Trend Setters: The Mechanics of Trend-Following

In this episode we dive into the reasons why trend following strategies work and importantly, when they work. Joining Peter is Otto Van Hemert, Director of Core Strategies at Man AHL, co-author of the award-winning paper Best Strategies for Inflationary Times which highlights trend as one such solution.

Trend Setters:

What makes a trend in asset prices? Which environments offer the best returns for quantitative trend-following strategies?

Introducing Trend Setters, the new multi-part podcast series from Long Story Short. Hosted by Peter van Dooijeweert, Head of Multi-Asset Solutions at Man Solutions, Trend Setters is demystifying the world of trend-following strategies and why they work in volatile markets.

Articles mentioned in this episode:

• Best Strategies for Inflationary Times
• The Best Strategies for the Worst Crises

Recording date: August 2022

Episode Transcript

Note: This transcription was generated using a combination of speech recognition software and human transcribers and may contain errors. As a part of this process, this transcript has also been edited for clarity.

Peter van Dooijeweert:

So, joining us today is Otto van Hemert. He is director of core strategies at Man AHL and co-author of a couple of award-winning papers, The Best Strategies for the Worst Crises and The Best Strategies for Inflationary Times. Otto, it's great to have you.

Otto van Hemert:

Thank you.

Peter van Dooijeweert:

We spoke with Russell in our first episode a lot about an introduction to trend for newbies. And I guess I'd like to do a little bit of a deeper dive on the mechanics of trends. So, maybe let's just start off getting an understanding of how trend has worked over time.

Otto van Hemert:

Sure. Trends often arise when there's a big change in the macro environment. It may start in one corner of financial markets, but then it has ripple effects to other corners of the financial markets. And it just takes markets time to digest that information. There's an initial under-reaction and then trending type behavior in order for full reflection of the news. An example is the global financial crisis. It started around 2007 with some badly underwritten subprime mortgage loans, house prices under pressure, but all in all, initially the problems were underestimated. It was supposed to be contained to the housing markets. Even Ben Bernanke spoke about it in that way. But really in the end there was a full blown crisis. All kind of markets were impacted and good trending in markets arose, and we can benefit from that using trend type models.

Peter van Dooijeweert:

So, I guess going back to the global financial crisis, people didn't see it coming. Nobody predicted it. So, how does trend go about, I don't want to say what predicting, but how does it capture momentum and how does it get positioned into these sort of events?

Otto van Hemert:

In macro markets, it's usually determined in a time serious fashion. You look at the price history of the security you're trading in isolation, not versus the price history of other securities. This is in contrast to single name stock momentum, which is more typically a cross-sectional concept. So, for macro markets like oil or equity index or government bond future, you look at the price history of that security in a time series fashion. You can look at the return over a certain window, one to three months. You can look at the moving average crossover, the average price recently versus the average price over a longer window. And if that crosses, it indicates that prices are going in a particular direction. Or you can even look at more esoteric formulations, like the breakout model where the price goes outside of a certain historical range of the prices. If it breaks out from above, it's a positive trend. If it breaks out from below, it's a negative. So, there are many different formulations that you can consider, but in macro markets mostly in a time series fashion is how they're defined.

Peter van Dooijeweert:

And that makes me curious, because on TV news shows you'll hear moving averages talked about a lot. Is it as simple as just picking a few moving averages or are they really crowded? Is everyone doing this? I mean, I hear 200-day every day when I wake up and it's below or above that. How should I think about that? Is it something you guys do a lot more work around?

Otto van Hemert:

The starting point can be straightforward in that it uses price data and a relatively simple formula applied to the price data, but there are several portfolio construction steps subsequently. How do you combine trading trends on many different markets? How do you size the position after you've established you want to be long or short based on the trendiness of a security? How do you combine all the different formulations? And indeed you can come up with slightly more complicated formulations that make use of, for example, intraday sampled price data, and not just daily price data. So, while the starting point is intuitive and to a degree something you can implement fairly easily yourself with an Excel sheet or so, there's a lot more that comes to it when you actually want to implement something like this.

Peter van Dooijeweert:

And without revealing special sauce, are there ones that the trend industry likes more than others or are their ones that are more profitable or more unique?

Otto van Hemert:

I would say things like breakout signals are a bit less common and therefore perhaps more additive compared to what others are doing. So, indeed, the most simple implementation is simply one month trends or 12 month trends. And once you go for the different formulations, like breakouts, it's a bit more different.

