What Happens When AI Starts Trading the Markets ft. Rob Carver

[00:00:01] Intro

Welcome to Top Traders Unplugged. In markets success doesn’t come from predicting what happens next, it comes from being prepared for what you can’t predict.

In each episode we go deep with some of the world’s most thoughtful minds in investing, economics, and beyond to understand how they think, how they prepare, and how they decide, and the experiences that shaped how they see the world. No noise, no short-cuts, just real conversations to help you think better and invest with confidence.

[00:00:35] Niels

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

Rob, it’s great to be back with you this week after hopefully a long, relaxing summer holiday. How are things in the UK?

[00:00:55] Rob

Yeah, it has been a long summer, and we were having a discussion before we started and my brain’s really not got into gear yet after having such a long break and not thinking about markets or trend following or programming or anything for a few months.

But yeah, it’s been a hot summer. The last few days it has started to rain again, which is very welcome indeed. Although it wasn’t particularly welcome for me on Tuesday when I went on a very long bike ride and got very wet. But yeah, swings and roundabouts. So, yeah, apologies if I’m a little bit slow during this podcast because as I said, my brain’s still not quite in gear yet.

[00:01:31] Niels

No, I’m sure you will. I’m sure get into the speed of things very quickly. Now, we’ve got a, you know, great lineup. Again, thanks very much for sending over some papers that I certainly had not noticed. So that’s great.

But before, and I’m sure you remember this, Rob, before we do anything else, we talk about what’s been on your radar the last couple of weeks. Now, granted you’ve been away, but maybe something did catch your attention.

[00:02:03] Rob

I mean, I think we both started in finance in the fixed income markets.

[00:02:07] Niels

We did.

[00:02:08] Rob

So, I think we’ve both got a sort of a bit of a love for the bonds, you know, a bit of nostalgia about when we were pumping out, you know, throwing around yield curve trades and whatnot. And it’s very interesting to see that normally the bond market gets nowhere near as much coverage as the equity market does because the equity market is cool, and sexy, and interesting, and the bond market is just boring maths. Right?

But I’m very pleased to say that bonds are getting a lot of coverage at the moment, although not necessarily good coverage. So, people are very worried about, particularly about government debt, although corporate debt as well, particularly by the sort of AI complex. And they’re issuing a lot of debt to fund all of these things that they’re making.

But people are very worried about government debt, especially in the UK is particularly worried, and they’re also very worried in the US. And I see that the current leader of the US is promising to give everyone some money if the Republicans get re-elected. I’m not going to comment on the politics, or the ethics, or the morality, or the legality, of that, but economically it seems a pretty interesting idea given they’re already quite a lot in debt in the US. So, all this is pumping up yields and making the life of central bankers quite difficult.

And I noticed just this morning that, in particular, that the Norwegian sovereign wealth fund, which is the biggest fund in the world (basically they have all the money), are proposing to reduce the allocation of government debt in their bonds portfolio by 20%. So, that’s not their whole portfolio, so it’s, you know, from whatever they’ve currently got about 25% in fixed income. So, it’s 20% of 25% which is, you know, sort of 5% obviously, which doesn’t sound very much, but 5% of an enormous amount of money (which is what the Norwegian sovereign wealth fund has), is going to have quite a big impact.

So yeah, people are worried about interest rates, and worried about inflation, and worried about government spending and about existing stocks of debt. And so, it’s cool to see fixed income in the markets news again.

[00:04:27] Niels

Yeah, I mean, you did send me a quote, and I just want to make sure we are looking at the same article because in the quote it says last week it proposed chopping the weight of government debt in its bond portfolio from 70% to 50%. So that seems like a little bit more than 5% reduction.

[00:04:45] Rob

So, their bond portfolio is about 25% of their total portfolio. So, it’s a 20% cut in 25% which if my math is correct means it is a 5% cut on the overall portfolio. But yeah, 5% of US$2 trillion and change is, you know, it’s still money. It’s a bit of money. It’s a bit of money.

[00:05:06] Niels

It’s a bit of money.

[00:05:06] Rob

It’s US$100 billion or something.

[00:05:07] Niels

I mean, if they’re selling bonds, I know someone who’s buying bonds because it came out that Bessent is boosting his buybacks. So, on August 19th he came out announcing an expansion of the treasury buyback program in the long end. And then yesterday, I think it’s yesterday, we are recording on Thursday, I think yesterday he unveiled a 6 billion buyback operation in the 10 to 20-year sector.

And from memory, I could be wrong here, but from memory, I thought it was a little bit of lower amount that he mentioned only just a few weeks ago. So, I wonder if this is just the beginning, and when he was saying also that well, we could always use the TGA account that’s got US$1 trillion in it, maybe he wasn’t joking even though that would be certainly, I’m sure, putting the bond vigilantes in a bit of a tailspin.

But it is interesting and, of course, bond yields (I mean you know this better than I do), I don’t see bond yields coming down a lot at, the moment, despite all these commitments being mentioned. But the signaling of this, which we’ve been discussing recently, actually, on the show, is interesting and obviously should not be taken lightly.

So, and of course we know, from the CTA world, that when you get big trends and bonds that’s usually kind of an interesting time for trend followers because it is a big part of the overall portfolio. Last time we saw it was late 2021 into say early 2023 where bonds really were the main driver of returns. They haven’t really been for the last couple of years. In fact, they probably detracted. It’s been a difficult sector. So maybe we are up for, you know, some interesting trends in bonds. Maybe not interesting for the one trying to keep the bond yields down, but interesting for those who can be short.

[00:07:14] Rob

Yeah. I mean US$6 billion in the US bond market, it’s not even a drop in the ocean. It’s like. a drop inside a drop in the ocean. I mean we’re talking about relatively tiny amounts of money.

Full disclosure, I am short US bonds, specifically 10-years. I’m also short term Korean 1- years what it’s worth. So.

[00:07:34] Niels

Yeah, actually Robert, I think most CTAs are.

[00:07:37] Rob

Yeah, I just thought I’d disclose that.

