big sell off. People realized that there is risk in owning bonds.
They had maybe not seen that for 20 years or so, but that's purely performance driven. When I read some of the news, the reports about the debt levels we are running, continuing to run, and the accumulated debt that we have out there from government side, I mean, that in itself should also, in my opinion, be part of the conversation as to why maybe 40% in bonds is way too risky today compared to what it was 20 years ago.
But that's not where our focus is today, because so far no government, big government, has defaulted on their debt. But when you read some of the I was reading, not that I spent much time on Sopstack, but I did see whatever was on the free level from Richard Werner in terms of the debt levels in the US that he posted this week, and it's absolutely terrifying when you read those numbers. Anyways, it's a discussion.
It's a big discussion for another day, but it should be part of why the 60 40 is maybe not the right measure to start with. No, I agree with you. Look, 60 40 was a carry trade and a yield reduction in a turning point. Now which of those components can still be there? I think that's a big question, right? And I couldn't agree more with you. Well, let's end on a point of agreement. That's always a good place to end. Nick, this was tremendous, as always, so useful, so insightful.
And I'm sure everyone listening today will have picked up a lot of new and important stuff from you. And I hope for people who listen in that you're going to show some appreciation to the work Nick has put into this by going to your favorite podcast platform, leave a rating and review. Tell Nick how great he is. That's always a good way to bring him back. In a few weeks. Another great person will join me next week. That's going to be rich.
So if you have questions for him, probably more hardcore trend following questions I could imagine do send them to me. Infobtradersonplugged.com is where they should go and I will bring them up from Nick and me. Thanks ever so much for listening. We look forward to being back with you next week and until that time, take care of yourself and take care of each other. 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 infooptradersunplugged.com comma 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.