Peter van Dooijeweert:

And you alluded to lag and how it affects returns. Should investors look at returns over a shorter time, say the past month, or is that not really enough time to determine a trend?

Otto van Hemert:

So, we extensively researched this and written about this as well in some of those papers you mentioned. Predictability of past returns with a positive sign is there for the last 12 months. This is in line with the academic literature on single name stocks, where there's often 12 month momentum that's being considered. Interestingly, the 13th month ago is not predictive with a positive sign. If anything is predictive with a negative sign. So, it's really about exactly the last 12 months of returns that you would consider in a trend formulation. If you look at within those previous 12 months how the predictability varies, it's more recent returns that are quite relevant. And you can imagine that an under-reaction to news is particularly well measured by more recent month returns, but also returns almost a year ago are relevant. But I think of those more as a seasonal pattern that you start picking up on. And then returns that are around half a year ago are a little bit less predictive because they're neither the short term under-reaction to news, nor a more seasonal pattern. They're in between and therefore less predictive. Oftentimes people focus more on the more recent months returns, not necessarily because those perform the best but they do lead to strategies that are a bit more defensive in nature that are providing more crisis alpha, I suppose.

Peter van Dooijeweert:

It's interesting you say that because I think a lot of us, when we think about trend, we just think of everyone's piling in, moving an asset in a fast way. But what you're describing is of a couple different markets in a way where there is this herd mentality and then there's this longer dated piece. Is the longer dated piece something academic maybe, or how should I think about the longer dated versus this shorter dated piece exactly? Who's driving it, in a sense?

Otto van Hemert:

I do think there's very different driving forces behind the predictability of returns more recently, and the predictability of returns almost a year ago. The predictability of returns almost a year ago, again, I think it's more seasonal related. A lot of things we do are related to the annual calendar. You can think of tax effects. You can think of 52 week reporting cycles. A lot of things follow an annual cycle and therefore have price pressure affects that follow that same annual cycle. And that shows up ultimately in a trend formulation if you've go back as far as a full year in historical return data.

Peter van Dooijeweert:

So, then every now and then we see certain asset classes have big moves. Equities might have a sharp up move as maybe people are chasing returns. Who knows what the cause might be. Do trend strategies just let them run, or do you take profits? Do you intervene manually? What do you do?

Otto van Hemert:

So, indeed, sometimes you are in a very big trend, a very big and long lasting move in a particular asset. Obviously, trend signals will pick up on that and you position yourself in line with that big move, but there are a couple of ways in which you do take some profits along the way. First of all, when there's a very extended long move in a asset that also often means that volatility is going up for that asset. And often a trend model will divide by volatility in order to allocate in risk terms. So, you invest a smaller notional amount when a market becomes more and more volatile. And in this case, that can mean taking profits on the market. Another feature of trend models is that you can explicitly codify that when a trend is becoming very strong and very long lasting, you may reduce your positions a bit. You may program in that the signal becomes a bit less strong because you're more at risk of a reversal of the trend at that stage.

Peter van Dooijeweert:

And is the reversal issue just because it feels like a higher risk of loss if the market's up 30% or is there something more to that?

Otto van Hemert:

I would say it's mainly a pattern that you can empirically establish, that when trends in a sense have been too strong, you empirically see a bit of a give up and that's somewhat predictable and you can modulate your trend systems to pick up on that. What the actual force behind that is, it could well be that there's been a bit of a herd mentality and at the moment that stops, there's a bit of a give up of, say, a price increase.

Peter van Dooijeweert:

Yeah. I guess it sounds to me a little bit like you're assuming some mean reversion in a trend and that seems to be paradoxical.

Otto van Hemert:

So, what is less common is to actually bet on mean reversion. What is more common is to bet less on trend continuation, still betting on trend continuation, but not in as big a size as you do during a more moderate trending environment.

Peter van Dooijeweert:

So, I'll pick up on the vol scaling a bit as well, I guess, because it's somewhat interesting this year. Different trend managers have different approaches. I guess, could you talk a little bit about what vol scaling means to, say, just a single asset, you touched on it already, and what the two sides of the argument might be?

