254 Systematic Investor Series ft. Nick Baltas – July 29th, 2023

[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.

[00:00:45] Niels

Welcome and welcome back to this week’s edition of the Systematic Investor series with Nick Baltas and I, Niels Kaastrup-Larsen, where each week, we take the pulse of the global market through the lens of a rules-based investor. Nick, it is wonderful to be back with you this week. So much continues to happen between our conversations. Now we have the highest rates in the US since I don’t know for how long but quite a long time. How are you doing? How are you keeping up?

[00:01:10] Nick

I’m doing very well. Happy Friday, Niels. Just looking forward for Summer. It has been a very, very busy summer so far, I would say, a significant amount of planned activity. But yeah, I’m looking forward to break. How are you?

[00:01:22] Niels

Yeah, doing well. It’s interesting because it’s not like it feels like the markets are particularly busy, but I do sense the same kind of activity underneath everything, which is good I imagine.

Now, before we dive into all of the things we have lined up today, as usual, you’re bringing some really academic points or angles we’re going to be talking about. I’m always interested to hearing what’s been on your radar sort of big picture, anything in particular that had caught your attention. It could be whatever you want to bring up really, Nick, before we dive in. Where are you focusing your attention at the moment?

[00:02:09] Nick

Sure. I think client activity is the one that typically determines how we become busy, but also, I think it’s a proxy of how the market performs as a whole. So I think last time we spoke a couple of months ago, it was this pendulum between being more defensive or risk on, this puzzling dynamic since the beginning of the year. I think more recently, at least I personally feel more of an optimism, conservative but yet optimism. We start seeing a few more risk-on trades or some of the more defensive-oriented trades. We see clients wanting to reduce, if you like, the cost of carry or possibly focusing more into the deep tail.

Obviously, with the VIX being very subdued, it’s a big question mark as to whether we see in the real market what some of the fundamentals seem to be suggesting. So that’s the first point I would make. I think some of the actual activity on the trade side, I remember a few months back, we discussed some of the work we have done over the years about dynamic allocation between economic regimes and how beta assets perform across different inflation, growth dynamics, growth regimes. So we see some interest, if you like, in those beta models that incorporate economic regimes.

In our world, which is obviously the trend following/cross-asset world, this is still top of mind. We see more and more activity, I should say, since the beginning of the game, both in terms of asset growth but also in terms of client requests, analysis, and so on and so forth. Some of the work we’ve been doing recently looks into different sources of premia in the cross-asset world. Back I think in the first time that we spoke with Rob, back in November, he was asking me some of my work on skewness.

So this is something that we also spend a good amount of time. Is that trend following plus something? Probably, versus nothing, I think Jerry mentioned last week. But this is certainly a place that we’re looking to bring some value, if you like, both on the design phase but also on the trade side. Trend, as you know and as you keep on reiterating since the beginning of the year is kind of lukewarm year to date. Some good months, some not as good months, but it’s an average with barometer of 30, 40-ish or so. So that’s where we’re spending our time, but as you said, very, very busy still. It’s the last business day of July. Is it the last, or is it Monday?

[00:04:48] Niels

It’s not quite. Monday is the last one, but there we are. It’s interesting you say that. Of course, right now, I was going to update people on the trend barometer, which actually has gone up to 48 by close of business yesterday. Yeah, bring it on. We’re almost there, right? But it’s interesting you mentioned this thing about lukewarm trend following returns this year. I would kind of turn it around and say actually, I think it’s been extraordinarily good because, and this is a well-known thing that happens, and that is often when you have a really strong trending environment as we saw last year, there tends to be large givebacks, but I haven’t really seen a lot of that. A few exceptions. Okay, fine, but generally a flattish return so far for the indices, some managers up 4% or 5%, which I think is a fantastic effort.

Obviously, equities almost making new all-time highs in the US. In Europe, some of the markets are making new all-time highs. Actually, this is kind of a good example of why people shouldn’t always think about this as a crisis alpha investment because we can certainly do well even when there are no crises. But we’ll come to all of this good stuff. I’m getting ahead of myself.

[00:06:07] Nick

The good outcome of the year is that obviously the March event, which was largely idiosyncratic and largely impacted performance across the industry on the CTA side has been largely mitigated. It’s basically wiped out from the year-to-date performance. That’s a good thing obviously.

[00:06:27] Niels

That’s a great thing for sure. As I mentioned, 48 on the trend barometer performance so far. This is as of Wednesday because we’re recording Friday too early to get the Thursday numbers. Down about 50 basis points for the BTOP 50, down about 1% for the year. SocGen CTA down about 1.6%, down about 1.6% for the year. Trend down about 2% for the month of July, down about 2% for the year, and the Short-Term Traders Index struggling a little bit more, down about 38 basis points in July, but down about 3.6% so far this year.

Obviously, equities having another great month, up 2.2% for MSCI World, up almost 17% for the year. World Government Bonds struggling a little bit with the last few days of yields going higher, so down about 36 basis points as of yesterday for the month and the S&P 500 index up almost 2% and up 18.18% so far this year.

