Everyone’s Different. Until They’re Not ft. Rob Carver

[00:00:01] Intro

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

[00:00:23] Niels

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

Rob, wonderful to be back with you this week. Hope you’re well. How are things in the UK?

[00:00:38] Rob

It’s hot, which is a mixed blessing, I think. But yesterday I came up with the perfect strategy. I spent the whole day in a beautifully air-conditioned conference room at a conference. And then after the conference we were invited up to the roof terrace which had a fantastic view of the London skyline.

It was quite a pleasant evening and a very nice place to spend the evening drinking free wine and looking out on a wonderful view. And the temperature had got down to a point where it was very pleasant indeed.

[00:01:05] Niels

Sounds like a good day in the office for sure.

We’ve got a great episode lined up with lots of interesting topics, three questions, some articles. But of course, as we normally do, before we head into all of that, I’m always curious to know what’s come across your radar lately.

[00:01:22] Rob

Well, Niels, unfortunately I’m going to break one of the cardinal rules of TTU, which is, of course, the first rule, which is don’t talk about TTU. I’m afraid I have been unfaithful and I have been listening to another podcast.

Yeah, so it’s the Money Stuff podcast, by Matt Levine and Katie Greifeld of Bloomberg, which is, you know, obviously I would never normally even think about listening to the podcast, but this episode from last week had on one of my favorite people, Cliff Asness from AQR. And it’s a great podcast.

He talks a lot about backtesting, about risk factors, about momentum. So, all the things that I know other people who are listening would be interested in. So, perhaps just for once break the rule, turn away from TTU. Just listen to this one episode and then you can return back to the fold, and you will be forgiven.

[00:02:11] Niels

You know, I’ll be completely transparent with you, as I always am. And that is, I listened to it as well. And interestingly enough, there are very few other podcasts I do listen to.

And actually, in my weekly newsletter (for those who subscribe to it), there is a little thing called What I Listen To, and it gives access to a Spotify playlist where I add music and podcast episodes that I find interesting. So, I will add this one to that playlist.

[00:02:43] Rob

You’ve got to be careful. You might listen to some music that you wouldn’t want anyone to know about. You know, you’ve got to be careful with that kind of thing.

[00:02:48] Niels

Okay, I’ll be transparent again, there is less music than there is podcast episodes because I’m not sure people do share my taste in music. You never know. You never know.

Anyways, my radar was lit up this morning when I saw a story, I think on Bloomberg, talking about the Federal Reserve gifting Wall Street a big gift. So, I thought, what could that be? Interest rate cuts? No, it was actually something about capital rules and how they plan to roll back an important capital rule that gives banks, that they have had to comply with, to limit their ability to hold more treasuries from the US Treasury market, of course.

And of course, you see other comments about it, and people being straight away concerned about, well, aren’t we setting ourselves up, by lifting these restrictions, for another bank crisis, as we saw, whatever, 20 years ago or so?

And of course it would be kind of ironic, Rob, if 10 years from now you and I are going to say, well, remember back In June of 2025, just before the big financial crisis, that they lifted this rule?

[00:04:02] Rob

Yeah, it’s true. I mean, one of my theories is that the Basel II Banking Capital Adequacy rules were partly responsible for the last financial crisis. So, these things are important.

But actually, Niels, I think it’s worse than you think, because the conference I was at yesterday, the final speaker was a gentleman called David Sacks, who is the AI and Crypto Czar for the current US government. And he said that they’re looking at the treatment of Bitcoin on balance sheets, on bank balance sheets.

So, I’m not actually an expert on that at the moment, but it could be that they’re going to treat stablecoins and perhaps non-stable coins, unstable coins like Bitcoin and Ethereum and so on, as more like kind of proper assets, and allow them to sit on bank balance sheets and be part of capital adequacy rules. And I’m going to be honest, Niels, that concerns me a little bit.

[00:04:58] Niels

Concerns me a lot. And actually, I don’t know much about stablecoins. I’ve never really figured out exactly what their role is. But I will mention that I know one of our other co-hosts are working on trying to get an episode or two around stablecoins I think actually, it’s something we need to keep an eye on. For sure, it’s coming. It’s, it’s here already. So anyway, we should learn more.

But the other podcast that I can also be very transparent about, Rob, is Odd Lots. And they also have a newsletter that I read from time to time. They had an interesting fun little indicator they were talking about this morning, and it was actually something as “benign” – not unemployment, not CPI, whatever. It was containers and boxes – how much is being shipped in and out of the US. And apparently, they say that when you look at these indicators for the past few years it’s been very, very good at predicting the US economy, and so on, and so forth.

So, they refer to the latest bank of America’s box survey and apparently the growth of these numbers, in terms of prices, in terms of volume, are basically falling off a cliff (not to be too dramatic about it). For example, they had something around, back in March of this year 11% of the survey respondents said container bought prices were more likely to fall than rise. Okay, now we’re three months later now 61% believe they’re going to fall more likely than rise. That’s one thing, that’s just the prices.

But also, in terms of volume, there is a guy who I think writes for the Container Volume Observer, I think is the source, John McCrown, and he points to a 6.9% decrease in inbound US volume last month, which is a big reversal from the 11% and almost 10% increase we saw in March and April. I mean, these are not small numbers for sure. Even if it’s something we don’t normally talk about on this podcast.

[00:07:12] Rob

I mean, I can’t think what would have happened that would have caused people to suddenly want to bring a lot of stuff into the US very quickly before a specific date and then import less after that date. Nothing springs to mind.

[00:07:23] Niels

No, no.

