You Don’t Buy It for Returns. You Buy It Because It Works! ft. Yoav Git

[00:00:01] Intro

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

[00:00:23] Niels

Welcome and welcome back to this week’s edition of the Systematic Investor series with Yoav Git and I, Niels Kaastrup-Larsen, where each week we take the pulse of the global markets through the lens of rules-based investor.

Yoav, it is really wonderful to have you back this week. I hope you’re doing well and how are things where you are? You’re in an interesting place right now.

[00:00:42] Yoav

Yes, it’s a pleasure. I’m actually visiting my parents and in Israel. So, I didn’t quite expect to be in the middle of sort of a global crisis, but there we are. I get a lot of chance to spend time with my parents, which is really nice and to see how that impacts the oil market, which is also a little bit interesting.

[00:00:42] Niels

Yes, it is. And then you’re incredibly calm, speaking to you and obviously I can see you. So, I’m glad to hear that all is well where you are at the moment.

Now before we dive into… I mean, I will say this definitely is a trend following focused conversation today, but with a lot of great inputs from a few papers that we managed to find. And so, I’m really looking forward to that.

Before we dive into that though, as always, and it sounds a little bit weird when I say it, I’d love to hear what’s been on your radar, and based on where you are right now, the radar part seemed a little bit odd to say. But anyways, what have you been thinking of, or what have you found interesting in the last recent time?

[00:01:59] Yoav

Actually, if it’s okay, I’m just going to go back to our previous TTU. So, just a bit of housekeeping. You asked me about publicly available indices in the bond market, and I didn’t have it at the time because I use Bloomberg a lot but I’ve looked around. Market Watch has a website that gives you a lot of yield curves in data which is really nice, and you can download the CSVs, and it’s useful. And the other website which you need to register for, is also very good is Cbonds.com. So, there’s plenty of data there which is also about volumes and issuance and all things bonds which, actually, is very good. So that’s just one follow up.

And, interestingly enough, the other discussion we had at the time was that I told you I was looking at AI, and I mentioned base44.com which is a site, an application where you can build your own website using AI. And I was very thrilled about it because it sorts out the connectivity with databases, with emails, everything.

And interestingly enough, just this week they announced that Wix, which is a website designed for building websites, is buying that application to increase its AI capabilities. So, I was very pleased that you heard it here first, about a month ago. It’s quite amazing – the work of one person that, you know, is worth about $80 million within six months of work. Really impressive.

But the other thing that has been on my radar is actually a book I’ve been reading. It’s called your Life is Manufactured, by Professor Tim Minshall. So, proper disclosure, I actually know the guy and I’ve known him for a while. He’s a professor at Cambridge, and he’s a great speaker. He’s not just a nice person, he’s also a great speaker and he’s a great writer.

And what is interesting is that he’s writing about the process of manufacturing, not from the ivy tower of a Cambridge professor, but really from the ground floor, really understanding the production chain, and how everyday items like a kettle is produced, and the constraints, and the global trade. I think it’s very interesting. It’s a very different view to what you would get in a normal economics book, but it’s very, very enlightening and lots of anecdotes. So that’s a lot of fun.

[00:04:40] Niels

Good. Well, you certainly have been busy.

You know, when I was thinking about today’s conversation, I was thinking about what has really been on my radar. And one of the things I noticed was that it wasn’t so much what had been on my radar, it was how little had been on my radar from the G7 meeting this week. And, you know, very few things came across that I felt was interesting.

And maybe this is kind of just another sign that these big institutions, that we’ve been used to, where they have pretty significant power, and people would listen to every single word that came out from these meetings, are just becoming kind of less (respected is not the word I’m looking for but I can’t think of another word right now)…

But it’s in terms of their ability (maybe is the word) to make, a meaningful difference to the world. And maybe in a world where it’s dominated by very few individuals and by narratives, and social media, and what have you. So, I thought it was interesting that I hadn’t heard more about that, for example.

And also, I think the statement that Macron made (the French president) just saying that Carney (the Canadian Prime Minister) fulfilled his mission as a host, preserving the unity of the multilateral organization. I think that was a very weak compliment in terms of the achievement for getting these people together.

Then something else caught my attention. It was a Bloomberg report/story that came out and it was actually related to the UBS Global Wealth Report that has just been announced. And I think the number surprised me a little bit. Apparently 684,000 new millionaires were made last year according to the report. I mean that is a staggering amount, really. And they also talk about the average wealth in different countries. And also, the difference from number 1, even down to number 10 is, is quite extraordinary. And then, of course, what also came through my radar this week was the fact that there was a very lucky winner (I don’t know if you follow this), a very lucky winner.

[00:07:14] Yoav

Yeah, the Irish guy.

[00:07:14] Niels

Oh, damn, then I don’t need to…

[00:07:16] Yoav

Well, maybe a lady.

[00:07:18] Niels

OK, then I don’t need to check my EuroMillions ticket because apparently they won 233 million franks, Swiss francs, as they were the only winner this week of the EuroMillions. However, it’s a good thing because I think, in today’s conversation, we can actually help them in terms of what they should invest in with all that newfound wealth. So, hopefully they’re listening, and they’ll learn something from our conversation today.

[00:07:55] Yoav

Absolutely.

[00:07:56] Niels

Okay. With that aside, let’s take on some more serious topics.