All right, Nick. Now, there is a little bit of a theme today in our conversation. Overall, you could say we will tackle or we will talk about at least single stock trend momentum dynamics, which of course is relevant in light of what we talked to Jerry and Andrew about last week, as well as the next episode in our CTA mini-series with Alan, where he and I also invited Jerry in and talked more about his company rather than his new ETF and his thoughts on trend following in general. But as usual, I will not have read all the academic papers that we’re going to talk about so I will let you guide us when we get to that. I’ll do my best to keep up.

But before we dive into single stocks and trend following, there is one topic that we wanted to cover but from a slightly different angle, because it is a topic that we bring up on a regular basis and it is also a very important topic. But maybe we can shine a light on this in a slightly different way today. It relates to portfolio construction, risk management, but specifically also in terms of how we size and manage positions. So I will pass it over to you and then we’ll see where we go with this topic to begin with.

[00:08:50] Nick

Okay. Thanks, Niels. The topic is about static position or dynamic positioning. That’s what we need to talk about. I was putting my thoughts together the last couple of days as I was preparing for the call. I would probably start from my conclusion, and then I’m going to go back. My conclusion could be incorrect in some form, but I think stylistically is valid. So when we talk about dynamic positioning, ultimately, I think we’re thinking more about a top-level portfolio approach in terms of allocation.

When we think about static positioning, I think it is more about an asset-by-asset assessment that ultimately leads to us forming a portfolio as a collection of assets rather than being a portfolio that includes assets. It might sound quite vague, but as we’re going to go through some of the points I wanted to raise, I think in my mind, dynamic positioning is about a top-level portfolio management, whereas static positioning is asset-by-asset management that you end up holding like a number of assets that can ultimately call a portfolio. But that’s a consequence of you holding the assets, not starting from a portfolio approach.

So why did I get to this conclusion? Let me start from static positioning. Static position sizing basically means I have a number of assets, 10 for the sake of argument, each of which would have a trend signal. That’s not the focus of the discussion right now, but in the presence of a signal, we make a binary choice pretty much saying I’m going to go long whatever is going up and I’m going to go short whatever is going down. Then I would expect that there is some form of utilization of correlation and volatility, at least to size the initial exposures, because specifically when we put different assets together from different asset classes, it makes no sense to just equally weight, for example.

[00:10:52] Niels

If I may interject, I would say I think you’re right about correlation, but not at the time of the trade. I think that’s just from the choice of markets you choose in a static world.

[00:11:03] Nick

So long-term basically. It could be long-term.

[00:11:05] Niels

Exactly. But clearly, volatility assessment and normalization does take place at the time of entry.

[00:11:15] Nick

So I guess in some simplistic form, out of 10 assets, we would do $1 per asset, but we’ll allocate a bit more to those that are at least volatile historically or at the time of the initiation in those. Also, we would allocate more to those that seem to be more diversifying to the rest of the universe, in a simplified term, or it would penalize the exposure to those that are more volatile or possibly more correlated with the rest because they just contribute too much risk so we want to kind of downsize that effect. But that will be just the initial allocation to the assets which we then bring together.

I would suspect we would let the portfolio run. Obviously, the signals will be changing through time, but the allocation would not be affected unless either the signal shifts direction, so from being long, we need to go short, or possibly we get into a loss and some stop losses will kick in. I will expect at the portfolio level, once we build those positions, we also have to account for the fact that if commodities are like 20 and rates markets are 2, we would want to allocate relatively similar risks between the two asset classes rather than the assets themselves.

So I think there’s some sort of top-level or intermediate level, at the sector level, at the portfolio level as to how risk is ultimately split, at least on a maximum basis.

[00:12:41] Niels

Yeah. I will say, I think that is probably a slightly more advanced way of doing it. I think if you just go back to where it all started, I would say the only thing people are concerned about at the point of entry is just if I want to risk say $100,000 in dollar risk, how many contracts should I trade. I don’t think there’s any other thing that is accounted for at that point in time. But I think maybe further iterations of it could do exactly what you said. But I just want to make that clear that, for me, static right now in our conversation today is really quote-unquote the “original” way of doing it.

[00:13:18] Nick

Okay. Fair. I guess in a world of some sort of prudent risk management, there is some expectation that if the asset groups are not fairly balanced, there has to be some form of penalty or benefit from having a more diversified universe within a particular sector or otherwise, but I get your point. So to summarize, there’s a signal on day one which is a binary signal, we allocate on the basis of some longer-term variance and correlation analysis. There’s some stop losses per position, and optionally, some form of risk awareness at the sector or possibly at the asset class level, pretty much.

I think that was what kind of made me think that, hey, static positioning really starts at the asset level. At the time whereby there is no signal whatsoever, you’re just concentrating risk to those few bets that seemingly have a good signal or possibly. Actually, that’s a good question and I don’t really have an answer. If everything has a signal which is not even a large one, how is the allocation done? It’s a bit of a question mark to me.

[00:14:35] Niels

No, I think again, the original way would just simply be to say, well, if I want to risk 25 basis points risk per market, I know there’s a limit to how many markets I can trade. But if I add more markets, I’m going to limit that risk from say 25 basis points to 20 basis points. I think that’s the kind of thinking…

[00:14:55] Nick

Exactly. What’s the dynamic positioning sizing, I guess the story here? There’s no day one anymore. There is a continuous monitoring of a signal which becomes larger or smaller and at times changes direction. We typically utilize that signal to dynamically increase or reduce the gross exposure we have in an asset. I think this is done partly for turnover mitigation in case the signal becomes smaller, and smaller, and smaller and at some point crosses zero.

