CTAs: The Unexpected Heroes of Market Crises 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.

[00:00:22] 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 market through the lens of a rules based investor. Yoav, great to be back with you this week. How are you doing? How are things where you are?

[00:00:38] Yoav

Not too bad, thank you. Great to be back on TTU. I’m actually on holiday at the moment. We are getting ready for Passover. So, Spring Clean is in full swing at the moment.

We are hosting about 20 people for sort of Saturday night and of course my in laws, both sets of in laws are here with us over the weekend. So, if you think there is stress in the market, think again, there’s much more stress around here. So, that’s my life at the moment.

[00:01:04] Niels

That’s your life at the moment. Fair enough. Yeah, absolutely.

I think we’ll have to talk a little bit about what’s going on at the moment and we’ve got a great lineup of some papers we’re going to be talking about as well that you brought along. So quite a lot to tackle this week.

But of course, before we do all of that, I’m always curious to hear what people have been thinking about, what’s been kind of crossing their radar. So, what’s been on your radar the last couple of weeks?

[00:01:36] Yoav

So, the market’s being so quiet. I’ve taken the time to read a little bit about MCP, which is a new, essentially, protocol, an API for machine learning. So, this has been, actually, gathering a lot of strength in the industry and there’s a lot of development there. So, I think machine learning and being useful is becoming sort of much more industrialized.

So MCP is a protocol that allows ChatGPT or other machine learning to speak to a website and that allows you to create applications, essentially straight out, via ChatGPT. And I think that’s a really nice development if you’re interested in that sort of thing.

Have a look at base44.com that’s a new app that allows you to essentially create apps on your own. It’s really impressive technology. It’s actually becoming very useful.

[00:02:32] Niels

Yeah. I’m not sure how I feel about all this AI, whether I should just embrace it and go all in or whether I should be a little bit worried about it.

So, I’m not full in on this just yet, I have to say.

[00:02:45] Yoav

Yeah, we don’t use it for trading at the moment, but I think it’s important to keep abreast of what’s happening and it is actually becoming very powerful as a way of prototyping technology and really, really nice.

[00:02:58] Niels

Yeah, absolutely.

Now people might think that what has been on my radar is what’s going on in the stock markets, and of course it has had some influence on what I’ve been thinking about. But, actually, for our conversation, I think there’s something that’s probably even more important, oddly enough, even though everybody has been talking about the wild swings we’ve been seeing in equities. And they have been historic in many ways. But I also noticed that the long-term yields have been soaring and that the US 10-year note saw the biggest three day jump since 2001.

That is impressive at a time when you would think people would use that as a safe haven. Now, I’m not an expert in this area as much as you are, although I did start my career in trading government bonds. But I would love to hear kind of your perspective, your thoughts on this because I truly feel that this is very important for investors, especially because we have been, or investors I would say, to a large extent, have been relying on a certain correlation between stocks and bonds, especially during crisis periods. We saw in 2022 that you couldn’t completely rely on that.

We had been talking about this, actually, on the podcast for a while because we felt that it was the anomaly in the first two decades of this century, and that we would go back to more normal times. But, actually, even during the last couple of weeks worth of extreme market moves, to some extent, including yesterday (we were recording Thursday), including the historic pivot we saw in the markets when Trump announced a pause to the tariffs on most countries with the exception of China. Yeah, I mean maybe there was more to it than what meets the eye in terms of why that pivot came.

So, I’d love to hear your thoughts and your insights on all of this since you follow these markets much more than I do.

[00:05:16] Yoav

Yeah. So, it’s actually been fascinating the way that the bond market reacted, as well as the dollar for that matter. So, you’re absolutely correct about the usual response, that you kind of expected, running as a safe haven to the US treasuries. That certainly did not happen.

What we’re seeing is that the bond market in the US actually retreated, yields were spiking and the dollar, as well, the dollar retreated. And you saw quite a nuanced behavior across different countries. So, in Japan you saw unprecedented drop in yields, whereas in the US you saw unprecedented rise in yields. And in many ways that was more painful actually the long end at the 30 year, it touched the 5% briefly. And that means that somebody trying to remortgage their house will be facing an extra 50, 60 basis points of cost on the mortgage. And that’s very painful for American citizens.

I mean the important thing to remember is that the US has about 30% of its debt being owned by foreigners; by the likes of China, by the likes of Europe, used to be Japan, but a lot more China and Europe these days. And that means that if you start a trade war, this can quite happily escalate into a capital war.

So, we saw that in a statement in France saying, well, if these guys are going to war with us, why are we investing in their economy? And what we saw, we saw a route in the US bond market and there was a lot of doubt about the auction, the 10-year auction yesterday, on Wednesday, thankfully that actually went well.

But certainly, in terms of long-term, the US needs to refinance or to re-auction about 2 trillion worth of debt every year and they will need to do that essentially for the next 10 years. So that means that they rely on foreign support for the bond market, for the US bond market.

I think the route that we saw in the bond market, in the US, was, actually, potentially much more influential in causing the pause in the tariffs that we saw because you can cope with temporary reduction in sort of a price of goods, in the price of equity, but yields, the bond market collapse was actually very worrying. Luckily that hasn’t happened and we are probably getting back to normality, hopefully.

[00:08:03] Niels

Now we talked about yields spiking and all of that stuff. Now I didn’t follow it that deeply in terms of liquidity. Did you know if liquidity actually kind of dried up, and even to the point where we as managers in the futures markets would have felt some change?

[00:08:29] Yoav

So, the US market stayed fairly liquid. So, I don’t think there was a huge issue there. If you go to the emerging market, it was certainly the case that spreads were all over the place.

It’s been very tough. South Africa, for example, was affected by local news as well. And that was very tough to trade. I’ve got to say this week has been remarkable in many, many ways. And it’s certainly just tough to trade and tough to see the effects.