Otto van Hemert:

Yeah. So, to a degree, vol scaling is a necessity when you trade macro markets because some macro markets are very volatile. Think of, for example, natural gas can easily move 10% in just one day. And other markets are not so volatile. An interest rate future, very short duration. A fixed income security doesn't move that much on a day-to-day basis. So, just because you have so many different securities, you have to scale by volatility to bring them on the same basis and make them comparable. But vol scaling then also typically is used over time. If a security at this moment is not so volatile, you take bigger positions. And if that same security is very volatile a month later, you take smaller positions. In other words, you almost allocate a fixed number of risk units to the security rather than a fixed dollar investment amount. It's something that helps with the risk management of the overall strategy and of that particular security in particular. And it's something that also over time allows you to have a more balanced return stream coming out of the trend model that you're trading.

Peter van Dooijeweert:

So, it's basically an issue of being greedy versus a more balanced return stream if I were to bottom line it. In other words, letting an asset run when it becomes parabolically straight up and highly volatile hurts your ability to manage a complex bundle of assets. Is that about right?

Otto van Hemert:

Yeah. So, it's about risk managing the exposure to that asset. And to the previous point we discussed, there's an element of profit taking sometimes as well. For example, you're in a big positive trend of a particular security, the security then also starts becoming very volatile and you reduce the notional allocation you give to that security at that stage and therefore taking some profits on that trend.

Peter van Dooijeweert:

Changing gears just a little. Oftentimes investment bank research, especially derivative of strategists, put out pieces talking about systematic strategies and what they might be doing, trying to guess if they're buying, selling, and they imply that there's some crowding going on. Do you think there's crowding in some of the bigger markets, say equities and bonds? Do you think they're capturing this stuff correctly? What are your thoughts?

Otto van Hemert:

First of all, I do think you always need to be very careful about crowding. That's not just for trend type models. That's for any model. When it comes to trend-following big macro markets, I don't think the size of the trend rates are big enough to easily move an S&P 500 index or a brand oil contract. In addition to that, trend followers really can be quite different in the timing of their trades, depending on whether they trade slow or fast or crossing of moving average versus breakouts. But you do have to be careful in sizing your trades in perhaps somewhat smaller markets you may be trading. One other thing to note in this regard is that there's many other participants in the market as well, not just trend followers. And in fact, a very natural way to trade for people is with more of a value-oriented approach. And that often is the other side of the trade. They are often on the other side because of security that goes down in price. A trend follower would be short. It's on a downward trend. But it will more and more look appealing from a valuation point of view often, and therefore value-oriented investors may be on the other side countering any imbalance flow you may be getting from trend-following.

Peter van Dooijeweert:

So, moving on to this year and inflationary times, you put out a paper last year that seemed reasonably well timed.

Otto van Hemert:

Thank you.

Peter van Dooijeweert:

And it definitely has turned out about as you expected. So, maybe why don't we talk a bit about the, paper a bit about the basis that you made predictions that trend would do well for this year, because you're not into timing anything in terms of strategies personally, right? You don't say when to buy value or when to buy growth. And so maybe let's talk about the research over the last 100 years and how it applied to this year.

Otto van Hemert:

Yeah. What we often see is that a big macroeconomic shock is helpful for trend type strategies. A fast sudden rise in inflation definitely fits that bill. And we've seen exactly that. We've gone from a very moderate, not so volatile inflation level to one that's volatile and rising quite fast. We've looked at exactly those scenarios over the last almost 100 years and across three continents, the US, the UK and Japan, and indeed, exactly during those rising inflationary time periods, trend-following strategies have tended to do particularly well. It takes markets time to fully digest the new inflationary environment, just like it takes central banker's time to recognize the new environment. For example, Jerome Powell has talked about transitory inflation for a long time period, but he had to drop that at some point because it's not transitory anymore. It's a real issue at this stage. And when inflation gets out of control as it has over the past year or so, you do see consistent winners and losers in terms of securities. And there's clear trending in those assets. Clear winners are commodity related securities. Indeed, commodities have gone up a lot over the last year or so, and trend followers have benefited by being long commodities by and large. And clear losers are there as well in terms of government bond securities. As inflation goes up, yields go up and bond prices go down. So, that downward trend in bond prices is something trend followers benefited from as well. And also historically, if you look at rising inflation era time periods, it's exactly those two asset classes, commodities and bonds, where trend-following has been most profitable at those times.

Peter van Dooijeweert:

What about equities issue?