Partly, it’s also done to be more thoughtful as to where we deploy risk if there is an argument that more trendiness should have more risk in the overall portfolio. So this dynamic sizing from a signal perspective is about making my bets more extreme when the signals are stronger. In addition to that, there’s a risk management angle which says I’m going to vol scale my exposures dynamically, not just at initiation. In other words, if an asset becomes more volatile versus all the others, I’m going to reduce the overall exposure, and there’s a correlation as well adjustment which would again act in favor of the more diversifying assets and how their nature, all else being equal, evolves through time.

Now, there are a few subtle dynamics here which on the vol scaling side might be beneficial for the portfolio as a whole, which has nothing to do necessarily with trend in the first stage and that has to do with assets that typically have negative correlation between volatility innovation and returns. So risk on assets typically, mainly equities, we can argue sometimes commodities, but generally more risk on oriented assets, when they become more volatile, they end up underperforming.

Now, momentum does it by itself, but vol scaling might be quicker at times in terms of reducing your growth before the signal changes direction. So there’s a byproduct here of performance and risk management that can help the overall portfolio, or not by the way. In markets whereby vol is dropping but the asset is falling, this is actually counter-impacting performance. Rates had this dynamic back in the ’70s whereby they were kind of increasing but very gradually with price falling and volatility falling.

So that’s on the volatility scaling side. Whereas correlation scaling is more important for top-level portfolio management in terms of risk. That’s purely a volatility-targeting adjustment to make the overall portfolio hit some sort of vol. Typically, again, we should have some sort of minimum and maximum exposures, some sort of global at the portfolio level min and max exposures, which is primarily driven by liquidity concerns.

[00:17:55] Niels

Can I also interject a small thing here, because one of the many of the conversations that we have on this topic, I hear the somewhat intuitive point saying, well, if you’re hunting for an outlier, you want the static position because whatever happens, you’re going to have your “maximum position” on throughout. Intuitively, that makes sense. However (and I’m going out on a limb here a little bit, I’m sure there’s going to be people who can test this and argue against it or for it or whatever), in addition to what you just said before about often (and of course, we know that one of our guests in the CTA series, Doug from Florin Court, he has mentioned this and I’d love to hear what you think), many people would say that volatility tends to increase around turning points. Okay, fine. But I think there’s one thing we forget, and that is that, actually, when you have really nice, strong trends, volatility tends to fall. If that’s the case, then actually, dynamic position sizing allows you to have very healthy risk on in those trends.

Therefore, I’m not so sure that you can make the simple argument saying, well, clearly, if it’s an outlier, you’re going to make more money if you have static position sizing. I simply don’t think you can make that black and white argument. There will be, in some instances maybe, reasons for that, but I think there’s going to be many situations where actually in a very long trend done correctly, you might end up making more money from that because, one, you might during say correction phases or whatever, you might reduce a little bit, avoid some of the pain.

When the volatility falls again, you might increase, you might get back to a really nice full position, etc., and then again, around the turning point (again, this is in an idealized world), you might scale out much quicker, and therefore, capture more of the trend profit than you would otherwise do in a static. So I’m not saying one is better than the other. I’m just saying I don’t buy into the argument saying, “Oh, clearly if you’re only going for the outliers (because I think we’re all going for the outliers, frankly), then one must be better than the other. I don’t buy into that.

[00:20:40] Nick

I think I agree with you. I agree with you from a different angle. My angle is related to some of my philosophical approaches on portfolio construction design. One of the principles I personally like going for is concentration risk limitation. In other words, because losses are hurting more than gains; as we all know, there’s this asymmetric treatment of losses. Exactly, the Kahneman. Limiting concentration risk and smoothing the returns (effectively smoothing the returns), is beneficial from one in many, one in a million probably, I don’t know, there are very few cases whereby you would have had significant outperformance versus having significant performance.

I think risk mitigation at the extreme level to me is more prudent risk management. But I can certainly understand another philosophy, that being I want to have concentrated bets because these are the ones that pay off. It’s just unclear to me which one of the two, over the longer term, is better because I think the frequency at which those extremes are occurring is too little to make any econometric assessment whatsoever. Precisely because an extra dollar of gain in the extreme is providing less happiness versus a dollar of loss, and that’s the asymmetry that I was kind of talking about.

I always vote for more concentration limit, concentration risk limitation, rather than getting the extreme bet when the bet is there and that’s the only bet out of all. So I agree with you. It’s not clear-cut, but that’s my personal preference. I think it’s ultimately a philosophical point once you’re managing those portfolios. So summarizing on dynamic positioning, I think the objective of dynamic positioning is number one, as I just said coincidentally, mitigation of concentration risk.

The second point is some form of dynamic budgeting of risk across assets or asset groups in a portfolio context that happen to be trending more than others, and I think that’s purely coming from the fact that aggregating positions and therefore naturally having more risk let’s say in commodities because today they might happen to trend more and therefore the aggregation of those budgets that we allocate gives us a number which is greater than rates.