Trend following report, so to speak, update as we are kind of a little bit more than halfway through June. What strikes me in June is not so much that it’s moved dramatically in terms of performance. There’s been a bit, some have outperformed, some have underperformed. But actually, what I’m noticing, at least from where I’m sitting, is that the daily volatility is actually pretty meaningful in the returns. And, of course, this is depending on, to a large extent, what the exposures are, in the specific sectors, that are moving this month for sure. But it is maybe also a sign that the portfolios, given what happened in April and May, and where lots of positions and changes happened in response to that, perhaps trend following portfolios might be a little bit uneven (if I can put it that way) in terms of where the risk is allocated. So, I thought that was kind of interesting.

The other thing I noticed so far this month is that, while short-term managers did well in April (from memory), maybe even in May (from memory), they’re certainly catching up to the downside to the trend followers in June. They’re having a pretty rough time in June so far according to the SocGen Short-Term Traders Index which I’ll come to in a second.

I’m also noticing that, as far as I can tell (again I don’t have a full view here), but as far as I can tell from the models that I follow, commodities are doing so much better than financials this month. It’s really the financials that are causing a problem, well, maybe with the exception of oil. I think you were right in saying that oil obviously has been majorly impacted from what’s going on in the area where you are right now. So that is causing some reversals that are against the longer-term downtrend, at least for now.

My own trend barometer finished at 48 last night. Now that, in itself, is kind of a neutral reading, but I will say, comparing to the rolling 10 days, two weeks, that I also publish every day on the website, it is a little bit firmer, a little bit higher than what we’ve seen.

BTOP 50 index, and I believe because of the US holiday yesterday, these numbers will be as of Wednesday evening. So, BTOP 50 up 1.55%, that’s respect predictable, down 3.42% for the year. SocGen CTA index up 72 basis points but down 7.83% for the year. SocGen Trend up about 0.5%, down 10.8% for the year. And, as I mentioned, the SocGen Short-Term Traders Index is struggling. It’s down about 2% this month alone and now down 4.63%. And given its much lower volatility than the Trend index, it’s actually now pretty much on par, I would say, on a vol adjusted basis, with some of the other indices despite having a much better time during April for sure.

MSCI World up 70 basis points so far in June, up 5.9% for the year. The S&P US Aggregate Bond Index also up 41 basis points this month, up 2.8%. And the S&P 500 continuing its upward trajectory towards new all-time highs, up 1.26% in June, up 2.34% so far this year.

Before we dive into the first paper, any observations from your side in terms of trend, CTA, anything you’ve noticed?

[00:11:44] Yoav

Well, I’ve had a slightly different experience of May because SAFI, my fund, tends to switch between fast and slow momentum. So, it was having a sort of a much better time. It was feeling happy in May. But, in general, I think it’s been a difficult period all around and the industry has been struggling to explain itself. But there are a few good papers which are coming out. So, AHL put out one very good paper. The paper we’re just about to discuss is a very good paper. So, actually there are quite a few. I think the industry has upped its game in terms of explaining what it is doing, which I think is very nice.

[00:12:31] Niels

Yeah. Okay, well, let’s try and help that effort.

The first paper that caught our attention is called Turbocharging Your Portfolio with Negative Correlation. It is written by Dan Galanter from R G Niederhoffer Capital Management. It came out this month. You know, in a sense, to encapsulate what it’s about, you could almost say that what if the most powerful addition to your portfolio actually loses money on its own. And today we’re reminded why negative correlation, not high returns necessarily, may be the smartest asset that you can own.

And of course, why negative correlation is kind of relevant at the moment is that for sure, CTAs have provided negative correlation in the past year or so compared to the equity markets that we are obviously trying to encourage people to diversify away from. So, do you want to start out in terms of laying out the premise of the paper and what they’re finding in it in this new take on this?

[00:13:48] Yoav

Yes, absolutely. So, there are two parts to the paper. The first part looks at the theoretical sort of aggregation or allocation to two strategies. You have one, you have your existing portfolio, and then you have the additional strategy that you want to allocate to. And the observation that they make is that, actually, sometimes it’s better to have a lower Sharpe and have a higher correlation.

They actually have an exact formula which is well known. But it is good to have it in the paper and it’s about sort of long-term allocation. If you have long-term strategy with negative correlation, even if it is losing money, even if it is flat, it will add to your risk adjusted returns because it will reduce the volatility of your overall portfolio and that will increase its risk adjusted returns. So that’s the first part of the paper, the next one is about sort of merging, taking a specific example of recent period where we think about the investor having an equity portfolio and say, what would be beneficial to add to that portfolio over the recent history (I think it’s like the last two years or thereabout)?

And they look at two examples. The first one is a multi-strat (which I think if you look at the paper you’ll figure out what that one is), And actually, it’s a multi-strat which has both good performance but also relatively low correlation to the reference portfolio. So, it has, I think, a 12% correlation which seems on the face of it like nothing. And that’s alternative A.

And then alternative B, they look at another manager with performance which is less stellar than sort of a multi-strat. But it has two very positive aspects that are very good to you if you’re building up a portfolio. The first one is it has a negative, I think, something like 30% percent correlation to the equities. And the second aspect is the high volatility.

So, one of the nice things about high volatility products is that they give you capital efficiency. You know I’m a big fan of that when you’re building up a portfolio. And they examine what happens if you add the multi-strats to your portfolio and what happens if you add this negative sort of defensive strategy to your portfolio. Even though it added less standalone expected returns, it still added quite a bit to your portfolio.

And in this particular example they explain why. I mean you can see why, with real numbers, why that would have really contributed to a higher Sharpe portfolio. And because the strategy has got relatively high vol (the negative correlation strategy), it will have a meaningful impact even with relatively small allocations, it will start making sense to be included in your portfolio as a hedge. So, that’s the paper.

