The Alpha Most Systems Miss ft. Yoav Git

[00:00:00] 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] Alan

Welcome back to the latest edition of Top Traders Unplugged, where each week we take the pulse of the markets from the perspective of a rules-based investor. It’s Alan Dunne here, this week, sitting in for Niels, who’s away and delighted to be joined by Yoav Git.

Yoav, how are you today?

[00:00:38] Yoav

I’m absolutely fine, just enjoying the sunshine here in Cambridge and the break that I took for August.

[00:00:44] Alan

Good stuff, you’ve had a good summer? You’ve been away a bit?

[00:00:48] Yoav

Yeah, I actually went a little bit north. I went to Budapest first, but then I went off to the fringe with my family. It was a marathon of shows, and it was just absolutely… The city was beautiful.

[00:00:59] Alan

Nice to get away for a bit. And it’s back to school now, I guess, is it?

[00:01:07] Yoav

Yeah, the kids are back to university, the youngest one is back to school soon. So, it’s all back to business. And of course we’re here for CTAs, so, we’re back to business.

[00:01:20] Alan

Good stuff, well, plenty going on in the markets and we’ve a lot to talk about. You’ve brought some great topics, as we always say, but certainly a lot to get into. As we always do, we like to start off by saying and asking what’s on your radar? So, has anything been standing out, from your perspective, over the last while?

[00:01:41] Yoav

Yeah, absolutely. To be honest, I’ve been dreading what’s coming next. I’ve been working on comedy. So, obviously enjoying the fringe with plenty of very good comedians. But I’ve got my own little set which is coming up on Sunday. So, it’s the first time ever that I’m taking the plunge, probably the last one, I suspect.

And really interesting about the motivation. I was talking to people about why you do this, about the way that you do things, actually, because they are uncomfortable. So, sometimes you do things not because you kind of enjoy them but, actually, it’s a challenge. It’s something that you need to do. And so, this one has been on my radar, and if you want a joke, then I’ll probably give you one.

[00:02:33] Alan

So, yeah, I mean, have you done one? Is this your absolute first time doing it in public or have you done it before?

[00:02:38] Yoav

Well, we speak to clients, and they think that I’m quite funny.

[00:02:41] Alan

You’re a joke, is that it?

[00:02:44] Yoav

No, but this is the first official time that I’m supposed to be funny. So, let’s hope I’m funny for the right reasons.

[00:02:52] Alan

Okay, will this be recorded or available anywhere for our listeners to see?

[00:02:58] Yoav

So, I think the people from work are going there and I think they’re already preselling a bootleg copy of that. So, if you want to see me humiliated in public, then I’m sure it’s going to be available and it’s going to go on the quant network in London.

[00:03:13] Alan

Great, well, we look forward to hearing about this and maybe we’ll catch a clip of it somewhere along the way. But fair play to you, that’s a brave thing to do and I’m sure it’ll go very well.

In more mundane matters, I mean, obviously there’s been a lot going on in markets. We’ve got Jackson Hole going on at the moment and Fed Chair Jerome Powell speaking today. So, that’s a big market focus. I guess as a quant you don’t tend to think too much about what Jerome Powell might say, Yoav?

[00:03:44] Yoav

Well, of course we think about what he’s going to say. It’s a question of what we can do about what he says. So, I mean, it has been a very difficult period for trend in fixed income. So, if you look at the US it’s been oscillating. You know, I want to be long, I want to be short.

10-year yields have been moving around indeterminately but there have been other markets where you see a significant drift in yield. In Japan, if you go to 10-year yields, they’ve started a long climb, back in 2022, from zero and now sitting at an all-time high last night. So, that’s definitely one place where there is an impact. And you know, in markets like Tai baht, you see exactly the opposite where yields have been coming down quite significantly over the last year.

So, it’s interesting that the US and the rest of the world are doing their own thing, and it will be interesting, but actually for the economy and for the equity market, in the US, it will be very interesting to see what we get in Jackson Hole.

[00:04:48] Alan

Yeah, absolutely. A couple of things on my radar that I saw this week that were interesting. One was an FT story about fiscal dominance. And this is something people have been talking about for a long time, the idea that monetary policy will kind of have to adapt to the fiscal stance when debt levels get so great. We’re kind of approaching that level, but not quite there yet. But it’s interesting how it is becoming such a topic in markets. And, as you say, Japanese yields touching record highs, I guess, probably all-time highs, reflecting that concern and the general drift higher in yields.

And then the second thing that I saw today was BlackRock out with a report recommending that investors increase their allocations to hedge funds by 5%. So, good news I guess for all in the hedge fund industry. I haven’t read the full story, I just saw the FT talking about it. But are you sensing that from investors you speak to as well, Yoav, more appetite for hedge funds generally?

[00:05:51] Yoav

Well, possibly hedge funds, but CTAs, I think, we have the exact opposite. We’ve had six months of poor performance in the beginning of the year, and I think people were very comfortable with it.

But now you can see there was a Wall Street Journal talking about trend. So, actually, in the CTA industry it doesn’t matter how well or badly your own CTA is doing. I think there is a general malaise in the CTA industry. Investors are waiting to see what’s happening which is fair, and it’s a bit difficult to come to investors when you’re down.

But I think that’s why we talk about it because if you look at historical records, and all-time performance of CTAs, the reasons why they should invest in CTA actually haven’t gone away at all. So that’s my experience.

[00:06:47] Alan

Yeah, interesting. I mean certainly I think all the reasons that were highlighted for why investors might consider more allocations to hedge funds in terms of a more volatile macro environment and greater dislocations in macro variables and market volatility. These all apply for reasons why you might consider a CTA as well, obviously, as part of the hedge fund allocation.