And I think the bond market and the equity market correlation is something that investors should understand that when there is concerns about inflation then you actually get positive correlation. And that really has implications in terms of your portfolio construction because if you’re relying on the 60/40 to essentially have negative correlation, that is not always the case anymore.

[00:09:23] Niels

Yeah, but as you say, I mean even within the fixed income and even within developed markets fixed income, to me it seems like there’s a lot of divergence within the signals that we get as trend followers right now compared to what we normally see and certainly also between the long end and the short end. And so, it is interesting times to say the least.

The other thing I think it opens up for, when you see something like this (and we’ll see what the fallout is and how investors might rethink their strategies), from memory (I think we’ve touched on it a little bit on the podcast in more recent years, but without fleshing it out completely), and that is this big question, what are safe haven assets today?

I mean, is this also changing? Right now people might say well, actually, it looks like it could be gold. That’s the safe haven. I even wonder (and of course completely biased here), but I even wonder if trend following could be “a safe asset” because of the adaptability.

And of course, yeah, we’ve taken some pain the last few days, but nothing out of the ordinary compared to you know, what we normally see in these types of periods. And it has been pretty extreme, certainly in terms of the speed of these changes.

So, I really think that, in large, investors need to rethink what they classify as a “risk free“ asset for them to hold significant levels of investments in.

[00:11:06] Yoav

So, I mean gold, this time, outperformed other commodities. So, it went down just a little bit, but actually was very good. I mean I was very thankful that it wasn’t crypto. That would have been insane as a safe haven. In terms of long-term allocation, it’s definitely the case that CTAs offer a much more diversified portfolio.

And in the presence of inflation, when you have one factor which is driving the correlation for all the other asset classes, so commodities, equities and bonds, you do need a set of more diversifying strategies, and CTAs certainly offer one of those paths to diversification.

[00:11:49] Niels

Yeah. Well, let’s stay with the CTA and the trend following space as we normally do at this stage to kind of jump in. And, for me, it was kind of hard to prepare for our conversation today because there’s so much going on.

Sort of just this section where we normally talk about where managers might have made money or lost money, whatever I’m going to say now, it’s probably not very accurate because I think this time around, because of the way it’s played out, the speed at which it’s played out, the part of the portfolio it’s been playing out in, I would expect that managers who are normally fairly highly correlated will have seen very different results in the last few weeks. Now if we look at a portfolio of markets, where is it likely that trend followers and CTAs would have lost money? I’m talking mostly, I should say, from a trend following perspective.

Well, I think it’s fair to say that it’s probably equities that have been hurting in the portfolios, maybe with a combination of currencies, as I suspect a lot of longer-term managers were along the dollar going into the last few weeks. So, I would expect that those two areas have been challenging.

Fixed income could be kind of anywhere, so to speak. Although I do think, on balance, it’s probably a slightly positive part of the portfolio, albeit, as you said, Japan saw some extreme moves in their interest rates, bond yields. So that most likely will have caught some trend following models on the wrong foot. But the short end, in particular, probably did okay for managers.

If there’s one thing where I think there will have been some bright spots in most trend following portfolios, it’s actually energies. Now, again, it’s a little bit tricky because energies actually traded higher during the month of March.

So, if you were a little bit too short-term, you may have been exiting your short positions or even flipping long. And then you get the big sell-off coming when tariffs were announced. But I think, if you’re a longer-term manager, that’s probably the area that’s given you some offset in your portfolio.

Yeah, grains, softs are probably not where people have seen the most action, so to speak, in terms of P and L. But then you get into the metals area and here I think again, very much depending on time, speed of models, you could have had a very different outcome.

But I do think things like gold and silver, interestingly enough for the month of April, not necessarily great until we get some bounce back, which may be happening now. But, of course they have done well leading into April. So, a very interesting period to look at.

Now, I know you trade just fixed income in your portfolio. I’d love to hear what your view is from a trend following perspective within fixed income because, as we started out by talking about it, this has been a very, very interesting part of the financial markets right now.

[00:15:15] Yoav

Yeah, absolutely. I think one area of fixed income which has definitely been very painful for us has been credit. So, not just government bonds, but if we look at corporate debt, that really got hit as well. And, in fact, the hit was delayed. Equity started trending lower earlier. Credit held fairly well for a while and then in the crisis really, really didn’t do particularly well.

[00:15:41] Niels

Do you trade that from a trend following perspective?

[00:15:44] Yoav

We trade it from a trend following perspective. So, that has been painful and that is, essentially, an equity beta. So that’s a large part of it.

I think the different countries reacted very differently to the crisis. So, Australia and New Zealand, for example, bonds retreated, yields were on the rise. Japan was exactly the opposite direction, Europe as well.

So, what we’re getting is actually an interesting dispersion amongst different countries. We saw a lot of EM countries really suffering. And, in fact, the more you traded with the US the more you got hurt by the tariffs.

But even countries like Mexico and Brazil, in principle, weren’t really that affected. Mexico was really excluded. Brazil only got about 10% tariff, so you might have thought it would have traded nicely. But actually, both currencies retreated very aggressively.

So, it was very strange. Not always driven necessarily just by what the tariff was. But we saw a huge dispersion in terms of behavior across the fixed income and the FX space.

[00:17:09] Niels

I imagine it must be very hard to hide from what’s going on when even islands that are occupied by penguins are part of the tariff war. So, there’s absolutely no place to hide right now.

Okay, so, on a serious note, my own trend barometer, interestingly enough, finished last night at 50 and has actually been pretty strong. And so I’m very curious to see how the shorter-term managers are doing because the trend barometer has some much shorter time frames built into it compared to most long-term trend followers.

So, they might actually have been okay except for what happened last night. I think that could be a really tricky one for a short-term manager because you probably would have been short equities again, and so on, and so forth. Maybe even also in the energy markets. So, we’ll see about that.