[00:07:24] Rob

Yeah, I think I’m going to get myself a subscription to, what was it, Container Observer Monthly. That sounds like a riveting read.

[00:07:32] Niels

Sure. Well, speaking of things that have been challenging, Rob, I think we need to talk about trend following. It’s been a bit of a challenge, I have to say. My trend barometer, yesterday, finished at 43. Now it’s weak. It’s not a disaster, but it’s weak. But what I have experienced the last month or so, in June, is a fair amount of daily movement in performance actually. But it’s almost like been one step forward and one step back. Not a great deal of net-net movement.

But, on the other hand, I think what may decide a little bit about how people fall +/- so far this month is also what took place in oil. I mean the big surge initially after the strike on Iran and then the collapse only a few days ago.

I mean, I think it kind of demonstrates or visualizes the challenges we’ve been up against in the last period of time. Also, as you say, with the events in April and the announcements and what took place around the day, so to speak.

But whether it’s from your own performance or whether it’s just what your observations are in general, Rob, I, of course, am curious to know how you think about the last three or four months. And also, putting into context a little bit your experience throughout all of these years you’ve been involved in this space, does it feel different? Is there anything to be concerned about? What are your thoughts generally speaking?

[00:09:09] Rob

I mean we had quite a long discussion about this the last time I was on, so I won’t kind of go over that ground. But just to briefly recap, you know, the current drawdown is the largest I’ve seen in live trading and it’s kind of on a par with the biggest drawdowns in my backtest which go back to 1970.

So, you know, it’s not like sort of shockingly awful and terrible and you know, unbelievably bad, but it’s kind of a 1 in 10 year, maybe 1 in 20 year sort of drawdown, let’s say. Actually, just the last month’s been actually not bad for me at all. So, I’m actually up a couple of percent which, you know, it’s better than a kick in the teeth, as we say in Britain. And although, actually, there was sort of a good run from kind of late May to sort of mid-June. And then I think you’re right, I think the ructions over Iran have sort of caused, for me, a little bit of P&L to come off on there.

So, yeah, you’re right. I mean, in terms of like, is this a good environment for trend following, or good environment to be seeing the start of trends to put money on? Well, certainly my own system disagrees with that and is saying, well, actually, it’s still pretty low risk to be honest.

So, I’ve got a longish position in equities. So, I’m a little bit risk-on there. I’m actually a little bit short bonds. So, you could call that a kind of risk-on trade I suppose, but a very small one. And actually in energies, which obviously, as you say, has been in the news recently, I’m actually net flat. So, you know I’m short gas but I’m long crude. So, I’m kind of, I could have sort of hedged positioning energies effectively.

But there’s nothing there shouting out to me, you know this is the start. There are no positions being held on, no big risk coming on. I mean, it is nice to see performance stabilize after what has been, you know… I mean, obviously, this year, this calendar year has been not great and I’m still down for the calendar year. I’m still down sort of 5.50% but, you know, that’s a couple of percent up from where I was.

But yeah, after what happened in April and on the back of last year not being great either, it’s nice to (cross fingers, touch wood) maybe dream of a better future.

[00:11:43] Niels

Well, we can always dream, that’s for sure.

[00:11:45] Rob

Yeah, exactly.

[00:11:45] Niels

But let me ask you one thing. You mentioned this kind of long-term perspective that with your backtest, your live data, etc., and I don’t know if you know this offhand, but you might. In terms of standard deviations relative to your annualized volatility, what level are you at or what level is ‘the max’ you’ve seen in your simulated data, live data, just to give people a feel for what we’re talking about?

[00:12:13] Rob

So, one thing you can do, actually, is convert your P&L into an annualized Sharpe ratio. And that’s quite, for me, an intuitive way. So, my long run Sharpe is probably going to be somewhere between 0.5 and 1, something like that. For 2025, so far, I’m down 1.3 Sharpe – minus 1.3. So, basically if my performance carried on at the same average level it’s been for the last few months and the volatility is the same as well, then I’d end up losing 1.3 kind of risk units over the whole year. And that’s my actual total drawdown which is from July last year. So, I hit highwater mark in, sorry, in May last year. So, about a year ago.

My actual total drawdown from there is probably somewhere, again, equating to down about one Sharpe ratio unit.

[00:13:06] Niels

But that’s not bad.

[00:13:07] Rob

It’s not bad.

Yeah, I mean the nice thing about trend following, obviously, is if you look at the right time frequency, it’s a positive skew strategy. So, my kind of rule of thumb is, in a good year I’ll make two Sharpe ratio units, in a bad year I’ll lose one.

So, if you just think about converting everything to Sharpe ratio units, which is a multiple of standard deviation effectively, then you’re kind of going plus two minus one.

[00:13:34] Niels

You’re saying you’re down about one standard deviation of your analyzed volatility. And I think, just to put things in perspective, I mean when the S&P 500 is down 50% or 60% in what we’ve seen in the last 25 years, that’s like four standard deviation of the annualized volatility. So, to put things into perspective.

[00:13:52] Rob

Right.

[00:13:52] Niels

So, it feels like a tough time for the industry. It is a tough time. As I was talking on the podcast last week when we went through the paper written by Dan, over at R. G. Niederhoffer, even strategies that don’t make money for a while might still be a good strategy to have in your portfolio. So, as long as we can deliver sort of the non-correlated returns in the long run, I think we should be fine. That doesn’t dismiss the fact that it feels painful at the moment.