So maybe we’ll touch on performance first before we get into the main topic. So, just to update, as of yesterday, SocGen Trend Index up 1.67% on the month. SocGen CTA index up about 0.50%. And year to date the Trend index down 8.5%. And the SocGen CTA index down just under 7%.

So, as you said, Japanese yields trending higher. There have been a few trends continuing particularly in the fixed income side, I would say, over the month. Japan, European yields, US yields are a bit more choppy. Outside of that it hasn’t obviously been a positive period but looking across a lot of the markets they are still quite rangy; FX equities up and then a little bit of a correction the last few days.

Have you any perspective on performance this month or the general environment we’re seeing? Obviously, it’s been a tough year, but we’ve had July positive, August looking more positive now. Are you starting to see more persistent trends emerging?

[00:08:14] Yoav

I’m a pessimist by nature, so I’ll count my chickens at the end of the year, I think. But I think it will be a mistake to look at one month performance, two months, even six months performance.

It’s, of course, encouraging and I think what is interesting is clients are rooting. They really want us to do well, and this has been very encouraging. But I think if you’re going into trend, you know it is painful. It’s going down the stairs and up the elevator. And, as you say, the dynamic nature of the trading is what you’re buying.

You’re buying that beautiful zero correlation to the rest of your portfolio while you have a dynamic adjustment to market conditions. And that hasn’t gone away. And that’s a very fundamental reason to invest in trend.

[00:09:20] Alan

Well, let’s get into the main topics. You’ve brought along a couple of interesting papers, and I wanted to talk to you about a couple of your own blogs as well. So, plenty to delve into.

So, the first paper is one that you highlighted from Moritz at Takahe Capital. Do you want to give a quick intro to it and get into it then?

[00:09:39] Yoav

Yeah, absolutely, I love that paper. I really do like it. I mean, it’s a paper called Don’t Just Trend It, Curve it. And they’re talking a little bit about, about curve trading.

So of course, if you’re trading copper or if you’re trading corn, the natural thing is you say, well, there’s just one object called corn, but really there isn’t because you can buy corn for delivery at different months, and the different months may behave differently.

Specifically, for example, corn delivery, at the moment, is using this year’s crop, but corn delivered in December will already be next year’s crop. So, there is a difference between them. There’s also a difference in terms of supply but also there is a difference in terms of demand. And sometimes there’s a difference to do with financial conditions and regulations changing.

And I think that paper is taking one example of that which is looking at the copper, because we know that we’ve seen a taxation on copper changing over the last period to do with Trump deciding to include it or exclude it from the tariffs. And what they’re doing is they’re saying, well, actually, if you look at copper prices, they have a certain behavior. But if I look at the difference between two deliveries, two contracts, suddenly, sometimes you get a dislocation. And they behave differently precisely because one contract is before tariffs, one contract is after tariffs. So, there’s a very transient risk factor which you can take advantage of.

And I really like the paper, and I think there are a couple of really good points they make about this and I think it’s actually interesting what they do because it’s actually quite different to what traditional European CTAs are doing. So full kudos to them.

So, the first point that I really like about that paper is the general observation that they make, which is that the front of the curve is generally the one which is more volatile in the commodity world, and it is the one which is more affected by macro news flow. And we see that precisely.

The long-term price of copper is going to be affected by long-term supply and demand, and it’s going to be less affected by local dealers because the tariff coming on/the tariff coming off is, at the end of the day, a very sort of local… It’s a very local news story. So, that affects the front of the curve. And therefore, that’s why you see this dislocation.

And that’s something that we see not just in copper. We see that across the commodities. In general, what we see is that when other players come into the markets, they tend to congregate where the liquidity is, and the front of the curve is where the liquidity is. So, everybody congregates in the same place where everybody else is trading, which is good for them. And that allows a lot of volume to be traded. And we see that in terms of increasing volumes in the front contract.

The second point that they make, in the paper, is that spreads exhibit strong breakouts and reversals. And I like what they say about breakouts because, when spreads actually do break out, it’s actually quite violent. And I think an observation that I would like to add – it’s about risk management and about allocation of risk.

So normally, if you were to look at traditional CTAs, then they allocate to each of the individual markets and that allocation is relatively static over time. So, suppose if I were to allocate to outright copper trading, I will allocate maybe 2% of my portfolio to that market and I will expect copper to provide me with 2% of volatility, day-in-day-out (a little bit less when there’s less trend, a little bit more when there is more trend), but overall, sort of a 2% over a long period of history. When we look at spreads, you can’t do that for a couple of reasons.

The first one is, and they do mention it in the paper, which is, well, that specific spread is going to disappear when the front contract actually expires and there is a delivery. This particular object, which is the spread between the August and September contract, that just doesn’t exist anymore in the market. But more importantly, although it has existed maybe for the last three years, the volatility, itself, is actually very volatile. You will have a period where this spread is doing nothing.

And if you try to make this spread provide you with that risk, you will have to leverage your position to such an extent that when a dislocation does happen, you will be out of business. You cannot allocate to the spread on an ongoing, continuous basis, hoping I’m going to make money. Because when there is a dislocation, like the Trump trade, this is going to be very violent for you. It’s very left skewed.

So, what you have to do is to understand that you have to wait until actually a dislocation happens. You can’t get into the trade first. You have to wait until the spread can actually provide you with the minimum level of liquidity, a minimum level of risk because there’s a certain volatility. And then you can start trading it, and sizing the position, at that point, it is actually possible.

And I’m a big fan of this approach, I have to say, because it’s dynamic. You don’t know that this is going to happen, and indeed, if Trump hasn’t made a comment about tariffs, then maybe it wouldn’t have happened. But when it does happen, being able to have the framework to allocate that risk factor is really nice.