Okay, now a little bit of… I really like the paper, I have to say. It’s straightforward, it’s explainable, it’s understandable, but it’s very mathematical. And I think it’s important to understand the real-life constraint of an allocator when it comes to what you can really add to your portfolio and the problem people have with allocating to defensive strategies. And defensive strategies can be multiple. It can be tail protecting. So, it can be option buying. It can be fast trend.

So, fast trend has got much higher convexity and it’s a little bit like buying options but with realized volatility rather than implied volatility. You can have trend following which is where we cross the boundary between alpha and hedge. So, trend following is not only zero correlated to the underlying markets but also has a positive alpha. And you can have other defensive strategies, maybe a quality factor in equities, so long/short equity with tilt towards high quality stocks.

So, you can have a collection of different defensive strategies which can go into your portfolio. But I think the real trick is for an allocator to have that in their portfolio, it’s difficult because the standalone performance isn’t that great necessarily. So, even if you are a mathematician and you think, oh, that really adds value to your portfolio, you need to justify it to your investment committee. You need to come along and explain how can you justify having a low Sharpe strategy if it has negativity? For sure, it adds value to your portfolio, for sure, 100%. But you need to be able to justify that. And I think that’s one aspect of real life that the paper doesn’t do. But I think the principle that they’re showing is really, really nice.

[00:19:23] Niels

Now, obviously finding a strategy that, on one hand, is, I wouldn’t say negatively correlated because trend following is not really negatively correlated to equities. We’re non-correlated, but we can be negatively correlated at times as we are right now on a rolling, whatever, one, two, three year period. We’re probably negatively correlated at the moment, I would imagine.

And then you mentioned that there are other things that can be negatively correlated. But besides trend following, where we have a long history of actual data, which I think is so important, especially if you’re going to pitch it to an investment committee, being able to essentially provide evidence that’s not based on simulations and what have you, it should be, you know, valued quite highly. But what other strategies can you think of that have a long track record for doing this?

Because this is kind of where I struggle a little bit in terms of finding really good, sustainable alternatives, but also alternatives that can scale as well, which is the other challenge, especially if you’re talking about shorter-term strategies. Is there anything you’ve come across so?

[00:20:46] Yoav

Absolutely. So, if you look at the strategies that I mentioned, I mean quality in equity, I mean in some sense, goes all the way back to Graham.

[00:20:56] Niels

Right.

[00:20:57] Yoav

And you can run it, and it certainly is a benefactor since the ‘70s. If you look at options markets, you know, we have a reasonable history. But again, capacity will become an issue. And for fast trend following as well, you have those programs that run, you know, a billion or two, but there is a capacity because the amount of trading that you do, you churn, is a constraint.

There are certain types of sort of insurance products where you can do things by going to asset classes which are very different to the asset classes that you have exposure to. There is catastrophe bonds in my fixed income universe. So, you can get exposure to that.

So you can get exposure to things where the return stream is not necessarily correlated to what you would get from a 60/40. Interestingly enough, I don’t think that private equity is a particularly strong diversifier.

Private equity and private credit has been a big buzzword, but generally not 100% clear to me that they are not exposed to the same economic factors that sort of long equity is exposed to. But you know, things like gold…

[00:22:27] Niels

Sure, yeah, we’ll definitely touch on that for sure in the next paper.

So, just to kind of summarize, essentially what they’re saying with this paper, which again is relevant, nothing new, but relevant and always good to be reminded, especially in a time like this, is that a lot of investors are basically not necessarily approaching this or asking the right question when it comes to how can I improve my equity portfolio? They’re looking for something with a high Sharpe because that’s what they look for.

But what they really should be looking for is something that can truly de-risk their portfolio even if it, “on paper looks like it’s a modest performer”. But if it has some other attributes such as negative correlation at the right time, it is really magical when you combine that.

And it’s funny because a number of years ago, on our side, at Dunn, we started just to monitor, and I did as well, just kind of a portfolio where you just say, well, what if you took 50% of our strategy and 50% of say the S&P 500. I know later on we’ve had products being launched like this, but this is something that I’ve been looking at for a very long time.

And it is quite interesting to see that the combined product, and in this case I just took 50% equities and 50% of Dunn, and if you look at that, even despite everything that’s happened in the last year or so, that combination is still pretty close to all-time highs. And when you then start looking more on the return streams, what return did you get from this part of the portfolio? What return did you get from the other part?

You often expect that when you blend two things together, you get some kind of average, but in this case you almost get the sum of the two. It’s crazy. I mean it’s so powerful. And yes, I mean, I wish more people would be looking in that direction.

[00:24:32] Yoav

I think the important thing about trend following and about being systematic is that you essentially guarantee the zero correlation. So, a lot of the time one of the problems when an investor approaches a strategy which says, oh, I’m in a defensive strategy. It’s got that negative 30%. What they are worried about is, is it going to stay that way? Is this actually going to be minus 30%. So, in the case of mechanical buying of options you’re kind of guaranteed because you’re buying puts.

But a lot of the time if you have a tail protect strategy and they say, well, what would happen if you try to save money and you didn’t buy the put options exactly when you needed them, then when the crisis comes you actually don’t provide that protection. What is nice about trend following is that it’s not just zero correlation (that it so happened to be, historically, the data was zero correlation over history), it is that, by the way we construct the portfolio and we follow the rules, we are by construction sort of zero correlated to the assets that we trade.

And that gives the investor a lot more confidence in terms of allocating to trend following because they know that this is almost inbuilt. The symmetry between both long and short is built into the strategy and that gives you the power to sort of strap it on a long only exposure in bitcoin, in gold, in equities, in bonds, whichever asset class that you feel like, in your long only portfolio, you want to have trend following will still be zero correlated and will add value. I mean it’s just additive.