Otto van Hemert:

Equities have a slightly more complicated relationship with inflation in the sense that equities also don't like inflation to be too low. Too low as in almost deflationary. So, when inflation goes from a very low level to a more moderate level, equities by and large welcome that and benefit from that in terms of price rises. But then if inflation goes through a more uncomfortable level of perhaps between two and 5%, that's when equity markets start worrying about inflation as well. And only at that stage you can see equities historically having sold off. Again, the more recent time period has been fairly similar in that regard. Equity markets held up reasonably well during last year, 2021, and only now this year, 2022, they've really started to sell off particularly in the first half of the year. So, equities have a slightly more non-linear relationship with inflation, which makes it slightly harder for trend followers to benefit from it.

Peter van Dooijeweert:

Yeah, I guess that's kind of interesting. As you describe the returns from trend-following, I think some people think, "Oh, it's probably just doing bonds and equities," but this year really is a multi-asset story. And maybe it's bifurcated in two pieces. There's correcting the things that I own, potentially bonds, and then diversifying into the other asset classes. So, you've mentioned commodities a bit, bonds. Without giving percentages of where you think trend followers made money, how important do you think the diversification benefits are, commodities and FX, where it's hard for people to make choices, like, "I'm not going to wake up tomorrow and decide to be long," some currency cross, and how much is the corrective nature of what's wrong with our portfolios?

Otto van Hemert:

It's an interesting question. So, I guess the corrective nature is that if an asset you would normally want to be long, like equities or bonds, if that goes down, there's almost a stop loss nature to a trend system. It basically goes in the opposite direction of what you're normally holding long and therefore limiting your losses on it. But really trend-following gets a lot of its performance from just being applied to a wide range of assets. Many you wouldn't be holding normally in a portfolio, including currency crosses and commodities. And that's a major part of the performance historically. This time around really short bonds and long commodities has been a big part of it, but there have been other knock-on affects of this inflationary time period as well. For example, the US Federal Reserve has started responding to the inflation a bit before other central banks. In particular, the ECB has been lagging behind quite a bit, and that's been really supportive for the US dollar in a very consistent way. So, also trend-following the US dollar versus a host of other currencies has been a profitable trend to exploit.

Peter van Dooijeweert:

So, the next thing people are thinking... This year has been inflationary. Trend has done well. So, don't know if inflation has peaked. We'll get data throughout the year. Even the Fed has suspended forward guidance completely. So, they're not even trying to predict anything anymore. But given the big run, should I just take profits now if I've got my inflation bit done so I can just move on to something else?

Otto van Hemert:

So, the nice feature about trend is that it's self-correcting. So, if the environment changes from inflationary to dis-inflationary, trend systems will naturally turn around. Not immediately. They will need some confirmation from price moves, but ultimately they will turn around. And we've seen some of that happening already at the start of the second half of this year. So, it's very hard to time trend strategies. I wouldn't advocate that. The more sensible approach I think is to hold on to your trend exposure and let it position itself in the way it wants to position itself. And can be a very different environment yet again in some months. To give you another example for that, it looked like plane sailing in January, early February 2020. There were some rumors about a new virus going around, but really all of a sudden the COVID pandemic bursts and big trends arose again. So, it's very hard to predict when those type of events will happen and it's better to have trend strategies employed and let them do the timing in terms of the trend signals picking up on any market moving in a certain way.

Peter van Dooijeweert:

So, that's a good segue into a different paper of yours, which is just as applicable today as the inflationary paper, and that's The Best Strategies for the Worst Crises. So, you've alluded to it before, but let's talk a little bit about crisis alpha and trend and how it will work during crises, where the returns come from, that sort of thing.

Otto van Hemert:

Yeah. So, before I mentioned that transforming strategies tend to benefit from big macroeconomic shocks and a very uncertain environment, and equity selloffs often happen exactly during those or in response to big macroeconomic shocks. So, transforming strategies have tended to do well exactly during equity selloff time periods. The global financial crisis was an example. The tech bubble bursting was an example, as well as for faster trends strategies, the COVID-induced selloff. So, trend strategies tend to benefit from weakness in equity markets because that's the type of micro backdrop that helps for a trend strategy. But interestingly, it isn't so much trend-following equities themselves that perform the best in those environments, but it's trend-following all kind of other securities that may show a big trend during the equity selloff. It could be that bonds are benefiting from a flight to safety move and being long bonds is a very profitable trend. During the equity selloff, the US dollar may show a big move. Oil may be impacted a lot during the equity selloff time period. And it's really more trend-following those other securities that historically has generated most of the return during the equity selloff period.