And a lot of streamlined CTAs don’t trade spreads or just haven’t attacked this problem properly. Being able to allocate meaningful risk to risk factors that rise, and they are transient by nature, is something which I think is quite important in terms of providing diversification. Because when it does happen, it is actually trading something very concrete which is very uncorrelated to the rest of our portfolio. So, that’s one of the reasons why I really like this paper.

Anything you want to add on this one?

[00:16:31] Alan

To set the stage… So, we’re talking here about spread trading more from a trend perspective, a breakout perspective, isn’t that correct?

[00:16:42] Yoav

Yes, absolutely.

[00:16:43] Alan

I mean, because obviously a lot of managers trade spreads more, I suppose, for the spread to revert to more mean reversion, I guess. Wouldn’t that be fair to say?

[00:16:53] Yoav

That is absolutely correct. So, there’s a relative value and, if you like to think, the carry system is very much the traditional long-term, if I think about the difference between two spreads, it embeds a certain amount of sort of cost of carry.

[00:17:16] Alan

Yes.

[00:17:16] Yoav

Some convenience yield and definitely, over a long period of time, they tend to merge. So, if you ask me, what’s the difference between five year delivery and five year and two months delivery? There isn’t really a good reason for that. The only reason for that is essentially a cost of capital because there is going to be some interest rate differential, there’s going to be some storage cost, and so forth.

So, the play for the convergence of these two things is normally the carry trade, and it’s a kind of a nice and slow process. But then what you have is you have these dislocations which are to do with the two contracts behaving differently because they become not exactly substitutable.

So normally, if the two objects are fungible, then you will expect them to converge. But in a case where, for example, this year there’s a good supply of corn, but there is a drought, or there is a disease, and suddenly we have a supply shock. That supply shock will hit you at sort of next year’s crop, say, so, then what you have is you have a real reason for that divergence between those spreads. And that will create a trend. It will be a short time trend. It’s not going to be a very fast moving thing, but that’s kind of what’s going to drive those dynamics of that particular spread.

In crack spreads, so, suppose you have a spread between the oil, the raw oil, and you have the distiller (the thing where you distill and you get heating oil, and jet fuel, and all of these things), and what you have is sometimes you have a breakdown in sort of the processing of that oil. So, if there is a Gulf of Mexico, most of the work is done in distilling oil in the Gulf of Mexico.

Suppose there is a storm there or there is unscheduled stoppage in those refineries, right? So, suddenly what you have is that spread will start trending too much there. And those risk factors are generally transient, but they are very exciting, and they are diversifying because that refinery closing down in the Gulf of Mexico is nothing to do with maybe the overall long-term behavior of oil.

So, that’s what they’re doing. But yes, it’s actually very tricky and it requires some care. And it’s interesting that they use breakout rather than using sort of a smooth trading because you actually are looking for a breakout as an indication of something fundamental happening there.

[00:20:02] Alan

It is interesting that they are uncorrelated with the underlying. I mean, because as you mentioned there, the front of the curve is more volatile, more influenced by macro factors. So, you would have thought that those macro factors are driving the front month or the near month future, wherever you got your main trend position. You would have thought they would be more correlated, possibly, if that tends to be the driver. I mean I think they show that or they mentioned that in the paper, isn’t that right?

[00:20:28] Yoav

So, we have to be cognizant about what’s going on. So, in the underlying two contracts, most of the volatility in the underlying two contract will be very much correlated to general trends, macro trends, or trends in copper. But what they’re looking for is the residual volatility – the difference in movement between the two. And that’s important in terms of sizing. So, normally there will be a 90%, 98% correlation, say and there will be very little volatility in that spread.

Which means, actually, if you want to have meaningful allocation for it in your portfolio, you would have to take a 50x position. And that’s insane. It’s a little bit like LTCM. LTCM was betting on the ‘on run and off the run’ yield curve. These are two US yield curves, two different types of bonds that converge. Normally they behave completely the same. So, the correlation is 98%.

So they, in order to get exposure to that difference, that tiny difference, that 2% of residual volatility, LTCM bumped up the risk profile by 50x,100x. And then, of course, when those spreads refuse to converge or refuse to do what you kind of expect them mathematically to do, then it becomes very expensive. What you need to do is you need to wait until there is a structural break – news, tariff, refineries, whatever. At that point that spread becomes more volatile. So, instead of, say, being 2% of the overall volatility, it becomes 10% of the volatility.

And now it’s interesting. It’s actually a valid risk factor that you can trade with not too much leverage. And so, that’s the one thing which is good about it.

The second thing which is good about it, and the reason there is a breakout is because there is an underlying risk factor which is transient, but it is happening right now. And therefore, there is going to be a trend. We see that in the FX market in currency basis.

So, currency basis is the difference between funding in the local currency versus the US, and sometimes there is just a transient pressure. Maybe what we have is a company that issues bonds in US dollars, and it wants to bring it into the local market, and that brings a certain funding pressure, and you see the basis trade moving, and then you can trend that as well.

But you have to be very careful because you have to appreciate that risk management is key here. The ability to say, I’m going to wait until there is a sufficient risk to start trading and then I’m going to get out of this when there isn’t enough volatility.

[00:23:26] Alan

There are a couple of thoughts I had. One is, obviously we have the kind of justification or the rationale for why trend following works on markets in general in terms of behavioral biases, etc. And then does that still apply to spreads?

And then the second question is, do you then get into the whole realm of synthetic markets and trading all manner of different crosses to create these multiple spreads, which, obviously, some CTAs do as well, or where do you draw the line? What are your thoughts on all of that?