[00:26:11] Niels

Yes, I completely agree. I should, by the way, also just say a big thank you to the person who provided me (a good friend of the podcast, great guy), who provided me with a copy of this research paper which both of us, both you and I enjoyed reading, Yoav.

Luckily you mentioned that there are other things such as gold, etc. Well, luckily we have other friends in our industry who write great papers. And once again we’re very delighted to be able to stand on the shoulders of Quantica, in this case, who in their latest quarterly report, which I just saw it this morning and shared it with you. So, it’s not like we’ve read it in in detail, but I think we can talk a little bit about some of their findings. But another great paper.

And you know, a good reminder, again, a lot of the papers you said maybe we’ve become better in explaining to investors these difficult times. And I think, as you said, this is super important that we do that and we spend time doing that.

But for people who’ve been in the industry for a while, of course we may not learn something new from this, but it’s a great reminder of some of these things. And I really do look forward to digging into some other papers in the coming episodes for sure.

But essentially what Quantica did, and again we only literally saw this like an hour or two ago. but essentially what they’re looking at is I think trying to say, okay, if we have equities (and most investors, institutional investors, will have a bond equity portfolio) we use the 60/40 as the usual benchmark. I’m not even sure in Europe that’s so relevant because it’s almost the opposite, I guess, more bonds than equities, unfortunately. But anyways, what they set out to do was to try and see, okay, how does trend do, but how do other things such as bonds, which has been the traditional protector mitigator, and also what about gold?

And they go through different scenarios. They look at the best 10% of equity periods, they look at the worst 10% of equity periods to see how these various alternatives to equities would have done. And then they also do a combination of the three. So, if you had a 60% equity portfolio, what happens if you combine a third, a third, a third of trend following, gold, and US Treasuries in this case?

Feel free to jump in and talk about if you have some takeaways from this. Otherwise, I’m happy to kind of run through some of the conclusions, but feel free to jump in here, Yoav, if you want.

[00:29:06] Yoav

No, I think run for the conclusions and then I’m going to do a little bit of take.

[00:29:10] Niels

Yeah, so they also do something great which is they look at different time periods because, clearly, we’ve had some very distinct periods really in terms of financial markets. The 1990s were very different from the sort of period from 2000 to 2015, maybe even all the way to ‘19 where central banks were in control. I see that as a very distinct environment for the markets. And then they look at the most recent 10 years, which makes, sense where we start to see some changes. And then of course they look at the full period of 35 years from 1990 to 2025.

Now, in terms of trend following, they do prefer to use the SocGen Trend index. However, it didn’t exist in the ‘90s, so they are using a proxy for that period, it should be noted. And what they find is that in each of these distinct periods, of course it’s not always going to be the same diversifying component that adds the most value to your equity portfolio. But when they look at the combined or the full period from 1990 to 2025, there is oddly enough some consistency, interestingly enough.

Let me run through some numbers. So, the Sharpe ratio, using the risk-free rate of return of a three months dollar rate as the risk free rate of return, they find that The S&P 500, if you only invested in that, had a Sharpe ratio of 0.49. If you had, instead, a 60% equity portfolio and a 40% CTA allocation, you could increase that to 0.64 for your Sharpe. So, a massive increase really. You would also have increased it, to some extent, if you use gold as your 40. It would have had a Sharpe ratio increasing to 0.56.

And actually, no surprise really, the best single portfolio diversifier would have been bonds. If you had 40% bonds, it would have gone up to 0.69% during that period. And I think it should be noted that bonds for the majority of this time period, I mean obviously more or less for the full part it was in a falling interest rate environment for sure, up until 2021 I think it was. So, there was some benefit from that. But still as a diversifier it did best.

However, then they do the mix where they basically say yeah, we keep the 60% equities and then in the 40% part we now just take a third, a third, a third of gold treasuries and CTAs and lo-and-behold, it actually delivers the best overall portfolio Sharpe of 0.72 (not to be misconstrued with the hedge fund called 0.72), but 0.72 is the Sharpe ratio for this combination.

So, you know, again, I don’t think this is surprising, but I do think it’s relevant, especially in a time where we’re seeing signs of changing regimes, changing market environments. Having a diversification across your diversifier, so to speak, across your risk mitigating strategies, I think makes sense. Obviously all these three are very liquid, so I think they’re relevant. And I think also kind of instinctively it feels like you’re diversifying.

Like you said some people think that private equity is diversifying, but to me it’s still equities and we don’t exactly know where the “improvement” comes from. Is it just a lack of pricing or whatever it is? So, I think this is definitely worth a read and for investors to take note of.

In fact, you don’t even have to be an institutional investor really to benefit from this. I mean, this is something most investors, I would say, can do through ETFs, and depending on their portfolio, maybe they can make it a little bit more, you know, cost efficient also by finding deals directly with managers, and so on, and so forth. But it’s doable and it’s definitely worth paying attention to. So, what’s your takeaway?

[00:33:42] Yoav

So, absolutely. So, the first thing is that the reason why bonds were such a good additive aspect to the equity portfolio is exactly what we discussed in the earlier paper. If you were looking for a negatively correlated asset class throughout the period that Quantica are looking at, for a large part of that period bonds were negatively correlated to equities. I’m not sure if it’s going to persist.

Certainly, it’s been choppy about that recently. So, during periods of inflation you have issues about that correlation staying negative. But certainly, in the period of falling rates, falling inflation, bonds as a negatively correlated asset class to equities is very additive.