Peter van Dooijeweert:

So, you've written two papers in terms of crises, Inflationary Times and the Worst Crises, deflationary shocks. I don't imagine you're going to do a paper called The Best Strategies for Ordinary Times, but it does raise the question which you mentioned you shouldn't time trend, but what does it do in between these environments?

Otto van Hemert:

Trend-following strategies, they're particularly good during more distressed times. By implication, they are less good during more normal times when prices move in more range-bound fashion. There still can be trends in more particular esoteric markets. And it really helps to have a broad range of markets for which you follow the trends. And you just zoom into those markets where trends are to be captured. But all in all, it's a less favorable environment if markets are calm and equity markets are calmly rising. I would argue that that's also a time period when maybe there's less need for a diversifying strategy to pay off big time because your normal investment portfolio, say long stocks, is already providing you the good returns. I would not take this comment further in that you don't need to invest in trend strategies while markets are calm because of the point we discussed, markets may turn less calm in a very sudden way, and you can easily be too late if you don't have your more defensive strategies like a trend strategy in place already.

Peter van Dooijeweert:

Yeah. I think we've seen returns in some random periods where they're just long sustained moves, and FX don't seem to impact the market for one reason or another. I think we had that in 2006/7, 2014, those sorts of things. So, I do think it's interesting that there's just enough. I'm an options guy. I'm used to tail hedging for people. So, I'm a put with negative carry. And in many ways, what you're talking about with trend is you're a put for crises, but with some kind of positive return over time, which is, as we talked about it in a recent conference, a sharp distinction between tail hedging and diversifying strategies. And so I guess that's a good segue to a presentation we gave together in a conference in London, and then subsequently in New York. We talked a bit about convexity and how options have convexity and how trend develops convexity. Maybe you can, without walking us through the entire 45 minute presentation, give us just a sample of what that talk was about.

Otto van Hemert:

Yeah. So, it's about investments with a convex payoff, meaning that if, say, the equity market shows a very big negative or a very big positive return, that's relatively good for that investment, while it doesn't perform as well when there's more modest moves. So, investing in an at money put push call or straddle option position is an example of an investment with such a convex payoff. It benefits from a big move away from the current price. But the challenge with options, as we both know, is that you pay an option premium. And on average, you may not earn back that premium. Options can be expensive on average over time. Trend-following strategies share some characteristics with, say, a long straddle position in that it does well when there's big moves because often those moves are achieved in some trending type way. A big trend up or big trend down means ultimately it's a big move in either direction. So, trend strategies share that same convexity pattern that they benefit from big moves in the equity market, in other markets, and may do less well in terms of smaller moves in those markets. But the point you were alluding to is that on average over time, trend strategies have tended to pay off positively while as options have been expansive to hold on a continuous basis. So, trend is not as guaranteed to be positively convex as a straddle position, but on the positive side, the long run return historically has been better than that of just holding straddle option positions.

Peter van Dooijeweert:

And I guess that construction of position in trend accounts a lot for the crisis alpha and the alpha we see during inflationary times where maybe you didn't get it the first day, but then as the larger trend develops you've picked that up in a similar rate options, but without the premium. Is that about right?

Otto van Hemert:

I would say that's about right. You can try to draw the parallel between trend strategies and option investments in a more precise way. And you can try to delta replicate an option, which means in case of a straddle position, you're buying more of the underlying if prices go up and you're selling the underlying if prices go down. That's exactly what a trend-following strategy does as well. But we also know that delta replication of options goes wrong if there's a big gap move. That is the weakness of that method. And that's also the Achilles heel of a trend strategy. It needs a somewhat gradual and sustained trend. And when there's gap moves, it's less suitable for trend strategy.

Peter van Dooijeweert:

Okay. Well, great. This has been a really enjoyable conversation. Thank you for joining us, Otto, and thank you everyone for listening. We've linked Otto's papers in the episode show notes. You can find them on the Man Institute. And on our next episode, we'll be talking to Giuliana Bordigoni about trend-following in emerging and alternative markets. So, we'll look forward to seeing you then. Thanks again, Otto.

Otto van Hemert:

Thanks, Peter.