[00:07:40] Niels

Yes, of course. Now, so, on my radar, something I mentioned last week, but there was a little bit of coverage, actually, since then that I had not seen at the time. I’d only seen a few comments on X about it. But it is relevant, I think, for our audience, and that is what’s going on in the Simplify managed futures strategy with the ticker CTA where it now seems to be clear that the last few months, at least according to an article that was written by Justina Lee over at Bloomberg (and kudos for her to picking this up because I don’t think it was announced much, at least until recently). But clearly the behavior of that fund is really what I started paying attention to in terms of performance the last few months, pretty much after the Iran war started.

And as far as I can tell, it seems like, one, the sub advisor has now been removed. I mentioned that because it’s relevant, to some extent, because we just had the sub advisor on the show only a few weeks ago and there was no sign of any change in that sense. And in the report of this story, it seems like the manager… so Simplify basically stepped in and took some decisions based on what, you know, exposure should be rather than following the signals that Alsys was providing.

And the reason I bring it up is, you know, they can do, I’m sure, what they want. I’m sure they have the basis for that. But I find it interesting in the sense that the CTA space has always been known for its discipline, its systematic application of rules. And I think investors kind of rely on a process that’s what they’re buying. So, when things like this happen, to me it’s unusual for sure. And of course, I think it leaves investors in a difficult spot because they don’t know necessarily what it is they bought if things are changing.

And you know, we’ve heard, I mean, over the last few years we’ve heard a lot of good things being promoted by people who like ETFs. And you know, even though I have my skepticism against some types of ETFs, I think overall, for the smaller investors, I think it’s certainly a good thing. However, I think this is an example of an area that people should be, you know, aware of that there are also some downsides to these structures. For example, the underlying strategy can change, you know, in a heartbeat and the sub manager that you thought you were investing with can be replaced. So, this is something that I would argue probably wouldn’t happen if you’re investing in a fund that is managed by the actual advisor.

Now I know there are also structures out there where being a sub advisor is, you know, perfectly fine and you shouldn’t be worried about it, of course. But anyways, in this balance of pros and cons about these structures, I do think it’s worth, you know, highlighting that it’s not all pros. There are definitely also some challenges with these structures and I think we’ve seen it play out.

Luckily for the investors, it looks like the intervention that has been done, which is predominantly to go long energy, so far, at least the last few weeks has done well, despite, of course, the big selloff we saw where performance was very challenged at times. But right now, investors seem to be in a good space. But it is the second largest CTA ETF. So therefore, it does, I think… what’s the word I’m looking for here?

[00:12:01] Rob

It matters, right?

[00:12:02] Niels

It matters. It really does matter. Yeah.

[00:12:05] Rob

And it’s not just the fact this is second largest. It has the ticker CTA.

[00:12:09] Niels

I know.

[00:12:12] Niels

How can we not talk about it?

[00:12:14] Rob

Exactly. It’s like the personalized number plate of ETF tickers, you know. I think this is, I mean, just to say, all of my information on this subject is based on an article on Seeking Alpha. So, I’m assuming the information in that article is correct and if it isn’t, I apologize.

But it does seem that they’ve changed the thing very dramatically. According to this article, they’ve gone from having over 50 markets, which is a kind of reasonable amount of diversification. And that’s one of the reasons why we think that trend following, as a strategy, and these other fairly low Sharpe ratio strategies make money, is by lumping them together and diversifying across lots of markets. And that seems to have gone. I mean, as you say, it’s basically an energy fund now. Those are the only positions they’re holding.

And the other thing is, at the moment at least they don’t appear to have any short positions. And maybe that’s related in the sense that if energy’s going up, you should be long energy, therefore you don’t need any shorts. But it’s really worrying. And you know, if I held this particular ETF (which I don’t), and I was holding it to get sort of a proxy CTA exposure (which I’m not), but you know, I could see myself doing that in the future, I would 100% sell it now because it no longer ticks that box.

And you know, and my experience working in institutions is, you know, investors are very concerned about style drift to the point where actually it was, in many ways, an inhibition on research and innovation. Because when I was first hired into AHL, I was hired to basically create a new strategy and we were unable to get a meaningful allocation to that because the investors were like, no, no, we just want trend following, we want diversified trend following. That’s what we’re buying into. You can put a little bit of your weird stuff in if you like, but don’t overdo it. So, you know, whereas, as you say, in this ETF structure it appears you can just one day wake up and say, today I will no longer be, you know, a managed futures ETF, I’m going to do something else instead. And that to me seems, I mean it just seems nuts. It really seems nuts.

[00:14:34] Niels

Anyway, well, let me soften it a little bit and just say that I think it would have been better, in my view, if things had been communicated ahead of time, ahead of any changes. You would have gone out to the investors and said, listen, we are planning to do this in a month’s time and so, people have a chance to vote with their feet, so to speak. Anyways, I’m not going to say what they should do, but I think that’s how, normally, things would go, and certainly if you make such a big change, that’s for sure.

Anyways, just moving on, the other headline I saw, also about our space more broadly speaking, is just an incredible number, and that is that Millennium apparently is nearing US$100 billion under management. So, that is quite staggering and, you know, it is quite amazing.

I mean, I think both you and I, and many people listening to us, and certainly a lot of the co-hosts, we know how difficult it is to raise money but they seem to just be, you know, pressing a button and say, oh yeah, we’ll take another US$10 billion or we’ll take another US$20 billion, and the money flows, so, must be very nice. But of course, they work very hard at delivering an extraordinary track record. So, kudos to them.

Now, trend following is actually having a pretty good start to the autumn, so, a very strong August, as we talked about last week, and September isn’t too bad either. My trend barometer finished yesterday at 57. That’s a strong number and I think it seems pretty much in line with the numbers that I see, which we’ll come to in a second. I’d love to hear your thoughts on kind of just overall broad-based environment, trend, etc.

But what I am picking up a little bit from looking at various data is that maybe, maybe momentum in the equity sector is weakening a bit and that could be interesting for many reasons, of course. But of course, actually, equities have generally been one of the bright spots, maybe with some more selective equity markets, have been kind of a bright spot so far this year. So, it’s interesting to see if we’ve seen most of that momentum, at least for a while. Anyways, time will tell. We don’t know what the future holds. But Rob, what are your thoughts sort of overall so far this year, just generally speaking?