When I looked at the performance numbers, so this would have been as of Tuesday night, I saw that The BTOP 50 index was down for April 4.76%, and down 4.5% so far this year.

Now that’s a big number. 4.76%. That’s a big number for that index. Also, in an historical context, interesting to see what the bounce is from yesterday because I do think yesterday was probably, on balance, a little bit of a positive when I look at the managers that I can have access to.

SocGen CTA Index down 5.53 for the month so far, as of Tuesday, down almost 8% for the year.

Again, a big number but it doesn’t mean this is where we’re going to finish the month, but so far it’s definitely showing up.

The SocGen Trend index down 5.89%, and now down about 10% for the year. And the Short -Term Traders Index, as I mentioned, certainly doing better, down about 69 basis points, down 71 basis points so far this year. So doing a lot better, as you would expect, for sure.

[00:19:15] Yoav

I think the CTA index is not going to bounce that much. I think one of the problems for us as CTA is that we manage the risk very aggressively. So, certainly I would have expected a lot of the positions to have been cut. So, that sort of volatility where we have a large sharp down move will be followed by CTAs across the board cutting their risk. And that means that they will not be benefiting as much from the recovery that we saw yesterday.

[00:19:47] Niels

I agree with that. I agree with that view. We’ll see. And I’ll just add to that because you’re absolutely right. But actually, you know what might be cutting position size even quicker than the fact that markets are changing direction is actually the expansion or the explosion of volatility. I think that’s going to be really kicking in these days. Yeah, yeah, yeah, for sure.

MSCI World index down 4.52 as of last night, down 6.5 or so for the year. The S&P US Aggregate Bond index, I kind of flip around with different indices every week. I’m not really wedded to any one of them. Here’s one. You might have some better ones I should hear from you, but that’s down about 97 basis points for the month, but still up 1.63% for the year.

What, in your mind, is something that actually reports daily data on the Internet everyone can look at? Do you have like a favorite bond index that you look at?

[00:20:48] Yoav

I’ve got to say, I’ve got to come back to you because I normally look at the Bloomberg indices, myself. So, I’m not sure exactly what’s available on the web.

[00:20:57] Niels

I used to look at the World Government Bond Index but they stopped publishing that free-of-charge, so, I had to find another one.

Anyways, S&P 500 Total Return index down 2.73 as of last night. Down 6.89% for the year. But listen to this: up 9.52% yesterday. That’s the third biggest one day move since 1950.

[00:21:26] Yoav

Yeah, it’s unprecedented. Really unprecedented.

[00:21:28] Niels

It is unprecedented.

And so, in a sense, we are writing history at the moment.

Anyways, I want to stay with trend for a little bit longer because there will be people out there listening to us and saying, hmm, I knew it, trend following has stopped working. It should have made money. It’s a crisis alpha strategy, and so on, and so forth. Luckily, we have Katy Kaminski coming next week to dive into some new stuff on that.

But in the meantime, I did notice on Twitter, and I hope I’m contributing these statistics to the right guy, but I think it was a guy called Tyler Loving as far as I’m aware. If it’s not, I apologize. This is a little while ago, but he had looked at The SocGen CTA index performance through the first 10% of an S&P drawdown.

Okay, now the first 10%, it’s a little while ago, but it’s not, you know, that long ago we hit that mark. In fact, we hit it on the 11th of March. So, the first 10% was from the 19th of February to the 11th of March, 10% down on the S&P. At that time The SocGen CTA index was down 3.56%.

So actually, that’s not too bad.

Now, he does a great job. He goes through, I think there’s 10 of these events since year 2000 where the index has been down by 10% and he looks at how the SocGen CTA performance has been. What’s very clear is that when it happens very quickly, CTAs are struggling the most.

The two worst occurrences, or the two worst periods of a 10% drawdown in the S&P, in terms of SocGen CTA performance, was Volmageddon in 2018, February of 2018, and also what took place just before, or maybe it was during the great financial crisis, so, back in July through mid August 2007.

Those are the worst ones. But if you look at the average SocGen CTA index performance during a 10% drawdown, and as I said, I think there’s like 10 of them that he featured, the average is down 3.18% and the median down 2.81%. So, in a sense, you could say, well, that’s actually not bad because (and this is super important) as some papers have highlighted a couple of years ago by, the name escapes me, I think it was Mick Cater, the consultants that did the report, I’m pretty sure they were the ones who brought it out.

They talked about first and second responders during crisis periods, and made it very clear that people should not expect CTAs or trend followers to be the first responders. We’re not the ones who can deliver the first positive offset when a crisis begins. It comes later.

So, what I really liked about what, as I said, I believe this guy Tyler did was he then went on to say, well, let’s look at the performance when the S&P index goes from -10% to -20%, so, the next part of the leg.

Now, there are actually not that many times in the last 20 years where that’s happened but there are four that he highlights. What’s interesting, so in 2022 this happened and the performance of the SocGen CTA index during that -10% to -20% of the S&P was actually positive 20.38%. So, great. That’s exactly what we would expect.

In 2020, there was a relatively quick period, late February to mid January, obviously Covid, where it went from 10% to -20%. During that period of time the index was pretty flat, down 89 basis points. So, no big deal. But you know, there was no positive performance.

In 2007 through July ’08 it happened again. During that period of time the index was up almost 8%, so, it did exactly what we would expect it to do. And then in 2000, this is part of the tech bubble, it happened between, I think, April of 2000 and February of 2001, and the index was up almost 11%.

So, I think what Tyler managed to do was to perfectly describe, and kind of confirm, what you should expect from a trend following CTA during a crisis period because it all comes down to the speed of the crisis. We know that the positioning of a trend follower, when an equity crisis starts, it often starts actually from an all-time high. It really does. And at that stage we know that we’re going to be super long equities because that’s what you have to be as a trend follower.