Now, the interesting thing is exactly about how much trend you want to put on top of that. And the interesting observation is that even the mighty 60/40 portfolio, from 1990 has got a Sharpe of 0.7, 0.6 or so. If you look from 2000 onwards, in the last 25 years, it only has a Sharpe of something like 0.55. And that’s interesting.

People don’t appreciate it after this beautiful run that S&P has had, people don’t appreciate that even the mighty 60/40, it delivers returns for sure, but it also delivers volatility. And what happens as a result is that the Sharpe is sitting, say, let’s call it 0.6. And we have a very good track record for the CTAs.

So, CTAs, at the moment, are going through a very tough period, but you’re still looking at Sharpe, between say 2000 and 2025, of around 0.5. So, you have essentially two asset classes which by construction, as we discussed before, are uncorrelated to each other. And it’s not that the Sharpe between them, the difference between them is that great. We have say, let’s call it 0.7 and 0.4, or 0.6 and 0.5, depending on which period you are examining. But they are broadly sort of equivalent.

And because you do have the zero correlation, it’s very additive to add it. And it’s not surprising that Quantica finds that it will be additive to add it because even if you have an expectation, going forward, that the 60/40 will give you a Sharpe of 0.7 and the CTA will give you a Sharpe of 0.3, that actually means that you should have almost 30% of your risk in trend.

So, that’s actually very useful to be reminded because a lot of people are looking at very recent history. What have you done for me in the last year or two? And trend is going through a very tough period, no question about it. But it’s important to realize both the structural long-term benefits of trend and the zero correlation that we discussed earlier.

[00:36:43] Niels

Yeah, well said, well said.

All right, well, then we will move on to another paper which is also CTA focused and it’s from our good friend over at SocGen, who’s also a co-host on the podcast, Tom Wrobel, who came out with a report a few weeks ago, I think from memory. But you found a few things that you wanted to bring up. So why don’t you share with us what stood out to you in this paper?

[00:37:18] Yoav

So, SocGen has been at the forefront of CTA for a long time. So, they came up with the CTA index and the CTA Trend index that we all follow and enjoy. But they also came up with the trend indicator that they have, which is actually a very simple implementation of a very simple trend, you know, crossover, plus one, minus one, indicator. And that has been essentially out of sample for, you know, the last 15 years.

And the paper is that they’re celebrating 15 years of their involvement in the CTA industry. And they are doing a comparison of like what has happened to the constituents of the SocGen index over those years. And they’re trying to find sort of certain patterns.

And one of the nice things that they look at is that we’ve become a little bit slower. So, the industry has shifted a little bit to the slow side. And the way that they do that is by checking the correlation between the very simple indicator to the run at a different speeds, versus the SocGen index. And they’re seeing a very high correlation.

And I think it’s almost remarkable that the correlations that they see are so high with something which is essentially a very simple, straightforward, very easy to explain indicator. They’re seeing dispersion obviously during certain crisis points. So, during periods when there’s high volatility. I suspect April would have been one. Covid would have been one. Taper Tantrum would have been one. You go back historically and in those periods there is some decorrelation between the constituents and the indicator but that reverts fairly quickly. So, it goes back to very high correlation.

And I found it very informative. They’re not trying to do anything which is not in the data. They’re not trying to make any claims which are above and beyond what they’re saying. And they know the industry. And it is actually the case that we can see the slowness of the trend following industry in the SocGen index. We can see that in the correlation between say the COT report and positioning and the industry. So, we can, you know, correlate the COT report with different trend indicators, and we can see ourselves getting slower. So, I think that’s very worthwhile.

I think the interesting thing for an investor is the fact that the correlations we’re talking about are very, very high which, to me, almost sort of calls into question what are the people trying to replicate the SocGen index doing? And I find that very interesting because if you look at, you know, there have been papers about trying to replicate the SocGen index and we get that you can do bottom-up approach, you can do top-down approach, you can do a mixture of the two. There are quite a few papers on that. And I think the SocGen indicator puts them a little bit to shame because, without doing anything, it is 80% correlated to the SocGen index.

And I think that’s actually interesting that I think replication, if you’re coming from the spider replicating the S&P, you will have a 99.99% correlation between spider and S&P. But when people are talking about replication within our industry at the moment, one should realize that actually picking almost any CTA you will get that same amount of replication. It’s not a huge… It’s very difficult to get above 90% but it’s also very difficult to get below 80%.

So that I thought was very interesting from that paper. It’s again solid ground and it just reminds you about what the shape of the industry is like.

[00:41:43] Niels

Now you pick up on something I wasn’t expecting. So that’s good. And that is, you talk about, you know, what are people trying to replicate, etc., etc. Clearly this indicator (and Tom maybe can speak to this next time he’s on) is trying to replicate, or not replicate, but trying to show a trend following type of return stream. I was just looking at their website right now and I’m noticing that since October of 2002, so more than 20 years, the annualized return is 0.43%.

[00:42:22] Yoav

Yes.

[00:42:23] Niels

With an annualized volatility of 16, a Sharpe of 0.026. So, it may be correlated, but it hasn’t performed very well.

[00:42:35] Yoav

So, that’s the key. So, what we do in a CTA is to say can we up the Sharpe in each of the markets that we trade just a tiny bit? Be it with better execution, be it with some nonlinear filters, be it with some clever switching between fast and slow, be it from whatever way we do it, we try to make sure that we get a better performance. And indeed, we do. So, if you look at the SocGen CTA index, we get a Sharpe of say 0.4 historically. And that’s out of sample.

If you pick lots of bread and butter… like I’m sure Dunn Capital’s, Sharpe is much higher. So, if you pick a good manager, you can get more than that. But even if you just take the average, you get 0.4.