[00:23:58] Yoav

So, that’s very interesting. Generally, trend is very prevalent in all areas, but I think you’re right in terms of relative value trades are called relative value because it’s value rather than trend. So, making money out of spreads is more difficult. And I think that takes us to synthetic or just factor trending. That takes us into that universe.

And let’s look at the normal CTA universe if you break the factors, that you look at, down. So, suppose I’m looking at all interest rate markets, all bond markets, and I do a PCA on it and I look at, what’s the trend behavior that you see? And what we see is as follows. You see the, the front, the main dominant factor is it trends, generally, very well because that’s exactly what we kind of want. And, in fact, what we see is that maybe the first 25% of PCA factors trend very well.

Then we have 50% of the remaining PCA factors, and they do trend. So, although they might look, to you, like a spread between Japanese and US rates, there will be a trend behavior because there is an underlying macro behavior. They may be less dominant but they have a sort of a low level trend. But what is interesting about them is that the volatility available on them, it’s a much smaller risk factor. So, the first PCA covers about 70% of your money, the top quartiles – it covers a lot of the risk.

So, there’s a lot of risk, but it’s very diversifying and it still trends quite nicely. And then you get to the very, very bottom of your PCA factors, which are very pure RV trades, and they tend to actually move in the opposite direction. So, they will tend to be, actually, very strongly mean reverting.

So, there are ways you can actually incorporate that in your trading strategy in a way that sort of allows you to essentially get rid of those small factors, maybe. Or maybe you want to up-weight the middle factors because although they trend less but they still trend well, and they still provide you with some volatility. So, you just have to pick. We are just trying to create a system which harvests alpha in all sort of factors. But you just have to be aware about what’s happening.

[00:26:37] Alan

And presumably, in terms of speed, if these are transient risk factors, as you say, the spread will diminish over time. Do you have to trade them faster, presumably, than you would on a pure trend following basis?

[00:26:49] Yoav

Fast, but not as fast as you think you kind of have to. So, that’s interesting. Because otherwise there’s cost. So, we have to think also about costs. So, when we look at the outright, if I’m trading the outright, there’s going to be a certain cost on a Sharpe basis. So maybe 2 or 3 basis points of Sharpe. Okay, so, 0.02 of Sharpe.

But I’m giving up trading the underlying and, as you say, I can trade it slowly because slow trend is actually very dominant and that will be even less expensive. If I’m trading spreads, and I have a couple of considerations. The first one is that I have to trade a lot of contracts, in both directions, in order to be able to get the risk that I want. So, the notional amount that I trade is higher and therefore my trading cost will be higher.

So, per turnover, the spread will be more expensive than trading the outright. And then, of course, you’re saying, well, if only fast trend is making money at that point, it’s actually becoming economically unviable.

Now, there are ways you can still incorporate it into your main system in a way that I think, I suspect, the two Moritzs are doing, even if you think it’s too expensive to trade, it can feed into the process of which contract you want to trade.

Okay, so for example, if there is a spread which says okay, right now, the back of the curve looks cheap while the front of the curve looks dear, if momentum wants to buy something, then maybe you should buy the back, you shouldn’t buy the front.

So, although the contribution is not going to be as great as like trading, you know, going full gang-ho and trading the spread, it can still have incremental benefits to your overall trading by allowing you to sort of choose which point on the curve.

And I think that’s very interesting because when you think about what’s happening that’s related to the comment, in terms of the front of the curve, it is more shocky, it’s more exposed to macro, sort of the macro trends which are actually more to do with what’s happening with SMP and the US treasury. Whereas the back of the curve is more affected by supply and demand. So, generally you would tend to move slightly further out of the curve.

And I think one of the things that they failed to mention… not failed, it’s a very short piece. So, I’m not going to begrudge them on that. But it’s to do with who is actually trading.

So, as we said, when you know, if you were to look five, ten years ago there were very few shops, even multi strategies that were looking at trading copper. But these days everybody’s trading copper and as a result the copper has become sort of more speculative in nature.

If I look at who’s trading the front curve in copper, it’s a lot of CTAs, a lot of multi strats, a lot of speculators, not necessarily the people who are actually hedging, and producing, and buying copper for production. And we see that in terms of the ratio between… There are multiples way of seeing it.

The first one is to look at it in terms of the volume and the open interest. The open interest when the contract closes is what remains. Like, who is actually wanting to get exposure to the copper contract. So, we can compare the volume during the trading versus what who actually wanted to take that position. Or we can look at the COT report, the Commitment of Traders reports by the US, and see which ones are the speculators, which ones are the hedges. And you see the front of the curve becoming more dominant by the speculators. So, that’s your universe.

If you were to move a year out in copper (I’m not sure you can actually move a year out), I suppose we can move a couple of contracts out, then the people who are doing that are actually more the hedgers and producers, and they tend to trade more slowly. So, there’s less volume, there is more open interest. And that’s the difference between the stock, between position and volume, between stock and flow. So, the curves do have different dynamics. And when you’re doing trend following these different physical characteristics of the contract actually will feed into your trading system.

[00:31:39] Alan

Just to your point, like the front being more speculative activity, I mean does that suggest that it’s more zero sum than trading in that sphere? Obviously, you know, if it’s all specs, they’re just profiting from each other which, you know, there are some winners, some losers.

[00:31:57] Yoav

Well, the Efficient Market hypothesis applies everywhere. But it’s more about what is the fundamental reason for things trading.

[00:32:06] Alan

But if there are a lot of hedgers in a market then they’re risk averse. But in theory…

[00:32:13] Yoav

There is a premium.

[00:32:14] Alan

There’s something to it. Yeah. For specs to capitalize on.