And so, positioning in general, not just in equities but in other sectors is incredibly important in the initial stage because there’s no way we can adapt ahead of time, so to speak. And I think his little analysis shows that.

I’d love to hear your thoughts on this. I know you write a lot on your LinkedIn profile. I don’t know if you’ve ever written about this.

[00:27:11] Yoav

So, it’s right, what you say, in terms of CTAs not being the first responders, but it’s not just the speed of the crisis, it’s also the speeds of CTAs. So, in the early 2000s, I think CTAs were generally trading a little bit faster and they have migrated the bulk of the money. So, the SocGen index has migrated to be trading slower.

So, that’s one of the reasons why I think the SocGen CTA has suffered is because although we saw equities retreating in February, and maybe a faster manager would have switched position by then, a lot of the CTAs still had residual long position in equities going into the crisis. It’s not just speed of the crisis, but CTAs have migrated to the slightly slower speed month, you know, month to three month horizon trend following.

And in that speed, you don’t necessarily expect to turn over your position after the 10% drawdown, especially after like hitting an all-time high in February. I think what it highlights is, really for us as practitioners, the importance of diversification.

So, the way that you construct your portfolio is you have to be aware that although CTAs have positive convexity on a 1 month, 2 month, 3 month horizon, they have negative skew. We tend to own negative skew positions in all of our asset classes at any point in time.

And when a day one crisis hits, that’s actually when you suffer. You have your position and you have to live through that position. No amount of adjustment will get rid of it. So, the trick is actually creating a diversified portfolio so that, when that crisis happens, you just get hit in one part of your portfolio.

It’s actually very difficult in a situation like what we saw here, where the crisis is, essentially, across the board, where it’s not an idiosyncratic challenge because some disaster, some political turmoil in South Africa, some political turmoil in Japan, or elections in some other part of the country. But when you have such an across-the-board crisis, where we see oil tanking, and you see that in equity, we see that in effects we see that in bonds, I think it’s actually very difficult to create a diversified portfolio at that point regardless of how many futures you trade.

But I think this, a crisis like this, really highlights the importance of very rigorous risk management, which actually CTAs are very decent about doing it, so that on day one you don’t get hit so badly.

[00:30:03] Niels

Yeah, I think the point about risk management is super relevant and I’m hopeful that we don’t see any big fallout of this because this is truly something we have not seen in our historical data. And whatever backtest you did, you probably would never have seen anything like this showing up.

So, it’s a good test, it’s a good stress test, once again, for these strategies and hopefully they’ll come out okay. There are many lessons, actually, we can learn from this. You talked about diversification, so it’ll be interesting to see, at the end of the month, are people who trade the 400 to 500 to 600 markets better? Have they done better or worse than the people with concentration, or more concentration in their portfolio?

Again, it’s going to come… Also one thing that I think we’ll segue into shortly is this discussion about maybe some niche portfolios that leave out some of the, perhaps, financial markets instead of having more commodity exposure. And once again, I know this is early days, we have no idea where this crisis is going to go, but it confirms a study that was done a while back where the company had looked at which sectors are actually more consistent in performing during equity crisis. And it was very clear from that paper that it was the commodities. Which is also why we, at Dunn, we put a lot of weight or the importance into having a large number of commodity markets relative in the portfolio because we do believe that they’re so important.

But again, once again we talked about it already. I mean things like gold and other metals have done okay doing this. Energies have done really well, from a trend following perspective, offsetting some of the fallout in the financial sector. So, again, there are new lessons to be learned for sure. But there are also old lessons to be confirmed during events like this.

So, it’s tough for many investors and that’s unfortunate of course, but it’s also super interesting from an intellectual point of view to go through this in real time and just see how portfolios react, and as you say, the speed of the models, and I wonder, even down to the type of model, I would imagine that some models by design are faster to react.

I don’t mean the parameters in terms of speed, but even in terms of whether you are more a continuous system or whether you’re more like a breakout type system, there can be some differences in that. So, it’s super interesting to monitor closely how this all shakes out.

[00:33:00] Yoav

Yeah, I don’t think we’re going to see a material difference between managers who trade 200 or 50 futures. I think the paper from Quantica, three weeks ago, that you discussed with Alan, it’s not really about the number of markets. It’s really about the correlation structure, and which markets are less correlated. And that’s, I think, very important. The diversification for a CTA doesn’t come actually from the number of markets. We always say, square root of the number of markets, but that’s actually not where diversification comes from.

And as you say, being overweight in assets, which are less correlated to the rest of the market, is going to be much more material in terms of how this market, how this crisis played out, I think. So, if you were overweight in the commodities, that probably has done quite well for you in that respect.

What I think is interesting is that when you see a shift in correlation structure, that’s very painful, actually. So, one of the problems that you have is that if you think that you’re diversified to a certain amount, but then everything becomes more correlated very quickly, that is actually very painful. So that’s where we might see more pain.

[00:34:24] Niels

Yeah, yeah. There was something I was going to say to you, and I forgot it while you were talking, Yoav. Maybe it’ll come back to me.

But let’s jump to the first topic now. Again, this was written by one of our friends, of course, on the podcast, Moritz, something he posted on Substack. So, I don’t want to make it specific about his business, but I just want to make it specific about the topic that he raises because I do think it’s an interesting topic. And that’s this thing, that you also picked up on, and that is… Well, maybe I should let you explain what interested you about this particular article.

[00:34:58] Yoav

Oh, thanks. That’s a great Substack. So, if anybody wants to go into Two Quants Substack, I think they write really well and they write well about things that we don’t normally talk about when we talk about clients, and it’s about attention to details, which I think is really, really important in the CTA business.