Okay, anyways, not too much pain to talk about, at least for trend followers in June so far. The BTOP 50 is up 66 basis points in June, down 4.25 for the year. Soc Gen CTA index down about a 0.25%, down 8.75% for the year. The Trend index is flat in June so far, down 11.25% for the year. But maybe a little bit, to my surprise, Short-Term Traders index is down 3% in June and now it’s down 5.5% for the year. And given that it’s trading at like a third of the volatility or something like that, or almost a third of the volatility, that is a meaningful change because they were doing so well in April relative to the trend followers.

So maybe it’s exactly what we talked about, these news events over the last month or so, especially what’s happening in the Middle east and what we see coming out of the US Administration, that it has been exceptionally challenging for certain timeframes to operate for sure.

MSCI World up 2.3% in June, up 7.6% for the year, the S&P US Aggregate Bond Index up about 1% for the month and up 3.50% for the year. And the S&P 500 Total Return up another 3%, up 4.25%. Who would have imagined that around Liberation Day in April? Probably not many.

Anyways, let’s jump in and tackle some questions that came in. The first one is from William. He writes, “Greetings, in equity trading there’s a concept of decomposing returns as the sum of factors, market, sector, value, etc., plus the residual idiosyncratic return. I don’t often hear of the same approach being used in commodities and was curious if you could educate the listener on this topic.

For instance, if metals are all correlated, it would seem you could derive two separate trend signals for a given metal, one corresponding to its exposure to the metals factor, and another one of the idiosyncratic trend of the individual market metals return after subtracting the component of the return explained by the broader factor. Does Rob think treating these factor and idiosyncratic time series as separate assets would lead to meaningful different result in position sizing? And does his handcrafting and diversification scaling approach implicitly capture the same thing?” (It’s probably ‘his handcrafting’ as me misreading there.)

Anyways, I know you got the question, so hopefully you have some thoughts.

[00:17:11] Rob

I have many thoughts. So, I’ve actually done a bit of research on this and actually, I haven’t explicitly used a factor approach. Although, actually, one of the blog posts I want to write one of these days is about factor approach. It may even be next, you know, next month. I’ll see what mood I’m in when I sit down to write a blog post. It depends on the mood that I’m in as to what I fancy writing about.

But what I have done is to look at what happens if you do trend following in a slightly different way, which is to create an asset class index which you can think of as analogous to MSCI World for equities. The difference is that MSCI World, obviously, is market cap weighted, which means it’s basically America plus noise, right? The way I construct the index is I kind of normalize everything by volatility and then equally weight.

So, although I have more US indices in my basket, that still means the US weights probably still only about 10% rather than 70%, or whatever, in MSCI world in equities. And then the nice thing about that is you can use a similar approach in say metals for example, or any other asset class where a market cap approach wouldn’t necessarily make sense because what is the market cap of copper, for example?

So having got that index, what I then do is say, well, I’m going to trend follow that index. I’m going to, you know, if it goes up, I’m going to buy. And basically, what I do then is I take the signal that’s coming from that trend following, so, you know – long if metals are going up, short if metals are going down, for example. And then I use that signal in my normal kind of signal construction for every individual instrument.

So, what that means in practice is if I was only trading this signal, then it would basically go long all the metals, or short all the metals, on a given day, depending on whether the index is long or short. So, that’s kind of analogous to the idea of creating a metals factor.

Now the difference is, and I’m going to get a little bit nerdy for a second (I hope people don’t mind, feel free to fast forward). The way I construct that index is different from the way you do it if you’re creating a factor. If you’re doing a factor, you would actually do a… well, actually you could do it the way I do it, to be honest, because that’s sort of the way that market beta is estimated in equities.

You start with a market index and I’m saying, well, I think a better market index is not, you know, market cap weighted, but equally weighted, vol weighted, risk weighted. But most commonly, in factor analysis, actually what people tend to do is something like a PCA – principal components analysis. And that will basically, it sort of gives you, it tells you like what’s the main thing driving metals or driving equities. And then where does that come from? What are the weights of that thing on the individual things inside it?

So, for example, I recently did this exercise in cryptocurrencies and 70% of the weight was coming from bitcoin. So basically, crypto you can think of as bitcoin plus noise, effectively.

And if you were to do that exercise in equities, you would actually find that the S&P would probably be quite a big weight. So, you actually end up with a similar sort of index using a market cap approach and a PCA approach in equities at least. In other asset classes, it’s not so clear.

And in metals, for example, because metals are a very interesting asset class actually, because you’ve kind of got different things in it. You’ve got precious metals like gold, silver, platinum, that act as kind of risk-off assets like Swiss exchange rates and US Treasuries. They’re things people buy when things are going wrong. But then you’ve also got things like copper and other industrial metals.

Gold, those precious metals have industrial uses as well. But the majority of their price movement is driven by this risk-off factor. And their prices are more correlated to the general economic growth. So, you know, when economic growth is higher than people want to buy stuff. Stuff is made out of metal and therefore the price of metals goes up, broadly speaking.

So, I have not done it, but I would imagine that if you were to do a PCA for metals, you might actually find that there are kind of two factors that are quite big in there, sort of risk-off precious metals factor and an industrials factor. So, that’d be quite a fun exercise to do.

Anyway, it doesn’t really matter. The point is you end up with an index and from that you can trend follow that index and then you can sort of use that to trade everything together. The next thing you do is say, well, obviously I’m left with what economists call the idiosyncratic risk. But you can think of it as the sort of relative value momentum. Because if we’re going to be long all the metals at the same time or short all the metals all at the same time, we’re missing out on the kind of relative momentum of the assets.