[00:17:15] Rob

You know, the first kind of few weeks, even, of this year were incredible, and, actually, you know, my performance has basically gone sideways since then, I would say. So, you know, I’m sort of hovering within kind of a few percent of my high watermark. But yeah, always was… And it does seem that, you know, we’re kind of lacking in a, you know… It does feel like one of those environments (and this is purely a qualitative thing, I haven’t necessarily got any statistics or numbers to back it up), it does feel like one of those environments where the market’s kind of tottering on the edge of something. Right? It does feel like something’s going to happen. I’m not sure what. And you know, and that does mean you’re not probably going to make much money in trend.

And actually, the fact that I’m flat, you know, so, I made like 7.50% this year. Most of that was made in the first few weeks and then it’s been sideways. And most of it has been made in equities but, actually, also in energies. But you know, if I look at my risk now, and we talked about my short bond position and I’m still long equities, but you know, none of these positions are particularly big and actually I’m running at probably about a third of my average risk. So, it does, you know, there’s a lot of people who are worried about a lot of things, but the market reactions (except in bonds where we are seeing a reaction), but the equity markets, it does feel like something’s going to happen, things going to go wrong, but the markets are just poised, and teetering, and no firm reaction.

It’s the same in energies. I mean we’ve had this war going on now since the beginning of the year pretty much. Crude oil has been hovering just below US$100 there or thereabouts. But you look at the, so, the US Strategic Petroleum Reserve, and it’s still going down. Every week the numbers come out and they’re down a few million barrels. And although the pace of that reduction has slowed, it’s still falling and it is going to reach a crunch point where they can’t release any more from there. And so, you know, that’s another kind of thing that feels like it’s going to happen.

So, I don’t know, it could be the US midterms are the catalyst perhaps, it could be the SPR hitting the point where it can’t be withdrawn from anymore. It could be some, something that happens related to the new Fed chair who seems to be uncertain about exactly what his policy is except to tell us that, you know, he’s in charge now and he’s not going to tell us what he’s going to do which you know, is an interesting approach for a central bank governor.

So yeah, I’m happy with flat. It does feel like something’s going to happen, and I hope from a trend following perspective that it does. But I can’t tell you what it is.

[00:20:23] Niels

No, of course not, but I tend to agree with you that it does feel a little bit like the calm before the storm. But as we’ve seen, I mean the storm doesn’t necessarily need to be bad. I mean for all I know, I mean, and again I’m not a specialist in this area but it seems like the economy is doing pretty well in many places and companies are making money and AI is delivering and we don’t pay too much attention to all the CapEx, at least not right now, and all of that. So, who knows, it could just end up being a really solid year all round.

Anyways, let’s look at the numbers. So as of Tuesday, we record on Thursdays. As of Tuesday the BTOP50 index is up 1.15% so far in September, up 12.42% so far this year. SocGen CTA Index up 1.16% in September, up 12.41 so far this year. And the SocGen Trend Index up just shy of 1%, up 12.1% so far this year. Short Term Traders Index also doing okay, up 12 basis points in September (so fairly quiet) but the year is okay, up 5.9%, and obviously it runs at a much lower volatility, so, it’s a pretty good result. So overall, I would say for indices, it’s looking pretty good so far. Still a long way to go before the year is over, of course.

For the traditional markets, MSCI World Index down 55 basis points so far in September as of last night, up 12.78% for the year. World Government Bond Index up 20 basis points in September, down 32 basis points for the year. And the S&P 500 Total Return Index down 61 basis points in September, up 12.45% so far this year.

All right, before we jump into the papers, we had a couple of questions that came in. The first question came from Paul, a good friend of the podcast, enjoying himself hopefully, over in Canada. Paul writes, I run a DIY future system that owes a lot to your work. Thank you, Rob, and congratulations on your new book (Okay, so he did the plug for I wonder how much that cost. But there we are). I’m trying to set forecast weights for each rule on each instrument. Rule performance can vary quite a bit across asset classes. Does your recent series of blog posts on pooling say anything about the right level to estimate at? Should I judge a rule on its performance across all instruments or within an asset class? For example, if carry works in bonds and FX but not in equities, should I down-weight carry in equities or is that difference almost certainly noise and I would be overfitting too?

Anyways, I’m sure there’s a little bit for you to chew on here, Rob.

[00:23:32] Rob

Yeah, so over the summer I started writing a series of blog posts about optimization, which I kind of need to restart. Actually, that’s one of the things I need to restart after my summer break. And one of the things I’ve been thinking about. And this is a debate I’ve been having in my own head with other people for the best part of 20 years. This is not a new debate and it’s certainly not new within the CTA industry and probably been most recently revived by debates and discussions I’ve had with our good friend Yoav, who, of course, runs asset class specific funds, whereas I’m, you know, multi asset classes.

And this also goes back to how we structured the sort of research teams back in the day, the AHL days when both Yoav and I were working there, and we moved to an asset class based team. So, you know, there was a fixed income team which I was managing, there was an equities team, there was a FX team, and so on, and so forth. And the idea was that there was something idiosyncratic about each of those, you know, about each asset class, which justified having a team researching and managing them specifically. And there are reasons, there are many reasons why you might want to do that. So, it’s helpful, I think, to have people with domain specific knowledge about markets. And not many people can be an expert on hundreds of futures markets, whereas I think it’s reasonable to say you can have a good knowledge of, say, I don’t know, 40 or so liquid bond futures that there are in the world, say.

But another kind of debate we would have then was whether we should actually, for example, have more or less carry, or faster or slower trend in say, bonds than in equities, than in FX, than in commodities. And that’s actually partly a political question in a way because as people managing a specific part of our portfolio, we’d get very protective about it and say, no, if we’re responsible for the P& L we need to be able to pull all the levers.

And it’s also an empirical choice, it’s also a data choice. So you can actually test to see whether, you know, is it better to essentially fit everything across all asset classes together, or is it better to fit bonds separately, FX separately, commodities separately, or actually is it better to go even more granular and say, you know what, I’m going to fit everything individually, so I’m going to fit a set of predictor weights for US 10-year bonds, that will be different from those for US two year-bonds, which would be different for those for10-year UK government gilts, for example.