They’re talking about two ways that they are getting diversification. And not in terms of position diversification, but in terms of the way that they trade, their trade diversification, the way that they can ensure that they trade not at the same time as other players in the market. Okay, so to put that into context, that’s not going to be a big impact to your portfolio P and L. We know that trading costs are not that great. We’re talking about five basis points of Sharpe per year, give or take. So, improving that is not going to make a huge difference to your performance, but it’s still very important.

And there’s talking about two ways in which the fund and which CTAs other CTAs can potentially reduce their market impact by mixing a collection of signals. That’s the first way that they are thinking about it. And the second way is they are saying, well, we might be trading the second contract rather than the first contract.

And both these methods are a nice way of saying let’s trade at a different time to other people or potentially in a slightly different market, but the one which is co-integrated with the first contract. So, we can trade in a different market so we are making a little bit less of a market impact.

And both methods are perfectly valid and it’s really good. But it’s actually the throwaway comments that they make which I find really interesting. Okay, so the first throwaway comment that they make is about the way that CTAs congregate during exit rather than entry. They say, entry is rather smooth, but exit, we all tend to exit at the same time. And you’re thinking about this and you think, hang on a second, why is that? I mean signals are signals, price is price. Why is that?

And I think that plays out to exactly the discussion that we had today. In reality, that correlation of trade happens for CTAs for two reasons. The first one is about the signal, the signal changing. So, if price goes up, we all tend to buy. But the second method, which is actually very significant, is about risk management. So, if volatility spikes then what you expect, you expect all CTAs that risk manage, if you have a position on, you will trade at the same time.

Of course, at the beginning of a trend you don’t have any risk on. Okay, so you don’t have correlation due to risk management because nobody has any risk. There’s no trend. So, there’s no correlation. The only correlation you see has to do with sort of entry signals.

But when we look at exit, by the time we have a long trend and everybody, all the CTAs are in position, then you see exit correlation not just to do with signal changing, but also, exactly as you said, to do with vol spiking. We risk manage, we take down the position that we take the time size, and that’s very interesting.

And we can actually measure market impact based on that. So, what I mean by that is, if I try to estimate volatility in the market, suppose we use 30 days volatility, historic volatility, to try to estimate tomorrow’s volatility. And suppose this is actually unbiased, this is exact if there is no trend. What we find is that if there is trend, if trend is overextended, this same measure will tend to underestimate volatility.

There is additional volatility tomorrow that we will see which we have underestimated because everybody had an extended trend and everybody is in that position. So, actually, you can measure CTA impact on the market by measuring the sort of unexpected volatility in the market that we see when trend is extended.

And that’s actually what we saw in the crisis. What I mean by that is the CTAs were in position. The crisis happens. We’ve all traded down our positions regardless of whichever CTA that you’re trading. It doesn’t matter.

[00:39:35] Niels

Can I interrupt you in your flow. Sorry to do that. I think that what you’re saying makes sense, but we can’t be sure that it’s the CTAs necessarily. Right? It could have been anyone who just uses some kind of…

[00:39:49] Yoav

Completely, it’s closeted. It’s. So, there are plenty of closet CTAs. Right? Closet trend followers. A lot of micro traders are closeted.

[00:39:56] Niels

Or just people who want to de-risk because things are going absolutely crazy.

[00:40:00] Yoav

Absolutely. But I’m looking at it from a mathematical point of view. When we use prediction of volatility, what we have to be aware of is actually the level of the trend also gives you quite a good extra 10% accuracy in estimating, 5% to 10% accuracy. And that’s another way for us to measure our impact in the market or the impact of whoever is trend following it in the market at any one time.

So, that’s one interesting throwaway comment that they make, which I think people don’t really appreciate in terms of correlation between CTAs. The second almost throwaway comment is that they look at the various performance of the trend indicators. They say, we’re going to use a mixture of those in order to ensure that we trade at different times. And what I find very interesting, they’re showing the performance dispersion between trend predictors and it’s huge. It’s somewhere between up 100% to up 400%.

So, what is very interesting is that although we are all trend followers, and we all have different styles, we tend to think, oh, it doesn’t really matter. But actually it does.

If you look at the average overall, if you actually put all those together and you mishmash them, then you get to a long-term average where we know what the expected performance is. You can actually do a mathematical formula for what that performance should be.

But what it does mean, it means that there is a big dispersion in which trend predictors you’re using. And if you’re just able to tilt it slightly to the predictor that is more appropriate at a given time or at a given market, then you can actually potentially improve your performance.

So that is one area of research which actually is very interesting in terms of, as you said, which predictor has done well. Are they breakout styles? Is it the speed? Is it the breakout style versus moving averages? So, there is scope, you know, they don’t make that point, but actually it’s very interesting in that paper.

[00:42:22] Niels

Okay, so I think a lot of people listening to us right now are thinking, well, I mean, how do you know which one to use because we don’t know the future?

[00:42:31] Yoav

Absolutely, so, part of it has to do with what are you trying to achieve?

Okay, so if you’re trying to achieve sort of long-term trend in the market, then it’s what are you trying to harvest, which alpha? And there are two sources of alpha in the market, the first one is sort of long-term trend. So, if you look at fixed income like long-term secular trends. And the second one is short-term autocorrelations between price moves.

So, a market can be trending a secular trend, which is very long-term and, actually, short-term correlation can be quite flat, and vis a vis some other markets which are driven by sort of successive autocorrelation in the short-term. And I think the design of what you use for which market really gets affected.

And the third component is of course costs. So, the way that you construct your portfolio and the way you construct the predictor really impacts the cost and cost of the market.

So, these three things, like what are you trying to harvest out of that market, are actually quite important.

[00:43:48] Niels

And I think that the final thing I want to maybe touch on, in terms of this write up, was I guess the question of also a little bit about capacity, meaning the type of markets you can put into a portfolio that is a few hundred million dollars instead of $5 or $10 billion dollars will make a difference.