I think there is a theoretical optimality in the portfolio construction if we use some sort of mean variance, then we’re trying to achieve the best portfolios theoretically for using more like a risk-based oriented optimization in the presence of equal Sharpe ratios and correlations we have, again, optimality. I vouch by the way for more risk-based allocations. But ultimately, I think dynamic sizing of a portfolio as a whole, and how we treat a budget of risk allocated across markets and sectors and long positions and short positions, versus an asset-by-asset treatment that simply gives rise to a portfolio whereby each asset in isolation has an exposure with some max loss and some max exposure associated to it.

Having said all that, however, and I think that’s my main conclusion is that we should not disregard the fact that on anything we do when we build portfolios, there is some sort of duality or imperfect substitutes. That’s the very technical term of basically saying that I can easily see links between the approaches and sometimes we need to look through it. We look through those simple design choices and appreciate the fact that a stop loss rule is pretty much what dynamic volatility scaling kind of does. If vol is spiking, we’re reducing the exposure which is pretty much the same thing as saying okay, if that vol spike is actually impacting your trend, then your stop loss anyway would be hit.

Of course, we can make the argument the vol spikes and then you are performing so you’re not keeping a stop loss. But reality is specifically for more crisis alpha and risk-on assets, vol scaling is acting like a stop loss. That’s point number one. So we can make that link and say, hey, I’m doing static positioning with a stop loss. Fine, I’m doing dynamic positioning with vol, but implicitly, I’m saying I don’t want to take too much vol because if the risk is increasing, I’m going to just reduce it. It’s done a bit more dynamically but ultimately, that’s what it’s here to achieve. That is at the asset level duality or relationship.

At the portfolio level, if I think about some sort of a max limit exposure that I do not want to have in a static position in context, this is pretty much catering about concentration risk and penalizing for vol and correlation in a dynamic fashion. But again, done in a more kind of static basis, that’s my max exposure, whereas we say, hey, we want to be a bit more dynamic on how correlation and volatility is kind of moving.

So I don’t think there are two distinct at the end of the day narratives. It’s just different perspectives that however, we should accept the fact that there are linkages. That’s why I think if we do two backtests, they are going to look relatively similar, I’m not going to say very similar, but they share commonality.

[00:26:11] Niels

Yeah, I agree with that conclusion, I also want to just add one thing, and that is, even those who use a “stop loss”, in some people’s rules in calculating the stop loss, actually, volatility is one of the parameters. So, therefore, the stop loss does react to changes in volatility. It’s just not called dynamic position sizing. It’s just dynamic risk sizing you could say, so I think that’s important to mention as well. I agree with you. Certainly, I don’t hope I’ve ever said that one is better than the other.

I’m kind of pushing back a little bit and I’m sure the listeners who I’m pushing back towards, those who will say categorically no, no, no, no, no, one entry, one exit, one stop loss is better, I just don’t think you can say that and I don’t believe in it. I know the numbers don’t support that either.

So I just think that that’s important as I’ve always tried to promote at least, and that is, I celebrate the fact that they are both good and have done well for decades because it means for me at least that the underlying methodology of trend following and the underlying quote-unquote “golden rules” of trend following are the ones that actually keep this strategy so relevant, so potent, and so robust through different cycles. I’m glad you wanted to talk about this. Obviously, I’ve been beating this topic for a while, but now we have your perspective as well, and I think that’s great.

[00:27:58] Nick

One last point just to conclude, because you kind of reminded me that obviously Jerry was speaking last week. He started discussing about how he thinks about trend following. He said, listen, we want to do trend, and then just trend plus nothing. I actually like the title by the way. But anyway, he said trend. Then there’s a portfolio construction angle which I think is important, but I’m not going to comment on that. But one thing I’m going to say is the following, whatever portfolio construction technique we might not use, unless the signal is informative, we’re not going to get the performance right.

So I think ultimately, the signal itself is the one that is driving returns and the portfolio construction is the one that is more about how prudent or not we are about portfolio construction risk and risk allocation and volatility targeting and so on and so forth. But genuinely, for a strategy to perform, the ultimate link is that the signal is well-designed and identifies the trends. That’s the reason for it. No portfolio construction can go against a signal or can recreate performance out of a non-existent signal. That’s the point I wanted to make, which I think makes the two relatively similar at some point, because the signal is there.

[00:29:13] Niels

I think, yeah, they’re going to be different in terms of performance year by year, but if you look over a 20-year period, they’re not going to be vastly different because of that decision. They can be vastly different because of the quality of the signal, which is different. That’s the point. Because performance is of course not similar. But anyways, I think we’ve talked enough about that. Now we’re going to wear the professorial hat that you wear sometimes at Imperial.

We’re going to talk about another topic actually that was relevant to last week’s conversation, which is single stocks in trend and the difference between trend and momentum. It’s one of the papers that you found. I think it’s back from 2018 and I think- no, this is not the AQR paper. This is the cross-sectional and time series test of return predictability, what’s the difference paper that I think you want to dive in first. So as I said, I have not read it so I’m going to pass the baton over to you.