So, maybe the metals complex, as a whole, is doing really well, but gold’s doing better than silver. So, we’d want to be longer gold than silver. So, we’d have a relative value strategy on top of that where we’d go long gold and short silver and then add everything up at the end.

And the way I do that is I basically (it’s in my last book if you want to read it) look at the index, which is a vol normalized index. I look at the vol normalized price of say, gold and I look at how those things are diverging, and that gives me a measure of outperformance that’s risk adjusted. So, if over the last six months gold has been doing better than the index as a whole, I’ll go long gold. If silver’s been underperforming I’ll go short gold.

And by its very nature that’s a kind of market neutral portfolio for metals if you like. And that’s analogous to trading the residual. Although actually, most people, when they’re trading residuals in factor models, tend to trade the residual as a mean reverting process.

But if you’re looking at trend following and looking at the right timescale, you’ll find that that residual does actually trend – show trending behavior. In the very long run it mean reverts, and in the very short term it mean reverts. But in one month holding it’s a tendency trend.

Now the interesting thing is (and this again is in my last book), if you look at the performance of that, that’s called the index plus residual strategy, and you compare it to the performance of just trend following, like an idiot, and not caring about indices, they’re very similar. They come out almost exactly the same.

And actually, what that allows you to do is to decompose your performance into how much performance am I getting from trading the index and how much am I getting from trading the residual from trend following? And you find that the split’s about 70/30. So about 70% of your return is, you know, when you’re just trend following like an idiot, let’s call it, comes from effectively the fact that you’re just trend following the metals index or the equity index, the bond index. And then another 30% or so comes from the fact that you’re trend following, say, gold versus silver, and so on, and so forth.

So that’s quite an interesting finding and it has implications. For example, it means that if you’re an investor with limited capital, you can kind of shrug off the fact that you’re probably not going to be able to trade all the different metals if you had to pick just one. Well, you can pick the one that has the biggest contribution in the PCA regression, which would probably be gold in metals, I would imagine, or it definitely would be the S&P in equities, and US probably five or ten years in bonds. If you just trade that, you know, you’re probably going to be getting about 70% of the performance if you traded everything on top. So, it’s kind of a nice pat-on-the-back for that.

So, yeah, that’s kind of the stuff I’ve done. And I do want to do a bit more work on factors because there’s other stuff you can do with factors like all this has been about, again, being nerdy for a second, we’d call the first principal component the first source of risk. But there are other components below that less intuitive and maybe more interesting.

And once you start factoring those out as well, you will get to the point, potentially, where you’ve got a mean reverting residual, which means you can create a mean reverting strategy that may be fun. And the other thing you can do, which I kind of think would be a novel thing to do, I’ve never seen anyone else do this before, so maybe I should write this blog post quickly before someone else does. But you can run a PC or entire futures universe – so, actually, equities bonds everything together.

Actually, I tell a lie. I’ve seen it done by people in risk management teams working in CTAs. They’ve done it as a risk management exercise. But I think what would be novel is to do it as a trading strategy exercise. So, there we go.

[00:26:05] Niels

By the way, in terms of printable components, didn’t Quantica do a paper on this a little while back? I don’t want to suggest that they did, but I just seem to remember they did something where they also concluded that, actually, performance tends to be better if it’s just a few components, or I can’t remember, maybe I should relook it up before.

[00:26:29] Rob

I mean this is also related to one of our favorite topics on the podcast, which is replication. Because if you can kind of say, well, a lot of the performance of CTAs is coming from a few factors. But that’s slightly different because what you’d actually do is start with the returns of your trend following system across all the different markets and then you’d work out the PCA of that – run the PCA of the underlying markets themselves. So, that’s not quite the same. So, that might have been what Quantica were writing about, possibly.

So, one of the arguments is, well, if there’s only a few factors driving everything, we just need to trade those factors. We don’t need the other 100 or so instruments. And yeah, I’m not having that debate again because we’ve had it many times before and people know my feelings.

[00:27:15] Niels

Yeah, we need to tread really careful here, Rob, because we definitely don’t want to give the replicators too many solid arguments for that case. Right? I’m sure Andrew is smiling right now.

Anyways, all right, well, let’s dive into another slightly nerdy topic, but something that you can definitely handle and certainly Raphael wrote in about this, so he believes that you can handle it as well. It’s about estimating the true Sharpe and full Kelly of trend following.

So, he writes, “The topic is inspired by Rob’s brief comments on X about a manager running their program at little under full Kelly and how a program running at full Kelly with Gaussian, with normal distribution of returns, could run their program volatility to be equal to its Sharpe. For example, a manager with a sharp of 0.5 should run at 50 vol for full Kelly, assuming Gaussian returns and even higher for more positively skewed. Could you elaborate on this if I misunderstood anything from your comments?”

[00:28:17] Rob

So, you didn’t misunderstand. Am I allowed to mention the name of the manager?

[00:28:23] Niels

Can you just describe it as a trend follower?

[00:28:26] Rob

It’s a trend follower that’s often kind of highlighted as they have very big years. And this year I think they’re down 50% or 60%. I think we all know who they are. So, the point is it’s abstracted from trend following. It’s a hedge fund that runs at extremely incredibly high vol for a hedge fund. Most hedge funds do not run at this vol level.

Now, the Kelly criteria states that to maximize your geometric return, which also will maximize your log of final wealth, blah, blah, blah. If you’re not a nerdy person, think of it as the CAGR, the Compound Annual Growth Rate, which is a measure I think a lot more traders are familiar with, that’s basically what you’re maximizing. You maximize that by setting your annualized risk in standard deviation terms. And this is a purely theoretical result. So, you can’t argue with it. It’s just maths.