And I guess the question becomes, and I think Moritz has kind of raised that, at least on things that I’ve come across, which I think is an interesting question and that is, in some way shape or form can capacity equate to some level of extra alpha? Meaning, if you decide that you’re not going to overstep your capacity, you’re not going to multi, multi billion dollar fund. Meaning that it gives you the ability to trade certain markets, commodity markets for example, at a full size instead of just saying it’s in your portfolio, but in reality it’s very little. Can that be somehow regarded as a little bit of extra alpha that you can get from these not multi billion dollar managers?

[00:45:01] Yoav

100%. 100%. And let me just explain. So, for a CTA, the average market Sharpe is important. But then you look at the diversification in the portfolio and we say diversification for a CTA means something very different to what it means in English. In English diversification means I’ve got 500 markets in my portfolio. But for us, what we care about is the reduction of volatility that you get by investing in multiple markets. i.e. If the diversification factor is three, by investing in 100 stocks rather than one stock you’re able to get your volatility down a factor of three. That bumps up your Sharpe by a factor of three. And that number is really, really crucial for us. So that’s why.

Because if we think about the average market Sharpe being, say, 25 basis points, if you get a factor of three, you get a 75 basis point Sharpe. This is how we do that. So having 500 markets or 50 markets is less important than having volatility reduction. That’s really what is key.

And what is interesting is that diversification does not come at all. It’s nothing to do with the square root of the number of markets. This is really funny. So normally you would come, and the way you would try to explain to an investor, you will come on and you will say, square root, if you had 100 markets and they were all diversified, you will get a diversification factor of 10, which is absolutely right. But it is actually not what is the dominant factor which affects our own diversification. And what is that? And what is the dominant factor? The dominant factor is actually the average market beta to the first component to the index.

Let me just explain. So, let’s take an example. Let me do it in stocks rather than in the CTA space. So, suppose I want to have $1 of risk in an equity, in a stock. And let’s suppose it has a 30% beta today index. So, I have $0.30 of risk in in the index and I have 95% in idiosyncratic risk. That works out together 95 squared plus 30 squared – that comes up to essentially $1 of risk.

So, I’ve got 95% of my risk is in idiosyncratic and $0.30 in the index. And now, suppose I’m investing in 100 stocks. Okay, now what is my loading on the index? My loading on the index is like $0.30 times 100.

Okay, so, I’ve got $30 of risk on the index. What is my idiosyncratic risk? Well, the idiosyncratic risk is the square root of 100, right? So, that is 10 times $.95. So that’s $9. So, after I’ve invested in 100 stocks, I’ve got $30 in the index and less than $10 in idiosyncratic risk. So, what is my risk dominated by? By my index.

Okay, the risk has reduced. I don’t have $100 of risk. I have $30 of risk. But it’s to do with the beta, the original beta of the stocks that I had to the index. So, if I’m a stock picker, I will get a much more diversified stock portfolio. If I’m able to pick only 20 stocks, but where the beta is only 10% to the index, it’s a much higher quality portfolio than just investing $1 in all 100 stocks. You’re much better off investing in only 10 stocks which are genuinely diversified, because the beta to the index is the thing which dominates, becomes very, very quickly dominating the sort of the reduction of volatility that we see.

And in our CTA universe, it’s exactly the same in the commodities. If we are able to allocate more meaningfully to markets which have low beta to the rest of the CTA universe, okay, so, as you said, Ags, metals, energies. Okay, energy is less so these days, but Ags and metals, then what you are able to do is you’re able to reduce your overall average market beta to the CTA index and that gives you a better quality portfolio, a much more diversified lower beta to the first factor.

[00:50:03] Niels

Now we had two more topics.

I think we may only get to one but it’s very fitting to what we’ve been talking about today because it’s regarding whether or not there is, more or less for that matter, dispersion among managers, at the moment, than what we’ve seen in the past. I think even I’ve made some comments saying well, there’s a lot of dispersion between returns at the moment. But obviously I never sat down and did the math and looked historically. Luckily, we have good friends in the industry who did that – not long ago, our friends down at CFM. And you had a look at that paper and maybe you can tell us a little bit about what they found and what you think about their findings and so on and so forth.

[00:50:52] Yoav

Yeah, absolutely. So, I like CFM. They are very much data driven. And it’s a short note, it says, everybody’s talking about dispersion of performance (and I think this month will probably show some dispersion). But actually, if you look at the SocGen index and you look at various…

[00:51:12] Niels

Is that the Trend index or the CTA index?

[00:51:14] Yoav

The CTA index. And (I hope I’m getting it right), if you look at the members (so they said we’re not picking the members, we’re just going to use the members that SocGen picked for us), and we look at the dispersion of performance, it is actually not that great. And if you look at pairwise, 6 month correlation, that hovers at around 50%. And by the way, this is not just them.

I think there was a paper from Goldman recently, about three months ago, that looked at dispersion between different managers. So, if you look at the macro managers, the pairwise correlation is maybe 30%. If you look at multi-strat, if you look at long/short equity, you see reasonably low correlation between different managers.

If you look at CTAs, Goldman came up with about 60% correlation, pairwise correlation between managers. And that’s like completely shocking. I mean, if you think about it, I’ve been in macro managers and they will concentrate in trading, maybe, the G10 universe, okay. And we go into huge lengths to trade, you know, 50, 100, 200, 400 different markets.

So, you might expect that we should be much more diversified internally to each other. Okay. And yet we end up very highly correlated. And just to put it into context, 50% correlation pairwise basically means that we all have 70% loading on the SocGen index. And that’s like a big number.

And that’s part of what I think is very much related to what we discussed just a second ago. Why are we getting such a high correlation between CTAs is because we all load on the same first factor as everybody else. So, it’s not just that you’re getting a more concentrated portfolio. You’re not necessarily getting a more diversification, you’re also becoming more closely to your peers.