[00:30:20] Nick

Sure. Why did I think of this one? Obviously, Jerry mentioned single stocks and trend following. Then that kind of reminded me of this work by Amit Goyal, who I happen to know. He’s a great academic. Obviously, the other co-author is Jegadeesh, who was one of the two that came up with single stock momentum back in the ’90s, and single stock reversals as well. So a very good academic pedigree going into this piece of work. That is indeed a 2018 paper.

It basically comes up and says over the years and decades, we have been knowing that single stocks exhibit cross-sectional momentum, and I’m going to get to the definitions in a second. But very recently at the time, that was like 5, 6 years from the time the paper came out, that was 2012, 2011, AQR has obviously come with the very well-cited paper now that made time series momentum, in other words trend following, something that academics started looking at. Even if trend following at the time was primarily studied in academia outside of single stocks, it is easy to make the connection between a price continuation pattern applied either in a cross-sectional fashion or in a time series fashion.

So they basically came up and said, okay, why are they different? Are they different in the first place, and if they are different, what makes them different? How we can bridge the gap, should we bridge the gap, which one is better than the other? That’s basically the key focus of that paper. But because it starts from single stocks, it reminded me of the discussion we had last week, and I was like okay, it might be worth bringing it up and summarizing it.

I grew up thinking that momentum is a purely cross-sectional dynamic. So you have a number of assets, you rank them on the basis of some historical performance, you buy the winners, you short the losers, and then you just get on with this either dollar neutral portfolio or risk neutral portfolio. It’s typically done dollar neutral if the volatilities are similar.

Obviously, what is important with this portfolio is that the bet is on winners outperforming the losers. It’s not about the positively trending assets generating a positive return, negatively trending assets generating a return, I’ll come back to that point. But it’s more about a cross-sectional outperformance in a market that everything is going up, you still have winners, you still have losers, in a market that everything goes down and you still have winners, you still have losers. So at any point in time, you build a balanced portfolio.

Conversely, once we look into what academics call time series momentum, which is just trend following, you just listen to the past performance of the asset and you don’t really care about the rest. Basically, say if my asset had a positive day, I’m going to buy, if it had a negative day, I’m going to sell. If everything is going up, I’m going to buy everything and if everything is going down, I’m going to sell everything. It’s only at times that my portfolio is kind of balanced and some are positive, some are negative, that it is going to start looking more like a cross-sectional momentum portfolio.

So ultimately, there are links, but also significant differences among the two. The differences have to do with in a very, I guess, simplistic way how much dynamic beta I end up having in my trend follower. Sometimes I’m net long, sometimes I’m net short, even if there is some cross-functionality, specifically when there is dynamic sizing, by the way, vis-à-vis being always “dollar neutral”, or actually, dollar neutral always, “market neutral”.

That’s a separate conversation as to how much market neutral you are by going long at the top, short at the bottom. They look into those differences in the single stock space. Firstly, they say, well, if we do trend following in single stocks, yes, it works, and it works even if you follow a one-month speed or a 60-month speed. So there is this type of embedded market timing dynamic across single stocks in the same way as it exists with other asset classes.

But if you were to do momentum, as in buying winners and shorting losers, you only get positive returns when you have like a medium-term lookback window. Very, very short-term windows will give you reversals, very long-term windows will give you reversals which are more associated with value. So there are some differences there. Now, what I wanted to mention is that the differences between the two are important to be understood because ultimately, they drive the philosophical approach to building portfolios. So it might become a bit technical for a minute, but I think it’s worth doing it.

Cross-sectional momentum is driven by three dynamics, whereas trend is driven by two dynamics. Let me start from trend because it’s easier. Trend performas well if there is serial correlation, whatever goes up keeps on going up, whatever goes down keeps on going down, or if your universe of assets has unconditionally very large either positive or negative expected returns. So if I tell you here’s an asset that 90% of the time historically was going up, if you were to buy it, yes, you’re going to make money.

If an asset was going 90% of the time down and you’re shorting it, you’re going to make money. I think we’ve done this, we’ve had a similar discussion a few episodes back, probably three, four months ago. You can still have no serial correlation and perform because you’ve been for example buying bonds and they have been appreciating because there’s positive carry, whatever. So trend can be driven by those two things, either serial correlation in the returns, or over the longer term, large unconditional returns, positive or negative, whereas momentum has three sources.

The first one thankfully is the same between the two. If you’re going up and you continue going up, and you’re going down and you continue going down, momentum cross-sectionally is going to work. The second one, that’s I guess a trickier one. That basically is lead-lag cross correlation. If I’m winning today, I’m predicting you losing tomorrow. If you’re losing today, you’re predicting me winning tomorrow. So there is lead-lag between assets, and that’s the cross-sectional lead-lag dynamic that makes momentum as a cross-sectional feature perform.

The last thing, which is kind of similar to the trends one, when you have the universe not with high unconditional returns but high dispersed unconditional returns, it means that over the longer-term, you have some big winners and some big losers, so statically buying and shorting them, you’re making money. So there you are with two plus three reasons of existence, the first one being common, one in momentum being very idiosyncratic and then the last ones are much more longer-term and conditional. I don’t think this is really as important for the conversation we’re having.