[00:32:15] Yoav

Yeah, yeah, absolutely. So yes, so I would say that is the issue. I think, actually, I’m going to say something about the loading on trend at this point, actually. We’re having a lot of discussion with clients and an argument which comes up quite often is to say, oh, trend is at capacity. The reason why the trend hasn’t performed in the last year is because there’s just too much money in trend.

And what is interesting is, okay, I think the most famous paper is there was a recent paper by Man, which Russell talks about, oh, the size of the AUM of the industry has been constant at $300 billion. The industry has grown significantly, which is true. Okay, so, that means we are a small part of the market.

The problem with this argument is that there is tons of trend being done at other places where you don’t see it; in QIS, you see there’s tons of trend inside the multi strat if you are going into Millennium, QRT, any of the big shops, there will be pods inside that they will be trading trend. I can guarantee you that is happening. So, taking a global bird’s eye view of what’s happening is not, actually, necessarily informative.

But what is informative is that we can actually look at the speculators and the hedgers in the COT Report and we can fit the behavior to different trading strategies. So, it’s not a prediction, it’s much more a descriptive view. It says, I’m looking at the behavior of the speculators over the last couple of weeks. Does that load on trend?

If the industry was overly trend reliant then what you would see is you would see a good correlation, a beta between the speculators positioning in the COT Report to your signal. And we can fit that.

And what we find is that we can actually detect a few things that are very interesting. So, if you look historically, you can see that the speculators have become very sensitive to volatility. So, the loading-on changes in volatility. If volatility goes up, the speculators proportion goes down. That that is now, it’s quite a dominant effect. So, that’s the first thing.

We see the reduction in loading on fast trend. So, if you go historically and you go back to sort of the early 2000s, you see that changes in fast trend explained changes in speculators’ positioning. That has gone down over the years, and it’s quite nice, it’s quite reassuring that we see this because that actually reflects the reality that we know.

If I look at the SocGen CTA index, if I look at the way that the CTAs have moved, they’ve migrated to trading slowly and loading less on fast trend. What we don’t see is we don’t see, suddenly, a huge rise, or in fact any rise whatsoever, in terms of the loading on slow trend or mid frequency trend in the data. So basically, the industry has actually being very careful not to take too much risk in any particular contract.

And then that capacity argument, the data just doesn’t bear it. We just don’t see this. Where we can see this data, that argument just is not there.

So, the clients are always asking me what is causing this issue? Why didn’t we see trend performing in the last year? That we can have a long discussion about that. But in terms of the loading on, are we just too big in the market? No, we’re not. That data just doesn’t bear it out at all.

[00:36:46] Alan

Well, that’s maybe a good segue into the second topic which is a paper which gives a lot of really interesting granular data on different market participants and their exposures in the FX markets. And that’s a recent paper from the Bank of England, which is really monumental paper in terms of the data they have put together.

Do you want to give some context and an overview of this one?

[00:37:14] Yoav

Yes, absolutely. So, if you look at FX data. So, first of all a little bit about the FX market. By size, in terms of volume of trading, it’s probably the biggest, easily dwarfing the equity market, say. So, if I look at the BIS Triennial Survey from 2022, the amount of daily trading in FX is something like $7.5 trillion. Okay, that’s just like insane.

So, it’s a huge market and there are lots of players in it. Of course, anything from the retail people wanting to buy some dollars to go on holiday in the US, to Tesco, which is sort of the non-financial corporations that they want to hedge, they have assets abroad, they have income from abroad. Tesco is less so. But companies that export, they might have an overseas operation and they will hedge their position into the local currency. And then all the way to hedge funds and sort of our investors. So, there will be asset managers who are in this market. So, everybody’s playing the FX game. Okay, and $7.5 trillion per day.

It’s just an insanely big market and there are a few data sets for that, in terms of understanding the positioning. There’s CLS, one of the biggest providers, they do FX settlements, and they offer some data which is very anonymized. But now we’ve just hit the mother lode. So, this paper is just insanely amazing in terms of data.

Okay, so the paper is called Topography of FX Derivatives Markets; A View From London, and it’s a collaboration of a few people from the Bank of England, a guy called Daniel Ostry, a few people from Fed, there’s a guy called Sinem, (I can’t say his surname so you’ll have to excuse me). And it’s a collaboration of a few central banks and a few researchers, and they have data which is non anonymized.

They have 100 million transactions, millions of transactions in the London FX market from 2015 to 2020 and it’s non anonymized. So, it’s from the vantage of being the regulator. You can see everything, absolutely at daily firm level positioning. And that’s just amazing. And that allows you to do things that you wouldn’t be able to do otherwise.

What I love about this paper, first and foremost, is that they don’t try to predict cross-sectional FX returns. Many times you will see academic papers in our finance industry, and the researcher rushes into, okay, here’s an effect, here’s some data that I have, let’s see if I can use it to predict cross sectional equity returns. I mean it’s just like hundreds and hundreds of papers, thousands of papers which are trying to do that. And what I love about this paper is that it, in some sense, says, okay, I’m just going to let the data speak for itself.

I’m just going to look at the data, I’m going to do very little adjustment, a few things that we actually have to adjust to correct the way the data is reported or to understand what is happening. But I’m not going to try and make any predictions. I’m just going to describe the structure of the market. And I find it extremely informative.

And I think a lot of quants, again, rush into, here’s a trading strategy. Sit back. Just look at the data. Let it tell you a story. And when the story is done, I mean you can probably guess.

There’s a paper, there’s a famous paper that just went out in March, The Virtue of Complexity, which basically says, I’m good to show everything, I’m going to make it overcomplicated. You can probably tell in which camp I lie.

I think before we start doing anything with the data, it behooves us, as quants, rather than just shoving it into a machine, to really understand what’s going on. And this paper really does it beautifully.

So, there are a few observations. Actually, let’s start with the appendix because that’s the one that relates pretty much to our earlier discussion.

