Top Traders Unplugged podcast

Best of TTU – The Evolution and Future of Trend Following

By  | 

Via Top Traders Unplugged

Best of TTU – The Evolution and Future of Trend Following


Regardless of the investment strategy that you apply to the markets, your strategy or approach needs to evolve over time.  The way we did Trend Following back in the 1970s, 1980s, & 1990s is not exactly the way we do it today, and I think this can be said about most, if not all strategies.  Now with some strategies, the model decay is so rapid, that if you don’t adapt quickly you can lose your edge.  I think short-term strategies are good examples of this.  Trend Following, being a longer-term strategy is, in my opinion, a lot slower when it comes to Model Decay… so you need to have been around for a really long time in order to have witnessed this evolution.  

So when I looked through my list of guests to pull a few golden nuggets from on this theme, I thought that Marty Lueck, the co-founder of AHL and Aspect Capital would be the perfect person for this.  So enjoy these unique takeaways from my conversation with Marty, and if you would like to listen to the full conversation, just go to Top Traders Unplugged Episode 37Episode 38.

How Trend Following Has Evolved Over Time

Niels:  What has been the biggest changes over time and is it really small incremental changes, or is there something where you look back and you say in the last 15 years 2008, or 2009, or whatever it might be, we did actually discover something that we would say that was a big upgrade or that was a big find – key finding?

Marty:  By and large it is very much an incremental process and we make a virtue of that because you don’t want… the last thing we want to do, especially with our focus on institutional investors and a high level of transparency. The last thing you want to do is surprise an investor. With the benefit of hindsight, I highlight two particular features about the evolution of the approach. First, in an odd way… the first, Niels, is the importance of risk management and portfolio construction. I think this is something that investors and maybe managers that haven’t been doing it for that long may underestimate the importance of this in the process, and again I’m saying this because I did (laugh). After all of the shenanigans of looking at Chartism and distilling it down into to so fundamental tech models, you come up with a pretty robust diversified set of medium term trend following models, or we did anyway.

martin-lueck-01

The neat thing, in the 1980s was the range of markets that have sprung up around us, Niels, afforded us a level of diversification that essentially… the combination of trend following across that range of markets it risk managed itself. You didn’t have correlated risk shocks. You didn’t have … there was enough intrinsic diversification that if one sector was melting down you’d have opportunities in another sector. Risk management…I couldn’t spell risk at the time. Then a couple of things happen. First of all (I’m going to foreshorten this) you got to an era where I think some of those trend following models became less efficient. You go to an era where markets did become more homogenous, so there’s both a sort of macro effect as your pension fund manager in Japan begins to hold a similar looking portfolio to your pension fund manager in Sacramento. Whereas, once upon a time they didn’t, it was much more parochial.

”You got to an era where markets did become more homogenous, so there’s a sort of macro effect, as your Pension Fund manager in Japan begins to hold a similar looking portfolio to your Pension Fund manager in Sacramento”

You begin to get a greater coherence of both investor holdings and then also with the advent of VaR metrics and that approach to risk management, you also got a more correlated response to events, so that everyone around the world who thought they were doing independent things would react in the same way to an event. In response to those kinds of increasing correlations in the markets and increasing propensity for shock effects we – both Aspect and as an industry, began to look for sources of diversification. Once you start to diversify, obviously markets is one axis and time scale is another, but once you start putting in other models, then how you bind them together, and that ad mixture becomes super, super, super important. I’m sure this is kind of obvious. It’s been an area that we’ve focused a lot on. How do you put them together carefully? How do you make sure that you constrain.., because just the simple thought experiment – if I take two models which have zero correlation between them and I leverage them up to achieve the same standard deviation of returns as one model on its own, well hurrah! I’ve just improved my returns, but I’ve also let the kurtosis creep out. It becomes increasingly important as you make the portfolio more complex that you deal explicitly with all of the edge cases – the risk management edge cases. That’s actually stood us in good stead as the markets have gone really into strange places since 2009.

Niels:  Since you mention that, what have you learned in these last few years in terms of trading and systematic models and how the environment plays such a key role in all of this?