So, if you’re a CTA, if you’re pitching correlation as what you are trying to replicate, then that’s not enough. Correlation is when you remove the mean. What we’re really thinking about is how can we improve the mean of your overall fund vis a vis, as you say, the very highly correlated but with zero alpha sort of trend indicator that SocGen has been running for 20 years? And you know, you can do that in terms of cost reduction, you can do that in terms of, I mean the SocGen index is, of course, with costs.

So, you can do it by cost reduction, you can do it just with better signals, better risk management. All of those things of course come to play. And that’s actually what we spend most of our life doing (working life anyway), trying to do.

So absolutely, I think, but I just want to shift the discussion. When people talk about replication, I think that the discussion should be shifted towards, yes, but where’s the added value of the manager that you are picking rather than what correlation you happen to be with the SocGen index because it’s very easy to get to the 80%, 90% correlation to the SocGen index.

[00:44:44] Niels

Yeah, absolutely great point. Now, that leaves us with one paper, I was just going to say, to talk about, but it’s mostly going to be you talking about it. But it’s a good one and it is another friend of the podcast, Artur Sepp, and it’s called The Science and Practice of Trend Following Systems, and it’s just been published, as far as I can tell from your…

[00:45:14] Yoav

Just went on SSRN, just today, just this morning.

[00:45:17] Niels

Exactly. So, this is hot off, off the press for sure. And, and I’m going to try and follow along and hopefully come up with some commentary or questions along the way. But this is something that you have looked at much more closely than I have. So over to you, Yoav, and let’s hear about this interesting paper.

[00:45:38] Yoav

Yes, yes, definitely. So, Artur actually presented the paper in the London Quant Group meeting about two months ago and it was actually interesting, it was really interesting. And he gave me a copy of the presentation and then a copy of the early draft of the paper.

And it’s feature packed. It’s like packed full of stuff which is some of it is historical, some of it is mathematical, but it’s a real introduction to trend following. And let’s take it one at a time.

I think that one of the nice things is history and diversity in terms of the different types of people and different strategies that I run. So, for those who don’t know, Turtle Trading in the US started in sort of the 1980s, early 1980s, and that was very much rules-based trading.

You open a position when there’s a breakout. You wait. You have a trailing stop, maybe using average true range (which is essentially a volatility measure), and then you shift to being short, and then again you have a trailing stop, and so forth, and so forth, and that has been very successful.

There are a few books on turtle trading because the guy taught Turtle Trading to some, you know, a bunch of people, and then they went off and did their own thing. It gave them money and within sort of a year they came back with like tons of money that they made.

It’s been very successful in the US. And, concurrently in Europe, you had the AHL, the European school of trend following, which is a little bit more continuous change of position. It is based on moving average crossover, and some volatility adjustment, and risk adjustment, and that has been very successful as we look at AHL and the grandchildren – Aspect, and Systematica, and Winton, and of course the children, then there’s grandchildren – funds, everybody’s on the game.

And I think I’m going to start with a story because there’s the third school that Artur will describe, which is TSMOM, Time Series Momentum, which is championed by AQR, actually. So, that’s at Moskowitz and Pedersen paper in 2012 or 2011. And at the time I was in Winton and in AHL and we were not allowed to speak to clients just in case we reveal the magic sauce of trend following.

I remember Anthony Daniels was head of the client relationship in Winton and he said, oh God, we mustn’t tell them exactly what we’re doing because, you know, it’s a little bit like stirrups on a horse. Once you tell them, they say, oh, that’s completely simple, completely trivial, and they’re not going to want to pay for us.

[00:48:31] Niels

By the way, I’m sure many managers would have said that to their people, do not spill the secret sauce. Not just…

[00:48:40] Yoav

Everybody, I mean, trend following was pure alpha. And it was just after 2008, we made tons of money while everybody was losing money. It’s like, of course, you should be protective on your idea.

[00:48:53] Niels

Yeah.

[00:48:53] Yoav

And then I remember I was sitting in, I think it was an imperial lecture hall theater, and it was a conference. And Pedersen came, and he gave the talk, and I thought to myself, my God is like, you know, he’s opening the blinds, he’s letting the sun in. He’s just giving away the crown jewels for nothing. What is he doing? It’s going to be the destruction of the industry. And indeed, actually, in some sense it was.

Following that period, we became a lot more transparent and it became a lot more commonplace, the implementation. And a lot of people are kind of very reluctant to pay for trend following despite the fact that it’s very much one of the few pure alpha strategies.

So, TSMOM is actually a very simple implementation. It’s just like, let’s look at a year-on-year returns, adjust by volatility, and that’s what we’re going to do. And I think one of the things that Artur does very nicely is to summarize the three strategies, the three types of strategies. And then he goes on and he says, and how have they done?

And he scans the different speeds and different rules that you can apply to the American trend following, to the European trend following, and to the TSMOM school of thought. And what you find is that, in most speeds and in most directions, they would have done very well.

So over the last 25 years, it really didn’t matter which particular implementation you did. Some are more consistent, some are less consistent, but actually most of them have done well.

And I think that’s a testament to the fact that trend following, as a behavioral sort of phenomenon, is prevalent at almost all speeds and all asset classes. So, it really doesn’t matter whether you’re looking at commodities, whether you’re looking at equities, whether you’re looking at bonds. The nature of people to want to buy winners and to sell losers is something which is very prevalent. So, that’s one of the nice things that Artur does.

But what Artur does next is actually very exciting to me, is to say, well actually, can we think about what’s the optimal way? What speed should you run at? And I think that, do you remember, we talked about what are the optimal indicators when we talked last time on TTU?