And I think that’s part of the reasons why people don’t necessarily like to invest in CTAs because they’re kind of used to investing in multiple macro managers. But when it comes to CTAs, they will say, well, if I’ve talked to one, I’ve talked them all, they’re all the same.

And essentially, it’s our fault for making it because it’s true. We are 70% correlated to the index, we are 50% correlated to each other. And I think that discussion about trading off capacity to be able to allocate to markets which are slightly off the main path is helpful not just internally for you, in terms of getting a better quality, more internally diversified portfolio, but what it will also help you do is to get away from the rest of the pack in terms of being correlated to the SocGen index, to the other big managers. What am I offering in addition to sort of the standard CTAs? So, being slightly offbeat is very helpful.

And I think the tradeoff you will have to do is capacity. It’s exactly what the Quantica paper is talking about. Can you turn that into alpha? You first of all can turn it into alpha, but also it will make you more distinguishing and different to others. But you will have to content yourself that a certain market can take just so much risk. So, if you have only 10 stocks, if you’re a stock picker and you have 10 stocks which are truly diversifying, that’s great. But you have to put 10x of risk in each of the stocks. You can’t do $1 in each one of the $100, you have to put $10 in each of them. And that means that you have to start thinking about capacity and you have to have to basically run a smaller shop.

[00:55:34] Niels

Yeah, I think that those are all very important and relevant questions for people looking at this space. Certainly, if you’re a larger institution.

Now, just maybe to summarize all of this because you may have read the paper from CFM more closely than I did, but their conclusion is, as far as I remember, that actually there is no more dispersion today than what we’re used to seeing. Is that correctly remembered?

[00:56:01] Yoav

Yeah, I mean they go back all the way to 2011 or thereabouts, which is to do with the index, actually, if you go back historically. And that 50% correlation has stayed fairly stable. In some sense, that’s an indication that the industry has become a little bit more diversified because we’ve actually… You might have thought we would have converged by now, especially with the large AUM.

So, the industry has done work in order to essentially maintain. We’ve done a lot of work to stay in the same place. It’s like Alice in Wonderland. You have to run twice as fast just to make any sort of progress.

But yeah, that’s sort of 50% correlation pairwise. Average correlation has remained remarkably stable and that’s a testament to the fact that momentum is really a macro factor which can dominate a lot of your performance.

So, yeah, that to stay constant.

[00:57:13] Niels

Yeah, so, the main takeaway is, actually, that it’s not really dispersion that’s the story here. It’s the fact that the trend engine still works, so to speak.

[00:57:24] Yoav

Oh yeah. I mean trend works for many, many good reasons. And we see that in the autocorrelation of returns. It’s not just a way of transforming the payoff, but it also allows you to buy an option that’s at realized vol rather than implied vol and it allows you to harvest autocorrelation which we see in the data, you know, going back hundreds and hundreds of years.

[00:57:57] Niels

We’re coming up to the hour mark, so I would like to save the last topic for another day. But one thing I would like to maybe bring up and see if you’ve noticed this as well. And I’m not going to name any names here, but you know, investment banks, they love to… well, this is my quote, I guess, “front run” CTAs by coming up with these research announcements about, oh, but in the next two weeks CTAs have to buy $100 billion worth of equities, or they have to sell $100 billion worth of equities because so and so is happening. And some of these calls have been recently called out on Twitter where some of these investment banks have come out with these very firm statements.

But, of course, what turned out to happen was the complete opposite, thanks to the little tariff tantrum that we’ve been going through. So, every time they came out saying oh, they have to buy, in fact we’re probably selling because of markets, because of volatility, and so on, and so forth.

Now, one thing, maybe, as a final little challenge for you, Yoav, something that I’ve picked up in my own correspondence recently with investors and prospects, and that is I think people would have assumed, at this stage, that a lot of our equity signals would have been short by now. And, at least from my vantage point, I would have had to remind them by saying, well, you know, trend following is really about two things. It’s about the price, but it’s also about the time. And although prices have moved, time is not really long enough for these typical lookback periods. You mentioned it as well.

In the old days maybe we were a little bit shorter-term in our models, but nowadays we’re certainly more long-term. So, my expectations are that, for the most part, many long-term managers are not close to getting real short equity markets. Let’s leave out what happened last night. But even if it had stayed down 4% or 5% last night, it’s not really something that’s long enough in duration for our models to react to that.

What’s a good way, because I’m sure I’m not doing a great job at this, but what’s a good way to kind of explain to people this symbiotic relationship between time and price within the trend following strategy to help them better set expectations in terms of what our models, how they would react in what we’ve just gone through?

[01:00:38] Yoav

Absolutely. So, I think there are two effects here. First of all is time. But also, there is the fact that you started off at very much at an all-time high. What we’ve seen is, I mean, ridiculous performance of the S&P for a very long period of time. So, for any normal manager, the strength of the signals would have been many standard deviations from normality going into February. So, of course, you update your prior as the price starts trickling down. But yeah, I completely agree with you. I will have expected most of the CTA money to be very much still long the equity. But I think you explained it perfectly in terms of time. and in terms of the level, the signal where we started with.

I will say that a lot of those predictions by the investment banks about what CTAs will have to do, I think there is something in it. Obviously, we don’t know the future, but we do know the past. And what I mean by that is, if I look at a signal which is a moving average 30 days, then I know that in 15 days time the last 15 days will be in that sample. And that gives me a good value.

Like it’s not assuming… I don’t know what the market is going to do. But I actually have a very reasonable idea about where I’m going to… Like, if you’re a 30 day trend follower, I actually have a very good idea where you will be in expectation in 15 days’ time because 15 to 30 days will disappear but you are still left with the last 15 days of performance.

So actually, that’s something that the industry doesn’t do that much. But you can actually play forward our own positions, and especially as we become slower trend following. If suppose we’re a one month to three month horizon, you have a very good idea about where your signal is going to be in 15 days time, in two weeks time.