Obviously, it’s relying on some assumptions, relying on Gaussian returns. It’s assuming you know your Sharpe ratio, which we’ll get to in a second, and it’s assuming you’ve got infinite access to leverage. But basically, you should run your system at a standard deviation equal to a Sharpe ratio.

So, if you think that your Sharpe ratio is 0.5 which, as I said, I think my Sharpe ratio is somewhere between 0.5 and 1, probably. Then you should run, to get maximum CAGR you’d run your system an annualized standard deviation of 50%. Which, by the way, is more than twice what I actually run my own system at.

So, my system is much more conservative than that. And most CTAs, hedge funds, are running an even lower vol than I am as well. And you can see that if you look at things like the CTA indices, like you said, they’re normally not that volatile. So that’s that.

Now that assumes Gaussian returns. If you’ve got negatively skewed returns, you need to run at a lower vol target. And if you’ve got positive positively skewed return, you should run at a higher vol target.

On my blog, if you can be bothered, you’ll actually find that I discuss how much you can sort of change that risk target if you have positive skewed or negative skewed returns. And I think one misconception is that having positive skewed returns, having these lovely outliers that certain people like to talk about, is obviously a nice thing, but it doesn’t actually increase your standard deviation targets as much you think it does and therefore you can’t necessarily kind of turn that into higher returns.

[00:30:48] Niels

Interesting. Very good. All right, well, there’s a little follow up question from Raphael. It’s about position sizing and maximum CAGR. He writes, “Many discussions have already been held on the topic of position sizing and maximizing trend following CAGR, including Rob’s own post on the topic.

But I would like to hear your thoughts on this topic raised in the following articles through the perspective of maximizing CAGR rather than risk adjusted return metrics such as Sharpe and Sortino.”

And the two papers that he refers to is a blog post on LinkedIn by a gentleman called Carlo Zarattini. And then there is a paper from our friends over at Quantica called Let it Ride or Lock it In. Those are the two papers he refers to in the question that we got. So, I will let you lead from here again, Rob.

[00:31:41] Rob

Yeah, well I should say up front that I’ve only really glanced at Carlo’s paper and it’s quite detailed so I’m not going to probably do it justice here. But basically, he looks at a few different ways of position sizing. And I guess you could describe them as adjusting or not adjusting in different ways. So, you can be adjusting depending if volatility changes during the life of trade.

So, the way I trade, for example, is if volatility increases, then I reduce my position size throughout the life of a trade, not everybody does that. And the other thing to look at is whether you change the size of your position according to your conviction in the trade. So, what I would call the forecast. So, if a trend gets stronger, for example, would you increase your position size? Again, I do that, but not necessarily everybody does that.

[00:32:35] Niels

So, that trend strength is what you’re looking at.

[00:32:37] Rob

Yeah, I mean that’s sometimes called pyramiding, and actually in this article it’s called pyramiding. But pyramiding really only applies to people trading in a sort of discrete way. Whereas I trade continuously.

But, anyway, he finds that basically, over a long backtest, you want to kind of do both of those things, I guess. And I think that’s consistent with work I’ve done myself on the same thing. But there’s not massive differences between them. Actually pyramiding has a much bigger effect on the thing.

But I think overall it sort of agrees with the fact that, as a rule of thumb, if you can get a higher Sharpe, you can get a higher CAGR, assuming you’ve got leverage, unless you give up too much positive skew in the process of doing it. That’s kind of what my lesson message is.

And basically, I found that the increased positive skew from doing things like not changing volatility and just essentially hanging onto bets with sort of fixed stop losses (one entry, one stop), that does give you extra positive skew but you lose too much in Sharpe ratio. So, net-net it’s not a winning thing to do and you end up with a lower CAGR. So, I think that confirms the work that I’ve done.

The Quantica paper is exactly the same thing. They look at what they call dynamic versus sort of static position sizing and you know, I think…

[00:34:03] Niels

They conclude the same, actually.

[00:34:05] Rob

They conclude the same, exactly, yeah. And I think part of the problem is that there’s sort of what behavioral finance people call an anecdotal bias. So, people tend to look at one market like cocoa (just to pick an example that Carlo actually talks about as well), where if you had held on to sort of the bitter end and just allowed your position to get bigger and bigger, and net-net that would have given you lots of money.

But the problem is, at a portfolio level, all that’s going to do is increase the dispersion of your trade. So, your trade especially, will be massive and you’re going to have huge concentration in individual positions potentially. And when that position goes against you, you will see a significant loss, as the trend manager (we’re not naming) has done this year, and as other people will have done because the markets move savagely against them and the position they have is just too big a part of their portfolio.

So, basically, the portfolio has become very concentrated and that means that, although on average their standard deviation versus Sharpe ratio level of targeting is correct, if you allow your portfolio to become too concentrated in one place, then the standard deviation is going to be a lot higher, which means you’re actually going to be running at significantly above the risk target you should have given – what you think your Sharpe ratio is going to be. So, it’s in those kind of peak events where you’re basically over risked and you’re suboptimal.

[00:35:33] Niels

Yes, and also, just again to talk about that specific manager who has produced some amazing returns in the last few years, but also, as you say, has suffered this year. The volatility of that product or that fund reached something like 90% at times because of the static position sizing.

And again, it kind of becomes a slightly different question for investors as well. And that is just simply does that fit their own risk appetite? But of course, people can size it appropriately as we would always suggest.