And what he says, well, okay (let’s concentrate on the European case), the returns really depend on how you are able to capture two effects. The first one is auto correlation, and the second thing is drift.

And he says, well, if you know that the data is of a certain family of processes, time series processes, suppose you knew it was just white noise, or suppose you knew it was just autocall – what’s called AR1 (an ARIMA process which most of the correlation is between today and yesterday), or suppose we looked at a fractional ARIMA – so, it’s something which has a long-term autocorrelation. What would be the optimal speed to run at?

And he does the maths and he comes up with the numbers. And I find that really, really useful. Really, really, really, really interesting. And I’m going to speak about why this is useful.

Even though in some sense it’s converting one problem to another. It’s saying, well if I knew that the underlying process was X, I knew what speed to run. But the problem is that we don’t know what X is, right? So, in some sense, you might think it doesn’t do anything. But I actually think this is really important for the industry.

And then the last thing he does is talk about, again, the combination, about how you would combine this with a 60/40 portfolio. And he talks about the improvement to Sharpe, both in overall Sharpe but also in times of what’s called bear Sharpe – periods where equity has not done well.

And he looks at the portfolio, and funnily enough, he comes to a very similar conclusion as the Quantica paper, which is, given that over the period that they are looking at the performance, the raw Sharpe of the 60/40 and the raw Sharpe of the of the trend following strategies are actually ballpark similar and the correlation, of course, is zero. It’s actually very easy to come up with where the efficient frontier is, and how to combine things.

So, I really enjoyed that paper. I really, really enjoyed that paper. I wrote to him, I gave him, really, thumbs up on this one. And I’ll explain why I think this is important. Why is it important to have a theoretical result with what is the optimal speed?

And I think it comes to what we do when we backtest a strategy? So, one of the best tools that we have when we backtest and we stress test a strategy is to have a benchmark to say, given something, we will expect something else. And we know what to expect. If only the conditions were slightly different, we know what to expect. And if our strategy doesn’t follow that pattern, then it’s not tracking the benchmark the way we expected to track it, then there is an issue.

So, here’s an example, like a simple example. So, when I test a strategy, I always… Obviously you lag it to see the alpha decay, but I actually lag it in a negative fashion. I say, what happens if the strategy knows the data ahead of time? How much money are you expected to make?

And we actually have a very good intuition about, for example, if a trend following strategy, if it sees the future, how much Sharpe it should be able to pick up. If my strategy is not doing it, something is wrong.

Of course, I can’t run it in reality sort of ahead of time, but it’s still a very useful tool to me as a benchmark. There’s a new mechanism. Suppose if I’m feeding it noise, if I’m feeding it noise, then I expect it to make no money.

So, if I run my strategy, and suddenly I have a host of strategies that happen to be able to make money, then something is wrong. My strategy is overfitted in some weird fashion that I haven’t really understood what is going on here?

That, by the way, it’s a new technology, it’s called Rademacher Antiserum. It’s a new backtesting technology.

And the way I see this paper’s contribution, if I look at Artur Sepp’s contributions, then if we knew this, I can simulate some data. I can simulate the processes that we wanted to look at and now we understand how much I am expected to make with the best possible implementation. And now I can say to myself, oh, here’s my new idea – my new idea that is supposed to be a better trend indicator.

Is it up to par with that implementation? Is it able to pick up the fact that the data is behaving like that? So having a benchmark, which you know what it is, and you know what to expect, and you know what the performance should be, is actually very useful to testing new ideas without actually resorting to burning your data.

If I knew what the process was, I would expect X. Let’s see how my trend following is doing with that X. And that’s actually, I think, very important and I think it’s a very, very important addition to the trend following community.

[00:57:10] Niels

Having read it, how should people think about this thing about time horizon lookback periods, however we call it. Because from what I can see, based on the research that I have access to, there’s a massive difference between say a 20 day lookback and a 130 day or 260 day lookback.

You know, in a perfect world we would know when to use 20 days, when to use 60 days, 260 days, whatever. And to some extent maybe some managers, obviously you can just spread it out and say, well I’ve used 20 different lookback periods, I weight them equally. But of course ideally you would not just weight them equally if you knew.

So, I guess what I’m trying to get at is, is there any way, in your opinion, to come up with a framework that actually, with real success, meaning something that’s been done in real life (I don’t know if you’ve done this) whereby you can tilt the lookback and actually achieve a better result over time instead of just kind of I’m giving each lookback an equal amount of risk? Does that exist or is it where it gets too over-engineered and it only works in a backtest?

[00:58:40] Yoav

Okay, so let’s take with the simple stuff. So, the simple stuff is that if you don’t know what you’re doing. I think the TSMOM, just look at one year returns is absolutely fine and I’ll explain why. So, the most important innovation that we’ve seen in the past is actually the vol estimation. The fact that we risk manage things properly is actually very powerful.

That change, moving into the risk space, thinking in terms of the problem, in terms of the risk returns rather than actual returns, that has been the big innovation. If we think about the two aspects, the first one is the drift and the second one is the autocorrelation. And it only makes sense to tilt from a unit. I mean, if you think the TSMOM, which is just one year-on-year return, basically gives equal weight to all the returns in the year. And that is actually very effective at capturing long-term drift. It’s very, very effective. And this is one of the reasons why the TSMOM historically has done particularly well.

The only reason to change from that is if you have a very strong view about the tilt of the autocorrelation to you tilt the weights to match the autocorrelation. But even if the auto correlations are 0, or the autocorrelations are constant across the year, then TSMON just a year-on-year returns will be optimal.