So, if you go to the appendix, there’s a beautiful description about where do each type of player in the market, which derivative contract they trade. So, the paper is not looking at spot, it’s looking at FX derivatives. And FX derivatives can be anything from one day, like tomorrow next, or spot next. So, it can be either one day all the way to maybe two, three years. Okay, so we’re moving from FX forward to FX swaps further out of the curve.

And what you see is exactly the same phenomenon that we just described in terms of the habitat of different players sitting in different parts of the curves. So, the market makers, they try to hold most of the risk in the very front. So, if you look at their chart distribution of their positioning, most of it is up to one week expiry. And that’s the most liquid part (that’s the part that they use to hedge their risk).

And then you move to the hedge funds of the universe, and asset managers, and you move further out, and you move to sort of the one to three month maturity. This is not a huge surprise, and this is how we would do that. And I think most CTAs will be trading the next IMM expiry – every three months, sort of the March, June, September and December the 3rd Wednesday of the month expiry. And this is where the hedge funds congregate with the liquidity.

And then what’s happening to the back of the curve? Who is holding the two or three year FX swaps? These are real people, these are the non-financial corporations, the companies that have exposure. They have genuine long term USD cash flows and they need to bring those back into the local currency. This is all taken from the London market. So, the local currency is GBP, or euro (before the Brexit happened).

And what we see is that they inhabit that habitat – the two and three years out. These are by the hedges the people who actually genuinely use it. And I think a lot of the risk is basically rolling that swap – the activities. They just want to make sure that they hedge their cash flows and they don’t have the risk. So, it’s very interesting. So, that’s one amazing part of this paper. But there is plenty of other stuff.

[00:44:23] Alan

Yeah, as you say, I mean, it must be the first of its kind in terms of this level of detail that we see. There are a few things that were interesting, as you say, the time horizon, that different segments hold their positions or primarily trade as interest. And then they also do show the net exposure by the different segments over time, which is interesting as well.

Which, obviously, as you’d expect the hedge funds specs, their net exposures to the dollar, euro, and sterling, fluctuate a lot. But one that was interesting was, I think, the market makers or the dealers obviously tended to hold the opposite risk to the hedgers. Did you notice that?

[00:45:14] Yoav

So, that’s correct, and a little bit misleading in a way. So, what you have is one side of the picture. There are two types of people. There’s the market makers and the dealers. You will see that the market makers have relatively little risk overall. And these are the ones who are really trading like Duracell bunnies.

They are buying and selling and they’re really adjusting the prices to ensure that they don’t have net risk. But what you have is you also have dealers that will be taking the risk, the dollar risk. So, there is a structural short dollar for UK corporates because they hedge their dollars back to the UK.

They are long GBP, and short US dollar. And the dealers, on our data set, will look like they will be the opposite. They will be long US dollar and short GBP. But what we don’t see is we don’t see the positioning, on the other side of the Atlantic, which is what the position is of the dealers against US corporates.

So, the US corporates will have exactly the opposite position, which is that they will be all short all other currencies (GBP amongst them), and they will be long USD, and the dealers will be providing that. So, the dealers will provide a slower netting of this universe.

And I think what you get overall, and the reason why I’m saying they will hold less risk is because they can’t hold a lot of risk. So, these days…

[00:46:46] Alan

But I thought this was the net, the net exposure.

[00:46:49] Yoav

I may be incorrect, but I’ll need to check.

[00:46:51] Alan

You’ll have to check.

[00:46:53] Yoav

But the problem with dealer banks in the US is that you can’t hold, on your book, too much risk these days. So, one of the outcomes of the GFC, the Great Financial Crisis, is that the amount of risk, which is available on the market that you are able to hold, is actually not that great. And we see that in bonds, we see that in credit, we see that everywhere. So, I suspect a lot of that risk is actually transferred through US corporates.

[00:47:28] Alan

The others are, as you would kind of expect, really, non-financial corps, obviously holding the kind of exposures that you would expect from a hedging perspective. So, you know, effectively short dollars, I guess, selling their dollars generally from their positions overseas. Outside of the positioning does anything else stand out that might have an obvious takeaway for you?

[00:47:57] Yoav

So, there is one which is very much related to trend and carry. But, before we get to that, there’s a little bit about the market structure that’s very interesting, and that was a huge surprise for me. So, with a market that big, you kind of expect to see a lot of players in the center. So, there are a lot of disparate players. But what is interesting is how concentrated it is.

It’s like a star shaped domain. There is the center which is providing the liquidity. But what is insane is that the top five dealers, the top five market makers, really dominate everything. And it didn’t used to be like that.

I mean, 20 years ago we were looking at the EBS data, which is the interbank netting sort of FX trading application. A lot of banks were involved in the game, and I suspect there’s still local currency and there is going to be a local effect. So, you know, so there will be a Swedish bank dealing with Swedish krona.

But by and large these days the real trading is basically channeled through very few counterparties. And that’s interesting and potentially a cause for concern. The interesting thing is that this is actually very different to the structure in the FX spot market.

So, this is all the derivatives, and if you look at the FX spot market, the emergence of players like the nonbank market makers (so, like the XTX and the Jane Street of the universe). So, it’s actually very interesting. It’s a very interesting difference between the forward and the swaps and the spot market. But structurally that’s very interesting to me. I’m not sure exactly what the implication is but this was like really important.

[00:50:13] Alan

I think there’s a lot of data in this one. It’s definitely worth having a look at. I’m sure some more observations will come from further review, but I wanted to move on to something conscious of time.