Anyway, so the blog post mentioned by Paul, I looked at all of these options and I basically found that you probably shouldn’t fit everything individually because then you really do have a lot of noise and not much signal. So, you don’t have a very much robustness if you do that. Essentially that’s overfitting.

I found that the other two options, so doing everything across asset classes and also doing everything together (so ignoring asset class and just putting everything together), both did pretty well. And so, it’s probably fine to do either of those, I would say, assuming, of course, and don’t forget that I have a lot of instruments. So, if your portfolio consists of two bond markets, three equity markets, one FX, and a couple of energy markets, then I would definitely not fit by asset class because you just don’t have enough instruments to give you enough reasonable amounts of data.

But actually, what works pretty much the best and also, I think, is arguably the most robust solution is actually to say, well, there’s three different approaches here. We can fit everything individually, we can fit them in asset class, or we can do a global fit across everything. Let’s just take the average. You know, essentially that’s like a meta fitting decision, isn’t it? Which of these three strategies should I run: the strategy over on this, the strategy over on this, the strategy run this? Well, I could just put a third of my risk capital in each, and that’s equivalent to just taking the average of those three.

So, that’s actually the approach that I’m going to recommend going forward, I would say, purely because it is the most robust. And you know, that’s kind of saying, well, there are three explanations as to why these things work. Either each instrument individually does something that’s funky and idiosyncratic, or there’s something idiosyncratic about the way that bonds behave versus sequiturs versus FX, or, actually, everything’s the same, we can just treat them all the same. We’re saying, well, we don’t know which of these approaches work best, so we’ll just use all three.

[00:28:48] Niels

Are you planning to change your own strategy to implement that as well?

[00:28:53] Rob

I am, although I wouldn’t imagine the results would change that much, but they’ll change a little bit. But we’re talking about a correlation of about 0.98 or 0.99 probably with what I’m already doing. So, it’s not a big priority.

But you know, I can imagine circumstances in which it would have a big effect. Like for example, as I said, if you’ve got a relatively small portfolio, then it would make a bigger difference. But my approach previously up to now has been just doing everything together.

[00:29:27] Niels

Yeah.

[00:29:28] Rob

So, it’s moving away from that towards bringing in instrument and asset class specific information. But yeah, I don’t think it’s a debate that will ever be settled, but I think it’s quite an interesting one.

[00:29:39] Niels

Yeah, I think it was a great question by Paul, for sure. And it is something that comes up from time to time. I mean, we’ve also made a change in that regard, many years ago, at DUNN, going from individual parameter selection to actually we use global parameter selection, but then you have different models so you get other kinds of diversification in there.

But it would be interesting, actually, if we could do a poll (we can’t), if we could do a poll among the leading managers as to which approach they actually use, that would be interesting.

[00:30:17] Rob

I mean, there are other options as well, which is to do what I call a matrix fit, and that would do things like, essentially, you sort of fit everything together. And that would mean, for example, that you might allocate more to, say, fast trend following in equities, even if it doesn’t make money in equities because it’s very diversifying, and equities are cheap enough that you could trade fast trend following in them.

So that’s almost the third option, which, again, I’ve looked at and found to be in this blog post series. I did look at that and found that it, you know, it makes things very complicated, and it doesn’t really add a lot of value. But you know that that’s sort of the way perhaps a quant coming from outside the industry would do the fitting because, you know, that’s what you do in other sort of quantitative hedge funds, like in equity, long/short.

[00:31:11] Niels

Anyway, I mean, the challenge is as soon as we start doing too much work, we are looking at historical data and we’re trying to build something that will last in an unknown future. So, one should definitely be careful, that’s for sure.

[00:31:28] Rob

My prior is to do no fitting at all, and only if I see strong evidence it’ll add value will I do some fitting. But yeah, that fitting has to be robust. It has to be out of sample. That goes without saying.

[00:31:38] Niels

Yeah. Another question came in because Ian saw that Mr. Carver is on this week (I like the formal approach here). Does he have any views on combining (I know you love this word) discretion with systematic methods and can he recommend a good book that covers this?

But oddly enough, you had sent me, completely unrelated, earlier, a link about a discussion that was taking place on next, which I think is related. So maybe you can combine the two.

[00:32:14] Rob

Absolutely. And actually, if I’m paid money to ask a question, it would be this question because this is actually going to give me the opportunity, Niels, to plug my new book. And you’re very good. I think you allow me to do this once per book.

[00:32:33] Niels

Once per episode.

[00:32:34] Rob

Well, no, I don’t think I’ve plugged this book yet. But I’ve hinted at it. I’ve hinted at it.

[00:32:38] Niels

Okay.

[00:32:39] Rob

But yeah, I don’t think anyone wants to listen to podcasts where people just come on and plug their own books because there are enough of them out there already. And you know, TTU has always been much better than that. But I will spend a little bit of time plugging my book in the answer to this question.

But let’s go back to the X poll first. So, you know, this was the someone… There’s a little bit of a debate that goes on, on X, between quants and everybody else. And for the purposes of this subject, I would call myself a quant, although I wouldn’t necessarily always do that. And they’re basically like, well, you know, you shouldn’t just blindly follow a system. You should use discretion. So, in other words, you know, you basically have all your systematic indicators and whatnot. But then, you know, you look at them and you use your intuition to determine whether actually you should take a specific trade or whether you shouldn’t.

And this particular person on X, and I won’t mention their name, but they had this post and in it was this sentence (and I’m afraid the red mist descending on me when I read it), and it was this: Systematic trading does not mean blindly taking every setup. And of course, you know, setup is a sort of a retail term for, you know, the conditions that define. We’d say signal, they say setup.

And my response is, well, you know, systematic literally means exactly that. Right? It’s a word in the dictionary. You can’t redefine words in the dictionary because you disagree with them. And you know, to be fair to this person, they did engage with me and say, well, actually, yeah, you’re right. But, you know, the point I was making was ‘waffle’. But anyway, they did agree that, yes, all right, systematic trading does actually mean literally blindly doing exactly what the system tells you. That is the point.