So, in many senses, if you’re not looking for anything too complicated, that’s absolutely fine, and you should stop there, and it will actually do you very well. It is a testament to the trend following in general just working, you know, almost out of the box.

I think part of the reason why you might want to tilt between fast and slow predictors is not necessarily to do with what they do on a standalone basis, but what they do in your portfolio in terms of diversification. So, like when we run trend following and we happen to have a market with a trend in the opposite direction, when I look at my risk it actually gives me negative correlation. So, we’re coming back to this negative correlation problem.

If I have a trend following signal which is actually negatively correlated to the rest of my portfolio, that’s wonderful. That’s great. That’s showing me, ah, that’s something I want to allocate to. And I think a lot of CTAs don’t necessarily do that. Looking at their actual risk profile right now and thinking, oh, maybe we have a market which is trading differently, we want to over allocate to it.

Similarly, in terms of the signal space, you might have a situation where overall long-term I have the same Sharpe expectation from slow, fast, mid speed trend but right now we have a signal which is diversifying. You know, the market, one of the marketers has trended. Everybody’s going happy, and oil (let’s call it oil for topical example), oil seems to be switching very rapidly. So, you might say, I actually want to over allocate to fast momentum now. Not because I necessarily think that fast momentum has a higher alpha, maybe they all have the same alpha, but it actually will be diversifying.

In my portfolio I have a collection of very slow trend following, but maybe now it’s a time to trend in one of those markets. It might be beneficial to trend fast. So, it’s not just standalone, it is actually when you come to build a portfolio, you might want to allocate differently in order to get sort of the maximum diversification in your portfolio. So, it’s not a single market argument at all.

But again, coming back to what’s optimal on a standalone basis, I think TSMON, 12 months, year-on-year, it’s beautiful. It’s beautiful and you can show that it’s actually optimal in so many situations.

It doesn’t have that much of a positive skew because it’s quite a slow trend following. So that’s the flip side. If you look at its backtest, it will have a relatively negative skew, but from an alpha perspective, it’s actually very nice.

[01:03:14] Niels

Yeah. I mean what’s interesting about it is that, and again, there are obviously different ways of looking at these things, but because we are fortunate to have friends in the industry, and who all do something slightly different. Some are from the American school, and by the way, interestingly enough, before the price breakout that Richard Dennis brought along with the Turtles, a lot of the managers from the ‘70s, including Dunn, started out using volatility breakout, kind of a slightly different variation to what the paper looked at. And then of course the moving averages and the time series momentum certainly came later.

But what I found, just from looking at the hard facts, namely the performance reports, that they never lie. I think you is fair to say that a lot of them, over the very long run, often get to pretty much the same place.

So, I guess where I differ with some of my very good friends in the industry is that when you claim one is better than the other and criticize people for doing differently, I just don’t see the evidence of that in the public records of performance of managers. But be that as it may, we all have our preferences for sure.

And I think that’s the other thing that’s important, that whatever people gravitate to, whether they’re an allocator, whether they’re doing it themselves, it has to be something that you trust completely. So, that might tilt your decision one way or the other. Because if you’re not able to follow it through thick and thin, it certainly won’t work for you. That is for sure.

Yoav, I think we did a pretty good job going through a number of wonderful papers. We have to thank all the authors, of course.

[01:05:15] Yoav

Absolutely.

[01:05:16] Niels

Because they help us come up with …

[01:05:20] Yoav

Well, they did all the work.

[01:05:21] Niels

They did all the work and helped us come up with new content every single week on the same topic of trend following.

[01:05:27] Yoav

We are purely derivative.

[01:05:29] Niels

Right. Which is not always easy after all these years, but hopefully it was educational. I’ll try and get my producer to link to all these wonderful papers in the show notes so people can find them there, either on the website or if you get it on your on your mobile device, through a podcast platform.

So, anyways, this was fun. Thank you so much, Yoav. Glad that nothing took place, no sirens in the background during our conversation. I hope that you and your family will get home safely and that your parents will stay safe where they are and very much look forward to our next conversation in a few weeks.

[01:06:13] Yoav

Likewise.

[01:06:13] Niels

Yes.

[01:06:14] Yoav

Thank you very much.

[01:06:15] Niels

Sure. So, if you want to show your appreciation to Yoav and all the work he did for today’s conversation, head over to your favorite podcast platform, leave a rating and review. It certainly helps and it’s a good way to show your appreciation for the work that does go into these conversations, I can assure you.

Also, next week, which may or may not involve a new paper. I never know exactly what Katy has been up to, but she does write some great papers, so maybe she’ll bring another one along for our conversation next week. I’m pretty sure it’s Katy that’s coming on.

And following that, we have another person, a new co-host on the podcast that people will get very excited about, I’m sure, when they find out who it is and what they’ve been up to recently in terms of producing a paper.

So, I’ll keep that a little bit in suspense for now, but just say that if you have any questions for our weekly conversation, you can, of course, send them to info@toptradersunplugged.com and I’ll do my very best to remember to bring them up during the conversation.

From Yoav and me, thank you ever so much for listening. We look forward to being back with you next week. And in the meantime, perhaps more importantly than ever before, take care of yourself and take care of each other.

[01:07:36] Ending

Thanks for listening to the Systematic Investor podcast series. If you enjoy this series, go on over to iTunes and leave an honest rating and review. And be sure to listen to all the other episodes from Top Traders Unplugged.

If you have questions about systematic investing, send us an email with the word question in the subject line to info@toptradersunplugged.com and we’ll try to get it on the show.

And remember, all the discussion that we have about investment performance is about the past, and past performance does not guarantee or even infer anything about future performance. Also, understand that there is 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.