[00:36:07] Rob

Yeah, I mean just sort of from a theoretical point of view, if you think your risk is now 90%, that implies you think your current expected Sharpe is 0.9 or higher, which is, you know, probably getting close to twice what you’d think the Sharpe… Because, as I said, I think the Sharpe’s pretty close to 0.5 when I estimated it. So, that means that you currently have got so much conviction in your positions that you think that your Sharpe’s twice what it should be to have double that risk target. That’s pretty aggressive.

You know, I don’t think anyone can forecast the sort of near-term Sharpe accurately enough to make that kind of size bet.

[00:36:43] Niels

The good thing is though that there are so many different types of investors out there that for some people it’s the right thing and they know how to manage that exposure. And for others it’s the wrong thing. And so, we’re all left with a choice. And, of course, diversifying across different approaches is also as always…

[00:37:03] Rob

That’s very diplomatically put, Niels.

[00:37:03] Niels

As always.

Anyways, let’s move on to a question. Let’s see how diplomatic you can be with your answer from Dom. It says, “Hi TTU team, thank you for the consistently high-quality content. (That implies you need to be really nice here, Rob.) I have a question for Rob, and I’d love to hear him address it on the show. I’m exploring a semi systematic trading where strategies are primarily rules based but allow for occasional macro overlays. For example, adjusting risk or exposure based on a discretionary (I’m going to whisper it), ‘discretionary’ view that volatility is being underpriced over the next three months. Has Rob come across any useful frameworks, research and resources for designing systems that intentionally combine systematic structure with discretionary judgment? I’ll be interested in his thoughts on how to balance the two effectively.”

[00:37:55] Rob

So, this is quite an old topic in sort of quant finance. So, for example, there was a seminal paper written by Black-Litterman, in the early 90s, I think, where they basically took the kind of standard portfolio construction technology and said, well, what if you’ve got some people who you think can forecast returns? How do you incorporate that information into your portfolio?

So, the idea is you might start off with a market cap weighted portfolio like the S&P. And you might then say, well, my analyst thinks that Tesla’s going to go up by more than Nvidia. I incorporate that information in a quantitative way into my portfolio? So, there’s kind of long strand of people doing that.

My approach is a bit simpler, to be honest. So, essentially the way my system works is I convert everything into a forecast. And a forecast is just a number – it’s a scaled number. And you can think of it as a scale where 10 is kind of averagely bullish, minus 10 is averagely bearish, 0 is don’t know, +/- 20 are sort of super aggressively bullish or bearish, respectively.

And when you describe that to somebody, you can actually get them to do their own forecasts using whatever methodology they use that’s discretionary, whether it be reading news reports, looking at weird patterns or throwing the dice on the table. And then it’s quite simple to use some kind of weighted average of what those discretionary forecasts are combined with your more systematic forecasts. And the nice thing about doing that is you can assign them a weight.

You might say, well, I’m going to be 50/50, because I could have equal faith in each of these sources of information. Or you might say 70/30 if you have more weighted towards systematic. And because the forecast is a number, you can actually then independently verify how good that forecast is in exactly the same way you could verify a backtest, although of course you don’t normally have as much history. And you can use that to weight the returns. And I like this approach quite a lot to the extent that I’m actually writing a book on it.

[00:39:53] Niels

That surprises me a bit.

[00:39:56] Rob

I’m writing a book, yeah. It’s not exclusively about that, but a big part of it is the idea. I mean, the book’s got the provisional title that will almost certainly change because one thing I’ve learned about writing books (and this will be my fifth one) is that I’m terrible at coming up with titles. But the provisional title is Semi-Automatic Trading because it’s about the idea of combining these two things.

[00:40:17] Niels

Okay.

[00:40:17] Rob

So, yeah, that’s the way I would do it. Now, obviously, I think the question was phrased, talking about, I think volatility is going to be underpriced or overpriced. I’m less keen… Systematic methods of forecasting volatility are extremely good, and I think it’s tough to beat that. So, I wouldn’t probably mess around with my volatility forecasts.

My opinion is that, essentially, if we think about trading, I think there’s a small part of trading which is essentially deciding whether to buy something or not. Which could be discretionary if you know what you’re doing, if you’ve got that skill. Which I don’t think, really. I mean, I’m the guy that sold all my equities on April 14 or whenever it was.

But the rest of trading, so, everything else, so, deciding how big your position should be, as we’ve discussed, deciding how long to hold on to those, setting stop loss limits for your stop losses, and things like optimizing a portfolio for the most diversification, those things should all be done scientifically, I believe, using systematic methods, and the very, very best traders. So, hedge funds, like for example, Marshall Wace, essentially. Marshall Wace (other people do it now, but they saw that at first), they take what are, effectively, discretionary forecasts and they then put them into a quantitative portfolio optimization technique that comes out and says, well, these are the positions you should buy.

[00:41:51] Niels

The next topics, I should say, they came from you. We’ll see how many of them we get to. But the first one that you sent over was, to me, where I thought, yeah, why haven’t we actually discussed this before? It was so interesting. So, the background you said was that you had been at another conference. I mean, you seem to be invited to all these fancy conferences.

[00:42:14] Rob

I’m a popular guy.

[00:42:15] Niels

You are definitely a popular guy. But this one was more in our own territory, namely hosted by our friends over at AHL. And you said that one of the speakers came with a thought provoking comment. Maybe he came with an answer as well. I’d love to hear.

But he said something like, everybody uses the same methods, studied in the same schools, worked in the same hedge funds, but they all claim to be different. And it’s absolutely true, because that’s the first question we get from investors is, oh, can you explain to me how you’re different? So, I guess we all claim to be somewhat different, but I’d love to hear what he or she said and what do you think?