But it does kind of lead into a wider subject as well, which is, if you are to accept (and I do accept this) that there is kind of some power in intuition and discretion. And in particular, there are probably people on this planet, and I’ve met some of them, who do have the ability to trade the markets in a way that can’t be captured by a system. What is the best way of kind of combining those two approaches? So, I would say that quite possibly the worst way of doing it is the way suggested by this person, which is that you do have a trading system but then you feel free to ignore it when it doesn’t suit you.

And the reason for that, I think, is that, well, for starters it makes it very hard to test whether your system’s actually working or not, you know, because you’ve got a great back test. But actually, if you’re just going to ignore what the system does, well, you know, it’s not really a system anymore. So, I think that approach is terrible. And you know, I do know that there are so called systematic places that, when they feel like it will ignore their signals and, you know, perhaps they’ll end up making money and more money as a result. But I don’t think you should be allowed to do that and call yourself a systematic trader. In the same way, I don’t think you should be allowed to have an ETF with a ticker CTA and then change your policy. But that’s another… We’re not going to talk about that anymore.

So, I think there are better ways of doing it. And it’s a subject I’ve thought about for a few years, and actually, happily, I’ve decided to write a book about it.

[00:36:35] Niels

There we are.

[00:36:37] Rob

So, the book is called The Art and Science of Trading. It’s out on December 1st. That’s the plug out of the way, and it sort of covers and it’s sort of aimed at resolving, or directed at, a number of different potential audiences. So, one audience is people who’ve picked up my previous books and thought, well, this is way too difficult. And although I do my best to strip as much maths and Greek symbols out of my books as possible, you know, there has to be a little bit of maths in there. So, I’ve deliberately written this book to be as accessible as possible while still telling the truth, if you like.

It’s also aimed at people who are trading with non-leveraged instruments. So, my previous books mostly cover futures, as you would expect, but also things like spread bets and CFDs. And here I put an asterisk and say, not legal in all jurisdictions and, you know, FX trading and things like that. This book is designed for people who can’t use leverage, which actually makes things a little bit more complicated because you can’t just assume your position. You can take all the positions you want. You have to be able to rank and order positions and have sort of path-dependent rules about swapping in and out positions. It’s actually more complicated to run without leverage, in many ways.

But the other audience is for people who want or think they can trade with intuition, or with discretion, or with some other non-systematizable thing, which could be looking at charts or could be looking at data, but not doing it and following specific rules, I feel they can add value. And my approach is to say, well, yeah, you absolutely can do that, but you have to limit the discretionary part of a trading strategy to one tiny part, which is the ‘what should I buy and when’ decision. I mean, that’s quite a big part. Right?

[00:38:32] Niels

I was just going to say.

[00:38:33] Rob

It’s quite a big deal. It’s not that easy to do that. It’s very hard indeed. But I think it’s fine if you believe you have skill in that area. And one of the things I cover in the book is about how you can test that skill.

But then every other part of a trading strategy should be based on a scientific, systematic, rule-based approach. So, that would be primarily things like deciding how fast you trade, deciding when to close positions, and deciding how big positions should be. So, essentially the subject of cost and risk management, because actually, if taking poor buy decisions was the only mistake that retail traders made, then you’d expect that roughly half of retail traders would lose money and about half would make money, because, worst case scenarios, they’re flipping a coin. But the real number is actually more like 95%/5%. So, 95% of retail traders lose money. About 5% make money. And if you look just at day traders, those numbers are even worse. And that’s because they’re making errors with position, and cost, and risk management.

And that’s where there is no room for discretion. There’s no art involved in that at all. That is science. And the other science I talk a bit about, although not in huge amounts of detail, is of course, as we discussed, the science of overfitting in robustness as well.

So yeah, I think we’re purely systematic on this podcast. There are people out there who are purely discretionary. But I do believe that there is room for both, for sure. And I do believe also that, you know, there are ways potentially of combining those that will work. But it’s absolutely not the approach of having a system and then just ignoring it when you don’t like the look of what it’s doing.

[00:40:28] Niels

I mean, at the risk that you need to put an appendix in your book, I have a question. And that is, from what I understand, you think people should not be allowed to have any say about when to get out of a position. That kind of, to me, tells me that probably people are not very good at that. So, I’m kind of curious, why do you think, if they can’t figure out when to get out, why would they be very good at figuring out when to get in?

[00:40:55] Rob

Actually, there is research on this topic, Niels. I can’t remember who did it, but academics have looked at retail traders, but they’ve actually also looked at institutional managers. And these are mainly discretionary, long only stock pickers. Right? So, you know, like mutual fund people.

And they found, for example, that there are a lot of mutual fund managers with genuine skill at finding good stocks, but actually terrible skill in choosing when to exit those positions. So, you know, it genuinely can be the case that you can do one and not the other.

Now I do, actually, in the book sort of say that, you know… So, not in an appendix, but there is a later chapter where I say, right, well, if you satisfy these conditions, then yes, you can use your judgment potentially to exit from positions. But you need to have essentially a statistically significant track record that proves that you have skill in doing that. And I explain to people how they can measure their performance and actually, you know, say whether that’s correct or not.

But yeah, I’d say that it’s completely believable that people are, you know, that there are people who are good at knowing when to buy stocks but not knowing when to sell them, absolutely.

[00:42:17] Niels

Funnily enough, there is something a little bit from our world, I think, that goes in that direction as well. And I don’t know if you agree with this or not, but what I found is because people often ask, so why did you choose vol breakout, or why did you choose moving averages, or whatever? And my experience, so my answer has always been, I think a lot of trend models today probably get in, more or less, at the same time in a trend. But I think where the big differentiator is between managers is actually how they manage the risk along the trade, but also where they get out. And I think that’s where the real difference shows up. And you could say maybe even the real skill shows up.

So, there is, you know, in an odd way, some alignment, I think in my observations about purely systematic trading and what you suggest there. So, maybe it is ‘easier’ to find the entry point and much harder to find the exit. And we also, I guess we also know, to some extent, that the exit is emotionally more difficult to take, especially if it’s a loser.

[00:43:31] Rob

Yeah. And that’s why you know, the simplest system I propose in the book is that you should use a stop loss, which is something I don’t use myself, of course. But, you know, if you need to hand over your exit strategy to a rule, well, let’s make it as simple and as clear as possible. You know, you can’t argue with a stop loss level.