You published a blog recently which I wanted to touch on, which is around asset allocation. So, it’s obviously topical given what we talked about a little bit earlier with BlackRock suggesting increasing hedge fund allocations. And elsewhere I saw Vanguard kind of going against the tide a little bit. They’re suggesting 70/30 is going to come out for the next decades.

[00:50:51] Yoav

Let me tell you a little bit about my blog and a little bit about Vanguard. So, I didn’t know much about, and I still don’t know much about the way that allocators are allocating. And normally, when we go to an allocator, we say, oh, give us some money. You know, we are trend following, we are uncorrelated to everybody else. You just give us some money and put us aside, and so forth.

And I talked to an allocator, and I asked him, how do you take into account the fact that the positioning might change and suddenly all of your allocations are long equity, and suddenly you have a lot of risk in equity? And his reply was, well, we actually do try to figure out the beta, let’s say the equity beta, on each of the data sets and we fit the P&L of each of the people that we allocate to. And that gives us an idea about what the positioning is like.

And this is kind of how we do it. It’s quite messy, but this is how we do it, which is we try to find out the beta of the equity that we have right now with this hedge fund positioning. And we exclude you guys. We exclude CTAs because you guys, you change your mind too often. And we can’t really retrofit what your positioning is. Which, I think, actually, shows you how uncorrelated we are to other asset classes because it’s like you can’t do the fit. We have zero correlation. It’s very difficult to fit what our positioning is.

And that started me thinking about what the process of an allocator to allocate is. And I felt, hang on a second. What they do is they go through a process. The first things they do is they make a capital market assumption. They make an assumption, or in the Vanguard paper that you mentioned, it’s sort of VAAM, sort of time varying asset allocation model. What you do is you say, what do I think long term, in the next 10 years, what’s going to be happening to equities? What do I think, what is the annual growth I expect in bonds? What’s the income that I’m expecting from commodities?

There are some correlation matrix assumptions that I’m going to make. And based on that I’m expecting, based on long term growth, I’m going to have the allocation. I’m just going to create an allocation matrix. And that’s actually, probably, one of the most important decisions that you have to make: what’s the proportion between equities and bonds?

And not surprisingly, it’s going to be related to how much long-term, how much value you expect to extract from equities and bonds in over a 10 year horizon. And I thought to myself, hang on a second. Within this process, when you do this optimization and you do this correlation matrix and you do this optimization, then there is a gradient.

So, the idea is if you think that equities are going to go a little bit further up, then actually you should allocate a little bit more. So, there’s a delta to your assumption. So, whatever allocation that you’ve decided on, you kind of think there is a delta to your assumptions both in bonds, and equities, and so forth. And I thought the delta was positive. So, you might want to say, if you think that equities have gone up, you might think that they will go up much more.

So, I wrote this piece, and I wrote this piece about CTAs, but actually I think it is more about the role of systematic strategies. So, systematic strategies, you know, risk parity, relative value, trend following, there will be funds that do that. And you kind of know what they’re going to do. It will be a mechanical adjustment of the data.

And I wrote it about CTAs because, of course, we trade CTAs. And, of course, the feedback I got was, you got it completely wrong. What you got is, you got it completely wrong. In fact, most of these models are mean reverting by nature i.e. if equities go up, they will actually underweight equity. They will say long-term I will drop my expectations about what’s going to happen to equity and as a result I will deallocate from equity.

Now interestingly enough, that’s actually coming back to the positioning. This is exactly… Those long-term investors are the ones that take the opposite side. We’re talking about positioning, these are the ones who take the opposite side to trend in the market. And in fact, the paper, the Vanguard paper, is exactly what you sent me. And I thought, oh my God, yes, I’m going to have to own up to this one.

What they are saying is in June equities went up tons. So, they went up 10%, 11%. So, in July they came up with a paper and said, oh, actually now we are adjusting half. If you think about that 10% equities have done that 10% rise, we thought they were going to go up 4.5% over 10 years. Now that 10% is gone. So, I think we’re left with 3.5%. Actually, they only cut 50% of it.

So, they cut the long-term expectation to equity by half a percent based on actually a very short-term change in equities. And as a result of that, you basically under allocate. You want to under allocate to equities. And they came up with a very bold number to be honest. They go for 30% in equities and 70% in bonds. Okay, notice, no hedge funds here.

[00:56:33] Alan

Yeah, I mean, I think it’s just between the two assets.

[00:56:36] Yoav

Yeah, absolutely.

[00:56:38] Alan

But I mean, I guess the benchmark is probably 60/40 equity. I mean, I think, as you say, it reflects more the equity valuation more than the view on anything else. I think that’s just…

[00:56:49] Yoav

Absolutely, and it makes sense. And I’ve had a chat with my allocator, and he said, yeah, yeah but woe onto those models. We kind of don’t like them because they have been negative equity for years.

[00:57:05] Alan

Yeah.

[00:57:06] Yoav

So, it’s the naysayer that, you know, one day they will pay. They will make money, all of these very long-term 10-year horizon things. All of these maybe make sense if you’re investing for 100 years. But suppose, after the equity rise, you then decided to go and sell your equity. You wouldn’t have done particularly well. I mean, at the end of June I think the S&P was 6.2 and now it’s at 6.4.

So, equity has been rising quite happily regardless of what Vanguard is saying. And I think that comes into the role of trend, not necessarily as long-term, as a tactical way of you phasing in that long-term view. We talked about the way which spreads I favor, right now, as a way of improving your trend. We can think about the way trend improves the long-term allocation model.

So, from the perspective of the Vanguards of the universe, trend is one of those very funny, very fast-moving objects. But we can use that signal, if you’re allocating to CTAs, to say, oh, okay, right now I want to deallocate from equities but trend is really strong. I don’t really want to allocate right now. I’m going to wait. Maybe, by the end of the year trend will reverse. Now I’ve naturally deallocated from equity the way I kind of feel about it.