[00:42:57] Rob

Yeah, I mean, we’re an incestuous industry, right? Like, it’s not a very big industry. And you go to a lot of people and you’re like, oh, I used to work with your ex, or didn’t you used to work with so and so, oh yeah, I used to work with them. And it’s quite nice, actually, going back. It’s actually the first time I’ve been back in the AHL building since 2013. Not because I’ve been, you know, banned or something. At least I don’t think so. But just the opportunities hadn’t arisen. So, it was very nice to go back and see some old friends and old faces.

But yeah, the speaker, he wasn’t even a kind of… I think a lot of the really profound things people say are often just throwaway comments because actually from, I think it was the last podcast I was on, I made a throwaway comment and someone actually made a whole LinkedIn post out of it. Which made me laugh because it wasn’t something I’d spent any time at all thinking about, I just said it.

And I think this comment was the same. I think this guy just said it in response to a question. It wasn’t like he did a whole presentation on it. But yeah, it’s a very profound thing, isn’t it?

So, you know, it’s not just true of CTAs, because if we think about, I mean, we talked, we’ve talked about AQR, and we’ll talk about them again, possibly, in the podcast. But they tend to hire people who have been to a specific university in the US and have got PhDs in finance. And there are a lot of people in their corner of the industry (sort of equity market, mutual factor modeling, corner of the industry) who’ve done exactly the same.

[00:44:33] Niels

And maybe I can just interject, and you can correct me because I probably will get it wrong, but Winton was more linked to Cambridge and some of the other ones were more Oxford University as well.

[00:44:46] Rob

So AHL was started by three guys, all of whom did physics at Oxford.

[00:44:53] Niels

Wasn’t one of them a Cambridge guy? That was the funny part.

[00:44:56] Rob

Okay, well, they’d all done physics, that’s for sure.

[00:44:59] Niels

That’s true.

[00:44:59] Rob

And the joke was that their idea of diversity was hiring a physicist from Cambridge or a mathematician from Oxford. So, someone slightly different.

[00:45:07] Niels:

Perhaps it’s David Harding that’s a Cambridge guy, but he could correct me.

[00:45:09] Rob

Yeah, yeah, possibly, yeah, I think we need to Google that. But I think you’re right, certainly.

So, Tim Wong, who was CEO of AHL for many years, he was an Oxford physicist. Andrew Sinclair, who I think is still with AHL, was my first boss there, he was an Oxford physicist. I think Russell, who runs a place now, I think he went to Imperial, but he’s a physicist. So, there are a lot of physicists in AHL.

And so, if you go to sort of high frequency trading firms, you know, they prefer to hire maths, physics, computer programming type people. Obviously, that suits their skill set more.

And I think, actually, one of the advantages that Renaissance has, and it’s deliberate strategy of Renaissance technologies, is hiring people from mainly pure mathematics backgrounds with no finance experience because they want to look at the data with fresh eyes, if you like. So that’s that.

No, I think what Syed said was (Syed, by the way, was the name of the speaker who made this throwaway comment), what he said, and I sort of agree with to an extent, is it’s actually the small stuff. So, you know, you could give the same data and the same books to a bunch of people and ask them to create a trading strategy and they’d all do it in slightly different ways.

And if you looked at the people who are more experienced and more skilled, you’d see that they’re taking more care with certain things than the other. Like, for example, data cleaning, for example, that the other people hadn’t taken such care with because they didn’t probably necessarily think it was important.

So, in sports, specifically in cycling, there was this philosophy of marginal gains where the idea was you didn’t try and make a big gain, you did lots and lots of small things to improve. And I believe, I think, in our industry that’s where we are.

I don’t think there’s a kind of big, undiscovered, secret way of making money out there. And this isn’t just true of CTAs, this is just generally core to finance. Everyone has the same data, pretty much. Everyone has the same access to computing power. Everyone’s got the same mathematical models. The sort of market knowledge you need you can read in a book.

It’s not like, if you go back 40, 50 years or the worst sort of secrets in the markets that only insiders knew, that’s no longer true. So, it comes down to marginal gains, you know, slightly better execution, you know, cleaning your data better. You know, if in the high frequency world, then it’s marginal gains in terms of latency, just replacing your memory chip with a different memory chip, you know, stuff like that, really, really nerdy stuff.

And for us, I think it’s, yeah, things like sort of optimizing portfolios to try and make the best use of this vast variety of markets we’ve got, some of which aren’t very liquid, and doing so in such a way that minimizes costs. I think that’s one of the… If you look at the good managers and the bad managers… that’s a bit unfair…If you look at the profitable managers and the less profitable managers, because there’s a lot of luck. It’s not necessarily whether they’re good or bad. I think it’s hard to tell.

Often if you read the annual reports of the managers that have had a bad year, they’ll say, oh, well, you know, we did X, Y or Z. And in retrospect you go, well, why did you do that? It’s an obviously stupid thing to do now. So, was that really bad luck or did they do a bad thing by putting too much money into an illiquid market?

And if I think about mistakes I’ve made in the past, trading too quickly, over concentrating individual markets. Thinking back when I was trading fixed income OTC, market capacity can vanish like that. And it did in 2009.

So, it’s a lot of small details. Another way of thinking about it is rather than marginal improvements it’s avoiding errors.

[00:49:16] Niels

Yeah.

[00:49:17] Rob

You know, it’s doing, doing a thing that’s quite simple to describe. Maybe not simple, but doing a thing you can describe. You can describe everything. I could describe everything I do, to somebody, but then I’m doing it in such a way that, because I’ve got a lot of experience, I’m avoiding errors. So yeah, I think that’s what it comes down to, to be honest, marginal gains and avoiding errors.