Yeah, I mean, it’s true that, you know, there’s definitely more variation in how people handle trades after they’ve been placed. I mean, you know, probably still the biggest debate in the industry is over vol scaling after entry. Right? Do you adjust the size of your position to represent vol, or do you just kind of ride it and let it get as big as possible?

[00:44:21] Niels

Let’s, for the last 15 minutes or so, let’s jump into some of the other topics that you brought along. Well, actually, the first one was an article I saw. I know you like AI as a topic. I don’t know if you like AI.

[00:44:40] Rob

I don’t like AI, but I have to talk about it.

[00:44:41] Niels

Yeah, that’s what I meant as a topic it’s very important. And I saw that one of our previous guests, Marat Molyboga, from Efficient Capital, had put out a blog post, I think it was, about AI, where he argues that AI governance can borrow, maybe, some things from portfolio management. I would love if you have a few thoughts about that.

But actually, you can broaden it out wherever you want to go on AI, because, of course, AI is not only a topic that comes up a lot in our personal lives, it also comes up a lot in terms of, well, maybe AI can just replace all these investment experts.

[00:45:25] Rob

I mean, really, there should just be two AI people talking to each other on this podcast. Niels, I feel like the technology is at the point where we could both just be deep fakes. Right?

[00:45:36] Niels

True. Anyways, I’ll let you handle this.

[00:45:38] Rob

Yeah, well, so I think I’ve mostly been thinking about AI in the same way, actually, as in this article, as something that potentially could be used by, you know, a quant research or a quant hedge fund, a systematic hedge fund, a CTA, to sort of improve or develop the research and implementation process. So, that’s, you know, both idea generation, testing, and then implementation. Those are kind of, broadly speaking, those are the sort of three stages of quant research.

And I think there are clearly massive risks with AI. I think in the research phase, I think the biggest risk is potentially that it will just overfit and you need to put very strong guardrails around it to prevent that happening. You need to essentially wrap it in a lot of code that isn’t written by AI. And actually, that will be the subject of my next book, which I’ll no doubt be plugging in a couple of years.

But there are also risks, clearly, in the implementation. And there I’d be really concerned if people had live production code actually trading that had been written by AI with either no or insufficient human oversight. That would absolutely terrify me.

I have quite limited, because I actively avoid AI, you know, apart from what I would call silly stuff, you know, so generating meme pictures and whatnot. But I do work with, you know, I do consultancy for people who do use it. And one of them recently did a code review, got AI to do a code review of some of my research code. And it was a combination of good and bad. It did actually find, you know, a quite serious bug in my code and a couple of more minor ones, but it simultaneously introduced an even bigger bug that wasn’t in the original code. And when I say big, I’m talking, you know, that the code essentially would no longer have short positions, just long ones, because all short positions would be deleted by a check that the AI thought that should be in. And, you know, that just goes to show the potential implications of this stuff.

And I think, you know, if you’re using it and you are diligently checking everything it writes yourself, and if it reviews your code and suggests changes, you’re diligently checking them and checking them thoroughly, well, that’s absolutely fine. But I think it really worries me, this sort of, what I call agentic AI, where people essentially just let a bunch of AI talk to a bunch of AI and write the code that way. You know, I think that’s terrifying not just for live production code in hedge funds, but, you know, for the future of the world, if you like.

But anyway, so yeah, it’s quite an interesting article and it’s absolutely the case that, you know, we know a lot about risk management in our industry. And when we say risk management, of course, we’re not just talking about the risk of losing money, we’re talking about all the other risks associated with the financial organization. And as people who run things that essentially trade automatically, we know that the oversight, governance, and processes around code implementation are absolutely key. And just because AI turns up with its fancy bells and whistles absolutely does not change that approach and, if anything, makes it stronger.

Anyway, that’s that. The other thing about AI, that again (and this has been sort of prompted by reading something on X), is what effect? So right now there are an awful lot of people in the world suddenly deciding to trade the markets based on AI. Many of those are retail traders, many of them will be institutional traders. And I would say, of those institutional traders there will be quite a small percentage who are using AI correctly. And what will be the effect of a wave of people using, you know, consumer grade AI models on the markets? Will it, for example, will it increase turnover in markets? Will it increase risk? Will it increase volatility? ‘I don’t know’ is the answer. And it will be very interesting to see, you know, how things pan out.

You know, there were probably ways you could empirically test that. It’s still too early, I think, to get any strong evidence on that. There’s still not enough people doing it. But it does feel like, at least, inside the X Twitter echo chamber that I occupy, that there are a lot of people who want to do this stuff. And I think it will have an effect on markets.

I mean, it could just be like the meme stock craze. You remember that, Niels, remember the meme stock craze and GameStop and all that stuff? It could just be like that. Markets just could get quite silly for a bit and then maybe things will return to some kind of normal. But arguably meme stocks just haven’t gone away. I personally think Tesla is the biggest meme stock in the world. If you look at its price earnings ratio, it’s pretty silly. Will AI just intensify those effects? I don’t know.

Ideally, what I’d like to be able to do is be on the other side of those trades. Because, you know, you really want to be sitting at a table playing poker with people you know have got no idea what they’re doing. And I think that’s definitely going to be the, you know… I think we’re going to see a herd of uninformed AI driven retail investors and I think there’ll be opportunities there. I’m not sure how to capture them yet, but it is a bit worrying as well.

[00:51:43] Niels

But you do know that AI, actually, also has something really valuable to it, Rob, because Yoav, I think it was last week, said, actually every time he does, and I’m paraphrasing what he said, but he said something along the way that when you go and you ask different AIs about trend following and how it should fit into a portfolio, it comes out with a very favorable outlook and inclusion percentage, like the total portfolio approach type thing. Selective AI is maybe what we should be focusing on.

[00:52:15] Rob

Well, the reason is it’s been trained on reading the Internet. And you know, on the Internet, there are lots of people saying, including myself and you, that people should have more in there and more CTAs, more trend following in their portfolio. So, it’s read that and it’s unsurprisingly agreeing with us.

[00:52:30] Niels

I think it’s listening to TTU and then saying, oh, these are smart guys. We should have definitely much more into trend following.