So they go down the route of really understanding those drivers, and ultimately, where they get to is in the following summary if I try to summarize it well. When you build trend following, you have dynamic data embedded. Sometimes, you’re net long, sometimes you’re net short, and this has value historically. What they do is the following. They say, how much net long or net short am I in trend following? If I take that net short or net long and buy let’s say futures and I add that into my cross-sectional momentum, I kind of start bringing them together. Because the cross-sectionality is always there. The question is how much net positive or negative timing ability your portfolio has.

By bridging that gap, their point is it’s not that one is better than the other. Obviously, they run some sort of tests, they have a bit of bias in favor of cross-sectional momentum rather than trend following. But they ultimately make the case and say, well, we should not be comparing the two at face value because one has dynamic betas that are purely driven by the way of the portfolio being put together, and the other one is kind of dollar neutral. Those beta exposures, first of all, they seem to be rewarded, but secondly, make the comparison between the two unfair. Once you account for that in the overall portfolio, performance is very similar.

Anyway, a very long-winded answer or summary if you like to effectively say that yes, trend following in single stocks can work and it does work, and has been shown at least in academia to work. But the point I would make, and I think Andrew made the point last week, there is a dynamic beta component there which I think is going to be much more biased on being long. Obviously, the question is, how much value does the short bring?

I think last week, Jerry said not much historically, but I genuinely feel that there is some value to be brought around market corrections. The question is whether the cross-sectionality brings value. On that, I haven’t specifically done much work to be frank with you.

[00:39:38] Niels

No, that is interesting. In some ways, I think our conversation with Jerry is- well, we’ll have to see how it goes. But on the other hand, I would say, well, we should already know because the underlying strategy he’s employing now in the ETF is not new. He’s been doing this for a while. So we kind of should be able to just study and say, well, for example, during the bear market of last year 2022, did it actually help to do it this way versus the other way and so on and so forth? So, yeah, people can look these stats up of course, but it is interesting.

One thing I just want to add, and again, since I don’t sit and crunch numbers, I’d love to maybe hear your thoughts. These things do change over time naturally, of course. But I don’t recall necessarily equities, certainly from an index point of view because most CTAs trade the index, I don’t recall them being the best sector in the trend following world, even though it’s probably where most certainly DIY trend follows would start saying, oh, yeah, I’ve heard about this strategy trend following, I trade stocks, and yeah, I should apply trend following to stocks and it’s going to be great.

I don’t know about that, frankly, because I always felt that one of the secret sauces of trend following is the fact that you apply it on so many different types of assets at the same time, not just a handful of different equity indices because of correlation. Now, of course, if you have a big universe of stocks, maybe some of that is alleviated because the correlation will be lower than maybe between just a handful of equity indices. But we’ll follow it. That’s all I can say.

[00:41:41] Nick

Two points I would make. It is likely that some cross-sectional momentum is brought in by the fact that instead of doing allocation at the index level, it effectively neutralizes in some respect the sector biases. You allow the cross-sectionality. So to the point I was making on those academics trying to bring the two together, they said, listen, I can do my trend following being very similar to cross-sectional momentum plus my dynamic beta. In a way, that’s all they’re saying. So dynamic beta is basically what the index will do.

I think what’s happening now is bringing in some cross-sectionality or some cross-sectional momentum in what otherwise is an index trend. I think that’s probably some of the value. The second point I would make is that just going through those papers and reading through, it reminded me a bit the discussion you have had with Bruno from Quantica, because I think he was talking about not single stocks, but he was talking about spread portfolios between markets and doing trending in those spreads.

So in a way, as a philosophical approach to design, it’s similar to cross-sectional momentum because you’re now looking into out and underperformances between markets. So that’s the other thing that it reminded me of.

[00:43:09] Niels

There’s another dynamic in terms of I think challenge. I say challenge, maybe it’s not a challenge. But we were actually asked the question by one of our listeners, Oliver, the about if you have- even if you trend follow single stocks, given the fact that there is probably a long bias in long-term trend following in that, are investors prepared to pay 1%, 1.5% fee for being “utilizing a large part of the risk budget” to that specific sector? That’s a different question. Nevertheless, it is somewhat relevant, but we’ll see.

Let’s move on just for the sake of time. There was this paper that I mentioned actually also, I remembered it from AQR, which talks about single stocks, single stock factor trend, as far as I remember. This is also a paper that’s a few years old from recollection. Do you want to tie that into the conversation?

[00:44:22] Nick

Yes. So there’s obviously trend at the asset level, whether that is single stocks or I guess global macro assets that we have kind of discussed. Then obviously, that led me to remember, as you said, there’s this AQR paper, 2019, I believe, published in Journal of Portfolio Management. But also more recently, another paper came out last year in Journal of Finance and another one that is going to be published next month in August, all of which are talking about this factor momentum, this factor trend. So here we’re still in the single stock space.

We’re looking into equity factors, namely, value, quality, low beta. All these are cross-sectional factors. You rank the universe of stocks on the basis of some fundamental or price-based criterion. You buy the top, you short the bottom. Whether you do it with an optimizer or you do top this cell, bottom this cell, that’s less of an important point for our conversation. It’s a critical point from a design standpoint, but not for our conversation. They make the following hypothesis.