Obviously, I suspect the investment banks are not doing a very good job at predicting it because, for a start, they don’t really know exactly the models that we trade. But certainly from our perspective, if we design our trading, we should be able to think ahead and say, ooh, where are we likely to end up from a single perspective and can we get better cost based on that? Can we trade smoother into that path?

[01:03:23] Niels

But you know what? So, two comments to that, Yoav. One is I think it’s a dangerous game because it sounds like we’re now starting to predict something which we shouldn’t do, right? So, we’re kind of trying to predict where we might be instead of just saying, well, this is where we are today, and we don’t need to worry about where we might be in two weeks time. So, that’s kind of one thing about it. It doesn’t mean that it’s not happening. I’m just saying that I’m not a big fan of it.

And the other thing I’m not a big fan of is essentially people trying to front run our positions and trying to gain an advantage to themselves by encouraging their clients to trade, “ahead of CTAs”. Especially, as you say, they probably have some idea of how these models are built, but they don’t have the full picture.

[01:04:14] Yoav

I think first of all it’s being done, so it’s 100% being done. I was sitting in a macro shop, and the prop traders were interested in the CTA model, playing it forward. So, that’s not necessarily in order to front run the CTAs, but in order to understand the dynamics of the trade, to say how is this trade likely to play out? What happens if it’s like as a Monte Carlo simulation? What happens if price does go up? What happens if price goes down? Where is going to be the buying pressure at that point?

So, the first thing is people definitely do that. And we are in the market with everybody else and we’re in the business of making money. I think, from our perspective, thinking about where our signal is likely to be is not necessarily about trying to game ourselves or front run ourselves. There’s not a lot of risk you can put in that trading system. But in terms of being able to trade better into a certain position, how is that likely?

Let me just give you an example. Suppose our short-term trend has spiked through the roof. But you know, it’s coming back very quickly because it’s a short-term trend. And there is a trading cost of putting on that trade, because you have an expensive market, say. Then you might say, well, actually this signal is not going to stay positive long enough for me to recoup my trading costs. So actually, I don’t really want to put on that trade simply because it’s just becoming too expensive.

So, being aware of where you’re going to be is actually important, not for me in terms of predictive power, but in terms of knowing thyself and being able to incorporate costs in a better way into their own trading.

[01:06:12] Niels

Now, I don’t want to disagree with you Yoav, because you’re much smarter than I am. But when I hear that and I hear the words, like a throwaway comment saying, because you know it’s coming back. This reminds me of these, you know, what were they called in the old kind of, you know, type one era, type two era, not just doing the trade and then it runs away from you.

[01:06:35] Yoav

Oh no, no, no.

[01:06:36] Niels

So every time, so every time I hear this, the words, “and you know it’s going to come back,” I’m thinking, yeah, well until the day it’s not.

[01:06:44] Yoav

No, no, no, no, no. Let me put this way. So, there are certain aspects that we don’t know. We don’t know what the price is going to do, definitely. Okay. But what we do know is that all our trend predictors have a mean reverting property. So, if the signal is… There’s something mathematical called the OU process, the Ornstein-Uhlenbeck process. So, our signals, for example, are designed so they’re not going to hover at free standard deviation unless they are being supported by price. They will turn to zero unless they get supported by price. And that pullback to the origin becomes stronger the higher the stronger your signals.

[01:07:28] Niels

Of course, sure.

[01:07:28] Yoav

So, that tells you that you have to fight costs more the stronger the signal is. So, what I mean is committing capital. If I’m at the signal, the signal strength is, say, 0.2 and it’s going up to 0.3 you are committing into that trade, that trade is likely to stay for longer. But if the trade is going from 3 to 3.1, because of the pullback, there’s a stronger pullback for very strong signals, you know that trend is likely to stay for shorter. So, it’s just essentially a cost versus alpha balance. And that is slightly different depending on the strength of trend because the pullback to the origin is different where you are in the strength.

You’re absolutely right. If you’re running, for example, a breakout system, then that’s not how you work at all. If you do breakout, you put on the position and you stay in it. And then that’s related.

If you’re doing a moving average crossover sort of type of predictor, then you kind of know that you need a certain fuel to maintain the trend at a certain level. And, on average, the stronger you are away from the origin, the more the pull back to the origin is and therefore you are more cognizant about cost.

[01:08:55] Niels

Fair point. Fair point.

This was great, Yoav. I think in a week of chaos, I think hopefully people will be feeling a little bit more, how should I say, maybe making a little bit more sense of it from at least a trend following perspective. I’m not suggesting we can make any sense of what’s happening in the real world.

[01:09:16] Yoav

I don’t think we can.

[01:09:18] Niels

Or anything to do with terrorists and trade wars. But super fun, super interesting, very much appreciate all the time you put into preparing for this.

And, of course, if our listeners feel the same show some love for Yoav by going to your favorite podcast platform, leave a rating and review. It’s definitely the best way to support the work that we do here.

Now, next week, I think I already mentioned, we are joined by Katy Kaminski who just is putting final touches, maybe she’s already put final touches on a new paper, funnily enough, relating to crisis alpha, which will be very timely, very insightful, fun as well. So, I can’t wait for that conversation.

I hope that we’ll maybe encourage some questions from the listeners and send them, as usual, to info@toptradersunplugged.com and I’ll do my best to bring them up with Katy.

That’s it for now from Yoav and me. Thanks ever so much for listening. We look forward to being back with you next week. And in the meantime, take care of yourself and take care of each other.

[01:10:28] 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’s a significant risk of financial loss with all investment strategies, and you need to request and understand the specific risks from the investment manager about their products before you make investment decisions. Thanks for spending some of your valuable time with us and we’ll see you on the next episode of the Systematic Investor.