[00:52:37] Rob

Sucked in the transcripts of all of the episodes. And yeah, they’re recycling back.

[00:52:42] Niels

Yeah, good stuff.

Well, speaking of trend following, you also wanted to revisit… Well, actually, it’s a new article, but it is VIX and Trend Following Revisited, an article that I think was a blog post that was posted on Alpha Architect, where they look at things like a decade after the original research was done, I’m just remembering roughly, what did you find interesting here?

[00:53:10] Rob

I mean, first of all, I really like the idea of revisiting research. And I think 10 years is a nice period because 10 years is long enough normally to get some additional evidence.

[00:53:21] Niels

Yeah, we did that last week, actually. I know you were on holiday, but that’s exactly what Yoav did with Lasse Pedersen’s paper from 2016. Yeah, about trend following, the 100 year podcast.

[00:53:37] Rob

Yeah, an absolutely classic paper, as everything Lasse writes is absolutely classic. It’s definitely worth following. So, yeah, it’s a good thing to do.

And you know, it’s one of those things that you kind of think sort of has a nice logic to it. And essentially the idea is that, you know, we should trade differently when the market is more or less volatile. And actually I have a little bit of that in my strategy. So, when the market’s more volatile, I trade less momentum. And actually, these guys say, well, the market gets more volatile, you should trade faster. And how do you measure market volatility? Well, of course, we have this lovely thing called the VIX, and that’s really nice, and we can use the level of the VIX to do that.

So, I’m not going to go into the article in detail. It is an Alpha article. It’s really very interesting. I’m going to jump straight to the…

[00:54:36] Niels

Well, before you do that…

[00:54:37] Rob

But before I do that…

[00:54:39] Niels

Before you do that, here’s something that I just want to flag and that is, that seems so logical that you say, oh, things get more volatile, let’s be short term so we can ride the ins and outs of all these moves. However, I actually think that one of the things that has stood (not just trend following but actually replication in particular) well, is being slow. And I mean being slow during periods like the tariff tantrum, whatever it was called, Liberation Day I think it was, and other things.

So, I haven’t seen any evidence that short-term managers, and I know we decided not to necessarily the same paper again, but there is a good paper that is out regarding why maybe short-term hasn’t worked so well. So, I haven’t, again, not going too deep, but I’m just going to throw out a little bit of question mark.

[00:55:47] Rob

I think we need to distinguish different kinds of volatility.

[00:55:50] Niels

Okay.

[00:55:51] Rob

And this is nothing to do with the paper, this is just my own thinking on the subject, just to say. So, the tariff tantrum was a classic V shape, right?

[00:55:58] Niels

Yes.

[00:56:00] Rob

And it was an incredibly quick V shape. Go back to 2020, we had a V shape, COVID V shape.

[00:56:06] Niels

Correct.

[00:56:08] Rob

The high volatility in, say, 2008 though was very different. That was a secular turning point.

[00:56:15] Niels

Correct.

[00:56:15] Rob

So, the market went down and carried on going. That’s about equity markets. Now obviously fixed income did the opposite, as did VIX. The market went down and carried on going down. It didn’t bounce back dramatically. It took several years, actually, for it to regain its previous levels.

Now it’s pretty obvious that if the market is doing a V shape, you want to be either very slow or very fast.

[00:56:41] Niels

Correct.

[00:56:42] Rob

You do not want to be just fast because you’re going to get whipsawed by that V. And obviously, you know, if you think about COVID and you think about the tariff tantrum, because those V’s were different widths, if you like, the COVID one was a bit wider. That meant that during the COVID one fast trend following actually did work quite well because it was able to ride down again, very slow trend following also did well. Kind of medium speed would have done the worst, and I think we saw that in figures.

The tariff tantrum though, because it was so damn quick, you’d have to have had an incredibly fast trend following to have ridden that down and up again. And actually, fast trend following probably did worst of all because it would have got whipsawed, medium and slow, pretty much held through and didn’t do too badly.

So, I think we need to be careful about making broad brush statements when effectively high volatility could mean one of three different things. It could either mean 2008, it could mean 2020, or it could mean the tariff tantrum of 2026.

[00:57:47] Niels

I completely agree with you.

[00:57:49] Rob

Sorry… 2025 was tariff tantrum, wasn’t it?

[00:57:51] Niels

Yeah, I mean I completely agree with you. The problem is we don’t know in advance what kind we’re getting.

[00:57:57] Rob

Exactly, exactly.

[00:58:00] Niels

That’s what concerns me that yeah, you can speed up your speed because now it gets more volatile. But what if it is a super quick V shaped or whatever?

[00:58:11] Rob

I’m discussing the article. Personally, I have a prior towards simplicity, which means I have a prior towards not doing this kind of thing. Right? All I’m saying is that, you know, there is a story here and it kind of makes sense. That doesn’t mean I believe you should implement this idea in your portfolio tomorrow. I don’t think I’ll be implementing it in my portfolio tomorrow, for example. But, with all that, what I’m saying is it’s not just some random data mining finding.

And I think why it’s interesting is because, and actually this is true of my sort of parallel finding that when volatility is high, trend following doesn’t do as well, is actually counterintuitive because people expect that high volatility is good for trend following. They have that image in their head of the straddle payoff where if the market doesn’t move, you lose money and if it moves a lot in either direction, you make money. But as we’ve just said, that payoff will depend very much on, you know, the speed of the trend and the speed of any reaction and the speed that you’re trading at. So you know, it’s a much more complicated story than I, as I introduced it.

[00:59:23] Niels

Absolutely.

[00:59:24] Rob

Anyway, just to say anyway that they did find that it still this thing still seems to work. Okay.

[00:59:33] Niels

All right, so that’s fine.

[00:59:34] Rob

That’s. That’s the conclusion.

[00:59:36] Niels

That’s the conclusion.

[00:59:38] Rob

That’s the conclusion. But, you know, it’s an interesting lesson because I find these articles interesting from a kind of academic point of view. But that’s, you know, often the initial analysis is hiding a lot. You need to understand what’s going on underneath the surface.