Marty:  I’m making note to come back to a question you asked earlier, but, oh gosh, in this environment has had a…we could spend another session talking about this because, Niels. Clearly in a period of underperformance for the strategy, we can…it’s human nature, why is it underperforming? What’s going wrong here? Ah, I invested in you Marty, because I saw your 2008 performance, what are you doing? Have you all saturated the markets? Have market dynamics changed? Has trend following stopped working? It’s all of those questions, again, and you just hit the nail on the head. I don’t want to appear glib so of course we investigate all of those things. We look at both our market footprint and what we think is the footprint of our entire industry to satisfy ourselves that we’re not…this isn’t shooting ourselves in the foot that’s happened here. We look at the low volatility environment and what that is likely to do to both the opportunity set and to the risk management challenge. It’s been a trying time, but then I guess in one sense I’m fortunate or cursed with having lived through periods like this before. After 1987 – a lot of parallels. There was a great…after the October crash of 1987 there was a huge run up. Managed futures delivered its crisis alpha. We delivered our crisis alpha, and it went roaring profitable into 1988 and then basically hit the doldrums. The analogies between then and now in terms of recession, Savings and Loan crisis…remember that one? Government intervention, managing the yield curve, suppression of risk appetite – a lot of similarities. You’ve got to dig and scratch a little bit at the AHL track record, but it took from the high in the middle of 1988… I don’t think AHL was back in new highs until sometime in about 1993 or 1994.

ON-CS802_market_OR_20181102091718

It was a similar length to the doldrums that we’ve been in here. It actually didn’t end neatly. It wasn’t just a sort of pleasant recovery of favorable returns. Just when you thought things were getting better we got kicked in the teeth by a surprise rate hike in 1994. But that sort of presaged… it almost felt like the tubes had been cleared, and the starting gun went off and the markets returned if you will, to some sense of normality, whatever your impression of normality is, and there was a great round of performance. 

I’m not predicting that, but what I’m saying is that I draw comfort from having been through it before that actually, provided the models are able to adapt to the fact that no two days are going to be the same, and you know that’s the beauty of what we do, because it’s not scenario specific, so the models can adapt to whatever the markets present, number one. It does speak to the persistence and having confidence in the approach.

Niels:  I think the other thing that people should do if they want to sort of satisfy themselves about why these strategies shouldn’t actually make money in the environment we’ve just been through is just look at the price range compression that we’ve seen. Just looking at what’s the high and low been for the last rolling three, or six month basis, it is so clear what happens to the prices a few years back and when you do trade momentum and the price ranges compress as they have done, it’s very easy to visually see that we shouldn’t be making money. It is so difficult for investors to accept that. What I think is even more interesting is that now that we’ve seen a lot in the news recently and in the last years about trend following not working and all of these things that normally pops up, a lot of the longer term, maybe, trend following strategies they’re all calling back to all-time highs again. People are not noticing it. They’re just obviously focusing on when things are not working, but they’re not really focusing on the fact that we are back to all-time highs and I think you guys are as well, and many people. So it’s very interesting, and I completely agree with everything that you’ve said.

Marty:  Niels, the thread that I forgot there was just it reconfirmed my belief that a systematic approach is … I mean horses for courses – there are some great macro traders but I can tell you I’m glad I do what I do rather than be a macro trader because how many times do you think folks have said, “well yields can’t go any lower than this.”

”You can get overly focused on an individual Trend Following model at an individual time scale, but it’s the combination of all of the pieces put together, and the combination of all of the markets put together, and how you manage risk that determines your end performance.”

Niels:  So on one part I think many people think of trend following as, OK, but you get your signal to buy or to sell, and then you follow along for the ride, and you have some kind of position size algorithm, but when I listen to what you’re explaining, I think what you are actually saying, is that since you have smoothed out this process, the position size is more a reflection of the strength of the signal. Because the more confirmation you get, the bigger you will build your position and so on and so forth. In my mind, I think a lot of the secret to success of trend following is not so much where we buy and where we sell, are we a day late, or a day early? A lot of it is really the risk management and thereby the position sizing itself. That’s a big part of the secret sauce to the success or the robustness of trend following. How do you view that?