They say, well, is there momentum in the factors? So can we argue that a factor performing well predicts outperformance for the same factor for the next month or two? So they run some analysis, some sort of serial correlation analysis, some lead-lag regressions, I’m talking about the selection of all three papers, and the answer is positive. In other words, for the large majority of factors, there is some sort of positive serial correlation, possibly shorter-term than what we see in single stocks. So it’s not like a 12-month cycle. It’s more like a 3-month cycle, for example.

But they document that in the first instance, saying, hey, empirically, if value does well, kind of continues doing well for the next month or two, or if carry does well, it kind of continues. Then some of those papers go one step further and they even bring forward a quite radical in a good way hypothesis, that being that there is no single stock momentum. All there is is serial correlation in factor portfolios that happen to give rise to, if you select positively, if you select winners, and if you short losers, you’re going to outperform.

Because if let’s say value is outperforming, then it is likely that high-value names will be flashed out as winners. If quality does badly, it might be that high-quality names are the losers. Therefore, the serial correlation or the factor level is giving rise to equity momentum. Then they go even one step further and they say, well, if equity momentum has historically been crashing, and we know that equity momentum tends to crash, they effectively say, well, possibly, these are the times that the correlation, the positive correlation between factor returns is breaking. So they create an autocorrelation factor, not a factor to be invested, but a factor to monitor.

Whether that is kind of predicting momentum crashes, they say, okay, yeah, it’s coincidental with momentum crashes. So where do we go with that? In my perception, that’s now my view and I’m not quoting anybody else, equity factors for years have led academic research. I think a good amount of what we do in any systematic business is obviously inspired by those findings and then expanding across asset classes and learning about portfolio construction and so on and so forth. But they came to a point that factors themselves became entities.

So if I talk about oil price today, I can equally talk about value. But the former is an asset, it’s a commodity, the latter is a portfolio of seemingly behaving assets from some fundamental characteristic perspective. But the fact that they become entities and now we can have institutions as well as retail investors going out and getting an ETF that is proxying for equity value, for example, or equity equality, can create to some extent or can have consequences of behavioral patterns being brought at the factor level in the same way as historically were brought at the single stock level.

So if we used to say, hey, we digest information in a rather slow pace and therefore we impose onto prices momentum, then at the time that those themes or strategies or factors become commoditized, there is a hypothesis to be made that these now factors because they’re accessible at a format that was not possible otherwise. Ten years ago, you’d have to run an optimizer between 2000 names. Now you can buy an ETF. It is likely that institutionalizing the access or democratizing the access, we’re creating price patterns at the factor level. That’s the point I would make.

I think this momentum in factors could also be the artefact of the evolution of the offering and the access. It’s no surprise that I believe some of the CTAs are not utilizing factor momentum in their broader CTA offering. So this is not a single stock momentum anymore, it is factor momentum. So it somehow embeds the factor of selection already before we deploy longer or shorter exposure on that. So that’s a summary of the factor momentum, still in a single stock but in a different lens; subgrouping, basically.

[00:50:18] Niels

Yeah, exactly. Very, very good. Very interesting. Thank you so much, Nick, for taking us into the academic world. I’ll bring you out of the academic world now with one of my more simple topics just before we wrap up, also inspired actually from the really good conversation we had with Jerry last week, because ETFs, and I don’t know if you wrap some of your products within ETFs, but ETFs obviously have become incredibly popular. I’m not an ETF expert, but clearly, the ease of access is one of the very compelling points.

Of course, I don’t know if the same can be said about mutual funds in the US. I’m not entirely sure. Maybe ETFs are even easier to access. So there’s a lot of excitement about CTA strategies being wrapped in ETFs. Clearly, Andrew’s success last year was a big inspiration I think for people with the growth they saw. Obviously, there’s two sides to this coin, and I think this is the point that I wanted to bring up with you, you’re bringing in largely a completely anonymous audience into a manager strategy by going the ETF route.

We as managers have no idea really who the underlying investors are unless we’ve kind of already talked to them and know that they will go the ETF route. But generally speaking, it certainly opens up the avenue compared to say a fund where they have to fill out a subscription document and the manager usually sees who goes in and out.

Of course, the excitement on the manager side or certainly on Jerry’s side was also the fact that well, he believes that it’s going to make it more appealing for some investors because if they don’t like the product, they can just sell it. They don’t have to make it known to the manager that they’re actually not so happy. I understand that as well. I kind of understand that as well.

However, my concern, and I’m thinking more like this on an industry level, I’m concerned with putting and giving investors, at least certain types of investors, daily liquidity in a strategy that really needs to be held for 10, 15, 20, 25, 30 years. I think that when you do that, first of all, if the investors are unknown to the manager, the manager has no way of truly educating and getting to know the investor.

As I told you offline, and I also mentioned this in my conversation that I did yesterday with Alan and Jerry for an episode we’re going to release very shortly in the CTA series, I said that I listened to a conversation with Seth Klarman, the super successful founder of Baupost, the hedge fund, value-based hedge fund, where he talked about that he felt that one of the most important aspects of their success was that they knew their clients so well that they had educated them well before they even got into the product. So there was no misalignment, meaning there’s not going to be a year where the client felt, oh, this was a bad year, and Baupost people felt no, this was a great year because we did exactly what we set out to do. I think that’s a truly valid point.