And of course, you know, you should be very, very careful. Every time you read something on Alpha Architect, you shouldn’t, you know, go, oh, great, that works into my system. In it goes. I think that’s…

[01:00:08] Niels

You need to ask your AI first.

[01:00:10] Rob

Exactly that, ask your AI what it thinks.

[01:00:12] Niels

Yeah, okay. But we are running out of time, my friend, so I think we just pick one more topic that you wanted to maybe talk about to round things off.

[01:00:23] Rob

Yeah, well, let’s, out of loyalty to my previous employers, let’s go very quickly to the latest article from the Man Group Insights team.

[01:00:38] Niels

Yes, it actually came out this morning, I think.

[01:00:41] Rob

It came out this morning. So, this is live. This is quite nice because one of the annoying things about this podcast, I have to say, Niels, especially when you insist on putting people like Yoav on the week before me, is that, you know, any good articles that come out get stolen by the person, leaving nothing for the person the following week. So, I’m stealing this one. Sorry, Yoav, sorry, whoever’s coming next week, or whoever. This one’s mine, I’m afraid.

Anyway, so, this isn’t really specifically about managed futures or CTAs or trend following. It’s actually more on quants generally. So, you know, in quant worlds, we have these things we call factors that we think sort of explain performance and also are used to try and find mainly stocks that have some, you know, hopefully some idiosyncratic alpha.

So, inside these managers, one thing they’ll typically be doing is a regression, statistical regression. And when they do that, they have this number called the R squared. And the R squared essentially is a number between 0 and 1. That sort of says, how well do your factors explain your returns?

And of course, this is also something you can do externally. So, what you can do is say, well, I think there’s a bunch of factors that explain manager performance, and I’ll do this regression, and if I get a high R squared, that means that most manager performance is being driven by these factors. (Yeah, have I explained that well, enough? Yeah?) Most people listening to this will be familiar with that concept. So hopefully it’s not new.

Anyway, so they do a nice exercise where they do a little rolling regression, and they look at the R squared, and they see that it does seem to be the case that over the last few years (and they’re going back quite a way. They’re going back to the early 2000s). So, you know, nearly 25 years, not quite. There is some sort of reasonable variation in this sort of R squared.

And the other thing they look at is. Is this thing called the quant winter. I feel like… I don’t know… Do we have a sound effects guy on this podcast who can… so, when I say ‘the quant winter’, there’s some kind of filter on it that makes it sound spooky. But anyways, I don’t know. (Loud horn goes off.)

[01:03:15] Niels

Something like that?

[01:03:16] Rob

That’s inappropriate. I was thinking of something that made my voice sound really spooky. But I’ll just have to settle for going very close to the microphone and saying ‘the quant winter’.

Anyways, so the quant winter was in 2018 to 2020. And this was a period of time when these factors didn’t do very well. So, you know, the classic factors, you know, your value and your momentum, and they did pretty badly.

So, yeah, it’s quite a long article. You know, it goes into a lot of detail. It’s very well written, it’s very readable. There are lots of lovely charts. And the main lesson here, as we’ve already discussed with the CTAs, you know, quant equity, quant managers are not all the same and they are doing things differently. And you can’t just assume that they’ve just got these basic factors underneath. It does. They are doing things differently. And, you know, diversification is important, but of course, diversification has a habit of going away during events like the quant winter, when everyone does…

[01:04:36] Niels

(Darth Vader Like Voice) You mean the quant winter?

[01:04:36] Rob

That’s It! That’s exactly it! Thank you, Niels.

Anyway, so, yeah, my brain’s clearly completely gone now because I’m down to just doing silly voices when discussing a very good and very serious paper. But, yeah, it’s very interesting. Probably of most interest to anyone in an asset allocation, who’s thinking about where to put their money after they’ve obviously put 90% of it in trend following, and they’ve got 10% left to put into some equity long/short managers.

So, I would say that with any manager allocation decision, if you can afford it, you should definitely diversify because it’s very hard to pick who the best manager will be. And it does seem in quant equity land there’s an increasing amount of idiosyncratic performance and diversification available there in the last few years.

[01:05:29] Niels

Yeah, absolutely. All right, well, people should, of course, as always go and check out the papers we talk about themselves so they can make up their own mind about it. This was great, Rob. Thank you so much for doing that and preparing for all of this. It’s good to have you back and we’ll continue in a few weeks, I’m sure with some other interesting topics.

Now to all of you who are listening, as always, I will encourage you to head over and leave a rating and review on your favorite podcast platform just as a courtesy and recognition for all the hard work that the co-hosts do every single week to be prepared for these conversations. And if you have a question, you can, as usual, send them to info@toptradersunplugged.com if you do, then next week it’s actually Rich coming back. So yeah, send a good question about trend following. He’s a real professor in explaining some of these things. So, that will no doubt lead to a good conversation.

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

[01:06:51] Ending

Thanks for listening to Top Traders Unplugged.

If you feel you learned something of value from today’s episode, the best way to stay updated is to go on over to your favorite podcast platform and follow the show so that you’ll be sure to get all the new episodes as they’re released. We have some amazing guests lined up for you and to ensure our show continues to grow, please leave us an honest rating and review.

It only takes a minute and it’s the best way to show us you love the podcast. We’ll see you next time on Top Traders Unplugged.

This podcast expresses the views of its hosts and the guests appearing on the podcast as of the date of its recording, and such views are subject to change without notice. Top Traders Unplugged do not have any duty or obligation to update the information contained herein.

Furthermore, Top Traders Unplugged make no representation to its accuracy and it shall not be assumed that past investment performance is an indication of future results. Moreover, wherever there is a potential for profit, there is also the possibility of loss.

This content is made available for educational purposes only and should not be used for any other purpose.

The information contained in this podcast does not constitute and should not be construed as investment advice or an offer to sell or a solicitation to buy any securities or related financial instruments in any jurisdiction.

Certain information contained herein concerning economic trends and performance is based on or derived from information provided by independent third-party sources. Top Traders Unplugged may believe that the sources from which such information are obtained are reliable.

However, Top Traders Unplugged cannot guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based. This podcast, including the information contained herein, may not be reproduced, copied, republished or posted in whole or in part in any form, without the prior written consent of Top Traders Unplugged.