Marty:  I think you’re absolutely right, it’s what we do as a holistic challenge. You can get overly focused on an individual trend following model at an individual time scale, but it’s the combination of all of the pieces put together and the combination of all of the markets put together and how you risk manage the whole that determines, obviously your end performance. So I make two observations. The first is that the positions that we hold are, yes, a function of signal strength and conviction, but that, as you would expect would be modulated by what you perceive as the risk of the market, so for a given signal strength, if I see the volatility of a market, which is a cipher for risk if you will – if you see that say double, I will effectively have my position to maintain the same risk for that given signal strength – point one. Point two is that it’s not an inexorable line. The stronger the trend, the bigger the position I’ll put on because as you can imagine, Niels, that way lies madness. So there’s knowing effectively when to back off and perhaps when to be a provider of liquidity to the markets rather than a consumer of liquidity. That’s another delicate feature of what we do.

Dont-Lose-your-Patience-by-mrsmas

‘We’re not slamming the markets with very fast models, so our execution algorithms are predominately looking to make us obfuscate what we do, and to capture, if you will, a ‘patience premium’. We’re not in a hurry to get our business done, but it’s an ongoing effort.’

Niels:  Yeah, so in effect, just to be clear, you’re not actually using a stop loss per se, because it comes automatically as the strength of the signal changes, your position changes along with it.

Marty:  That’s exactly right. So it’s a gradually modulating signal. The other thing that I really like about what it is we do, and obviously we spent time researching super-fast intra-day models and … and there’s utility in there and some people do them very well, but one of the things that I like to stress, for an institutional investor, is the intuitive qualities of medium-term trend following. What I mean by that, Niels, is that if you read the Financial Times or the Wall Street Journal or your local financial newspaper and you follow roughly what’s going on in global markets, then you will have a good intuitive sense of the positions that we hold. Obviously the detail and the complexity of exactly the position size that we hold, which as you’ve said, is a function of signal strength, volatility, where you are in the development of that trend, portfolio construction, risk management – are you up against any exposure constraints, all of those pieces holistic challenge very complex, but by-and-large if you stand back just a short way and look at the ebb and flow; at the dynamics of how that portfolio is moving from day to day, it’s very intuitive. You can broadly, as an institution investor, understand why a trend following portfolio makes and loses money when it does.

Niels:  Which is always, actually puzzled me because people often criticize what we do from saying that it’s complex, it’s difficult to understand. I’m just puzzled about this because it’s really not that difficult to understand that when something goes up you buy, because you think it’s going higher, and if it goes down, you sell, because you think it’s going lower. Compared to a fixed income arbitrage or whatever they call these strategies, yet people seem to love those strategies more. I wanted to ask you just a final point about the program itself. Going back to research a little bit here, how much research do you actually need to do to overcome, or to improve efficient execution? Is that a big part of research when you get to your size, to make sure you can continue to grow and have efficient execution?

Marty:  Yes, as a research team we look at the problem of continuing to evolve and develop the program. We break it down into the core trend following components; the diversifying modulating strategies; the portfolio construction and risk management piece; and execution. By and large there is always someone working on something in each of those areas and over the sweep of time we will have periods of more concentration in one area than in others, so it’s a bit of an ebb and flow. About four years ago we embarked on the transition to a wholly box-to-box. If you look at the sweep of evolution of execution, back in the good old days of where we would trade binary in big clips, and we had to get it down to an open outcry market, our execution research, in those days, Niels, was to go to Chicago and meet the biggest and “baddest” floor brokers and you’d hire them, because they got to the front of the pack.

So over time markets have become more electronic, and that’s played into the technology led firms. Then actually taking the leap to a predominantly box-to-box world or be it monitored carefully, that has been an enormous commitment of research effort and investment. It requires ongoing monitoring. I would stress that we don’t do High Frequency Trading (HTF), so we’re not going head to head with sort of an HFT firm, where you’re swapping out algorithms every few minutes or every few hours. We have a suite of execution algorithms which are generally fit for purpose. We ensure that they are correctly parameterized for the liquidity. Markets change their characteristics, once the sort of resting a bit off our spread, what are the typical clip sizes that people are making available, those are the characteristics of markets that we need to review and re-parameterize on a regular basis, and obviously monitor if they’re changing more rapidly than we expect. It’s an ongoing monitoring effort. I think because we’re not slamming the markets with very fast models, our execution algorithms are predominately looking to make us (I’d love to say invisible) but certainly obfuscate what we do and to capture, if you will, a patience premium. We’re not in a hurry to get our business done, but it’s an ongoing effort.

READ ALSO  Multiple Income Streams (Podcast Episode 67)


Print Friendly, PDF & Email

Hold dit netværk orienteret