I’m thinking of my own situation where I would say, I could probably call up 99% of my clients and we know each other. I think it’s incredibly valuable. Because especially in a strategy where we go through these potentially long periods of flat or negative performance, it doesn’t mean the strategy is wrong or doing something wrong or it’s broken or anything like that. It just means that it’s not in sync with what’s going on in the markets or you could say the markets are simply not trending.

But if you’re an investor who just sees a daily NAV and you don’t really know what’s going on inside it and, actually, I also heard, I think he’s called Dan Ariely, the behavioral economics author came out with a new book recently talk about the importance of transparency. He had some kind a fancy term for it. But it’s to do with the fact whether you know what’s going on inside a product. If you do that, you are more likely to buy it and hold on to it. I think it’s the same with our world that if people really understood trend following, which is why I think a lot of managers do a great job in making themselves very available to the public and talk about trend following, they’re more likely to stick with it.

I’m not sure there’s a real question in here, I’d love to hear your thoughts about it. But I’m just generally concerned about our industry embracing certain wrappers that actually I don’t think really serve the investor nor the manager really, not just because it’s low fees, but actually also from the potential turnover of clients they will see in these products. That’s my concern.

I also know there are some pros because maybe some people who would not otherwise get exposure to trend following can get it. I understand that. But those are my thoughts, Nick, and maybe you can challenge some of that or not.

[00:56:16] Nick

The only soft challenge that I would have, I don’t think it’s the daily liquidity per se, because I think daily liquidity can also allow institutions to be a bit more-

[00:56:25] Niels

No, I said depending on the investor.

[00:56:27] Nick

I agree with you on this one. I 100% agree with you on this one. So some of your thoughts I think in my mind just can be summarized by the fact that any coin has two sides. There’s no freelancers we tend to say in investing. You give something for something that you have to pay for or you benefit from. So I think the minute we make some investment profiles available for more opportunistic in a good way, entry or exit, we have to accept the fact that it might become a vehicle that is not necessarily serving the purpose that it has been designed for.

Now, obviously, there are regulatory implications here as to whether we shouldn’t make sure that consumption of financial products is done with the appropriate level of understanding of what we’re investing in. That is not necessarily what we discussed today but is of ultimate importance. But having said that, I certainly share your view on some of those strategies being more long-term strategic allocations for total portfolio completion rather than I’m buying in a share of stock A and a share of stock B in a trend following ETF. I don’t think that this is the purpose.

Whether that is the right format for the target audience, potentially that being discretionary managers, it’s a good question. But I don’t disagree with you on opening up an architecture. Not making it more open but making it more accessible is certainly likely to become a victim of opportunistic bets that are not aligning with investment premise in the first place. Now, how do you achieve that? I think education is the most important angle. But I think there are some, as I said, higher-level or possibly separate discussions to be had as to what can and cannot be in those formats.

I don’t necessarily think that there is… There is certainly risk associated, but I don’t think it’s the ultimate investment purpose that can be met by just going in and out. I think we’ve had a lot of discussion as to how dynamic we should be about those allocations and how instead more strategic, rather than acting as a consequence of some sort of regret. I don’t think it should be, “Oh, last year I did well, I should buy it now,” because that’s not how we should be looking at it. So the question is how it’s ultimately utilized. So I agree with you.

[00:56:44] Niels

Yes. But unfortunately, as you also know, Nick, that’s how most people make the decision, meaning, oh, it did well last year, I better get in.

[00:59:15] Nick

We all do, by the way. All of us have behavioral biases. I tell myself so many times, I’m now a textbook example. I’m doing exactly what I should not be doing. It’s human nature, right?

[00:59:25] Niels

Yeah. I’m not an expert at all in private equity but I think that’s the only thing I can really point that where you’re saying, yeah, that’s a great advantage. You can’t get out.

[00:59:38] Nick

Exactly. Well, I think said in a different way, returns smoothing has some virtue. I think, who is it, Antti Ilmanen, he has been quite vocal. He is a great guy. I think he has taught all of us over the years as to how we should be thinking about systematic investing. I think he has made some comments historically about how return smoothing has more of a behavioral effect than an actual investment effect. Setting aside how risk management should operate, I think return smoothing just fights against behavioral biases, specifically for long-term mandates.

[01:00:18] Niels

Well said, my friend. By the way, we are lucky in the sense that two of our favorite CTA ETF managers, they make themselves incredibly visible and available to the public. We’re grateful for that.

All right, well, if you like these conversations, I would love if you would take five minutes of your time, maybe even four, and leave a rating and review. Just go to your favorite podcast platform and do that. It really does help the show grow and for more people to discover it.

If you have questions for what we do every week here on the Systematic Investor series, you can email them to info@toptradersunplugged.com., and of course, to any of the other series as well.

On that note, we’re going to wrap up the conversation. We hope you enjoyed it. We will be back next week or I will be back next week with Alan, so this will be your chance to send some emails or some questions in for him to tackle.

With that, Nick and I are going to say thanks so much for listening and we do look forward to being back next week. Until that time, take care of yourself and take care of each other.

[01:01:37] Ending

Thanks for listening to the Systematic Investor podcast series. If you enjoyed 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.

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.