So, I think, although the premise that I had in my mind, that the delta of long-term allocators is positive, is completely false. Nevertheless, I think the principle of what I’m suggesting is correct, which is that the systematic strategies or rule-based strategies, you want them in your portfolio because if you include them inside the optimization process that you’re doing, you will be able to create an optimal allocation over a much wider set of parameters.

So you know in your mind what would be your allocation if your assumption is 3.5% for equity in the long-term, or 4% in equity, or 3%, in equity, you should pick a collection of systematic rules of trading, which is actually available in the market, to allow you to essentially automatically autocorrect depending on how your assumptions changes.

[00:59:57] Alan

I mean, it’s kind of like the allocation to CTAs should be upgraded. Is that kind of what you’re suggesting? I mean, I understand what your point is, CTAs will modulate your overall equity exposure. So, if you have 50% equities and you have 20% CTA, you’re going to pick up some more equity exposure in your CTA allocation. And that exposure will fluctuate over time depending on the strength of the trend in the equity market. So, in a strong bull market it’ll go up, and in a bear market it might go short. And that has a favorable portfolio impact.

But, as you say, the way most people do asset allocation is they think about, okay, I mean, everybody produces capital market assumptions and then to feed that into some kind of optimizer or some kind of framework where they’re thinking about what’s the expected return from these asset classes and what’s the volatilities and correlations. So that would determine some level of exposure for CTAs. I mean, you have to formulate a return expectation. But are you suggesting that because of this additional role, CTAs should be higher? Because it does point to… I mean, this is something I’ve been thinking about. This kind of framework was originally designed when you’re combining different assets: bonds, equities, property. But now we’re adding strategies, such as CTA, which are trading some of the assets that you’re diversifying. So, this is the point we’re talking about. Sometimes you’re going to pick up more exposure. I mean, you’re allocating to them for a risk factor. So how do you reconcile all of that?

[01:01:36] Yoav

I think it is the time horizon that we’re talking about.

[01:01:39] Alan

Yeah, absolutely.

[01:01:44] Yoav

So, actually the entire time horizon and habitat. So, if you think about those capital market assumptions, these are about the 10 year horizon.

[01:01:52] Alan

Exactly.

[01:01:53] Yoav

And, in that horizon, the natural fluctuation of equity… In fact, what you should do and what we advise a lot of retail people, actually, stick with those numbers, have those allocations and go and play golf. You know, allocate for equity long-term. For the long-term allocation, which is what the CMA model is designed for, which is looking at long-term growth, what’s happening in the next three months is completely irrelevant from their perspective. And you are just going to ride this out, and you should go and play golf, and just ignore Trump, and you should ignore whatever, Covid. We should just go and enjoy our positioning over time.

And what we are seeing, as you say, is that those active traders, not just CTAs but multi strats and other macro hedge funds and so forth, they are playing in a different time horizon than you. They are playing at the three-month horizon that, for you, is almost invisible and it’s very uncorrelated. And that’s the reason why, over time, they will be uncorrelated.

So, you might be a situation where, right now, you are long equity and the CTA is long equity or this micro hedge fund is long equity. But that risk is going to change so much, and in a month’s time they will be negative. So, don’t worry about it too much.

The way you should think about that part of the… If you’re thinking about it in terms of alpha, that alpha is almost distinct to what you are doing from an allocation perspective. And that’s why it is, A, uncorrelated to what you’re doing and, B, is very additive to your portfolio. And I think that’s kind of the way that a big allocator should think about that.

And the actual positioning right now, you would use it in terms of modulating some small changes in your long-term allocation. So, like the trading that you would do from your CMA, you kind of want to modulate it by trend or systematic strategies. But, in terms of alpha, you just think of it as almost an independent play.

[01:04:07] Alan

Yeah, yeah, well it’s true because that’s effectively what the CTA allocation is doing for you. It is tactically adjusting your portfolio while, as you say, you’re gone, playing golf. It’s increasing your exposure to equities if they’re trending higher and reducing it if it’s trending lower.

[01:04:21] Yoav

Yes, absolutely. So, to be honest, it was an experimental thought process. I don’t expect the allocators or the consultants who are running those CMAs, I don’t think Morning Star, or Vanguard will all sort of shift to my methodology. I’ll be very surprised. But I think it’s important to understand the tools of the trade that you have at your disposal if you’re an allocator.

And you know, we normally think of trend as outside that universe that you’re living in, but you actually might want to consider embedding it into your process. Not just CTAs, but a lot of the systematic system. And, you know, we do embed some systematic process.

So, for example, you will manage your FX exposure. If you are an Australian allocator and you have structural FX exposure in the USD, then you would want to hedge a lot of that risk and you will manage it in a systematic fashion. So, some systematic strategies, like FX hedging, are embedded into your process. And there’s a great paper by Otto Van Hemert about the best ways to hedge FX exposure. So, there’s a really nice paper I can send you later about doing this as an allocator. How do you embed a systematic strategy?

And you know, FX engine is like the simplest one, but already there is a lot of stuff that you can do with it. And CTAs are just another tool of a trade that you might want to think about embedding into your process.

[01:06:16] Alan

Yeah, very good. And what’s the name of your LinkedIn blog on that one?

[01:06:20] Yoav

It’s What’s Cooking, this one.

[01:06:22] Alan

Very good. Good stuff. We’ve gone over a bit on time, but that’s all good.

So, Niels will be back next week. I’m not sure who’s with him, but send in your questions for Niels. Great to have Yoav this week. Thanks very much, Yoav. And from all of us here on Top Traders Unplugged, thanks for tuning in. And we’ll be back next week with more content.

[01:06:48] 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.