[00:49:37] Niels

Sure. One thing that it kind of linked to this question a little bit about being different and not. And maybe, once in a while, one firm will come up with a slightly novel new idea and they’ll be ahead for a little while.

But the other thing I was thinking of, when I saw the question, and that is kind of this potential contamination of IP if you’re located more or less in the same city. Meaning do people move from one firm to another firm down the road and then suddenly they kind of both, even though they’re not allowed to or whatever, but you can’t remove what people have learned. And so, I wonder if that also makes it harder to be truly different over time. I don’t know. Have you ever thought about it?

[00:50:20] Rob

Yeah, I mean it’s an interesting one, isn’t it? Because you’re right, I mean the sort of, you know, AHL, Aspect, Winton, you know that they’re all within a few…

[00:50:30] Niels

Plus many more in the same…

[00:50:32] Rob

They’re all geographically within a few miles of each other. So, you know, if you’ve got a job at one it’s going to be quite straightforward to get a job at another. And you know, most sort of quantitative hedge funds are concentrated in perhaps three, four places in the world. You know, there are exceptions obviously.

[00:50:50] Niels

Sure.

[00:50:51] Rob

Some like Transtrend, for example, springs to mind. But, you know, that’s just inevitable, I think. And I think the same is true in the tech world. And although we do have these kind of famously long noncompetes in gardening at least. I know people have spent most of their lives gardening and being paid for it, which is a nice position to be.

[00:51:11] Niels

Same is true, by the way (sorry to interrupt you), the same is very true among these pod shops, I understand.

[00:51:16] Rob

Yes. Yeah.

[00:51:17] Niels

Lots of people move from one pod shop to another and you know, so that’s interesting

[00:51:20] Rob

I think there are trading strategies that are less systematic, and perhaps also higher frequency, where kind of ‘secret knowledge’ Is more important and you want to take more steps to…

I mean there’s that case recently, was it Jane Street? I can’t remember they had a strategy trading in markets, options in the Indian market and they were very cross that some people had left Jane Street or come to Jane Street (I can’t know which way around it was. It might not have been Jane Street), and they were very cross about that.

But I can’t imagine, you know, someone leaving, say, a big CTA and going to another big CTA and taking some big secret with them. I think the advantage that they would bring, more certainly, rather than any big secret, is just a fresh approach and sort of different pair of eyes, and saying, well, the way you’re doing the stuff here is cool. Actually, this bit is better than what we did back where I used to work. But this, have you ever thought about doing it this way?

You know, just, just a fresh pair of eyes, just suggestions and creativity, and that can come just by hiring people from a different industry as well. You don’t need to necessarily hire a competitor.

[00:52:25] Niels

Let’s see if we can tackle at least one of the papers, maybe both, because they both come from AQR, one of our favorite providers of content, of course. The first one is actually from a previous guest on the podcast, namely Antti Ilmanen, who has great insights into many things when it comes to investing. And the paper is called Why Are Bond Investors Contrarian While Equity Investors Extrapolate? It’s kind of an interesting thing.

I won’t steal your thunder because I know you probably read this more than I did, but it is an interesting idea, and I want to comment on it once you’ve done your kind of overview.

[00:53:08] Rob

Yeah, and I think I’ll start actually by saying why I think this paper is important for people in the momentum industry.

So, one of the things, right at the start, I talked about the podcast that Cliff had done on the money stuff, one of the things he talked about was what are the explanations for momentum? You know, is it under reaction, is it overreaction, is it behavioral, is it a risk premium? And so on and so forth.

So, if we think about sort of behavioral biases, if people in a particular place, you know, a particular market, think that asset prices or economic factors, or something like that, will mean revert, will kind of always tend to return to normal and they don’t like, if they’re wrong about that, then that obviously is a fantastic place to do trend following. Because you’ve got people saying, well, you know, I don’t know, think of a random asset class, gold. You know, gold always mean reverts in price. Like, you know, it’s always going to be in a sort of $500 to $1,500 window. You know, it’s just the way it works. And they make forward looking expectations on that, on that basis.

And you know, Bloomberg can get these economist forecasts, and things like that, for economic numbers. And they thought, oh yeah, gold always mean reverts. And then if it doesn’t mean revert, then trends, and essentially the consensus is wrong, well, that’s a fantastic place to be a trend follower.

Conversely though, you can imagine a market where people expect trends, they expect things to carry on in the same direction, and where they’re wrong, the other way around. Well, that’s clearly going to be a good market to not do trend trolling. And, in fact, you want to do some kind of mean reversion.

So actually, this idea of how people form consensus expectations and make forecasts is actually core to a potential explanation as to why. Well, firstly, why does momentum work at all? Why does trend following work at all? But secondly, is there a plausible reason why it might work differently in different asset classes?

So that’s the introduction out of the way. That’s the explanation as to why you should spend the next five minutes listening to me described what is not an especially long paper. AQR and Antti, in particular, are good at writing quite terse, readable papers with lots of nice graphs. So, it’s worth spending a bit longer reading the paper yourself. You can find it on the AQR website, perhaps link from the show notes, I don’t know.

Anyway, so there’s a couple of really nice graphs in that really illustrate exactly what’s going on. So, one is a graph of yield expectations in bond markets. So, what’s expected to happen to yields in the future?

And, actually, these are quite often used, these charts, to illustrate how terrible central banks are at forecasting. So, something that people will do is say, well, this is where the bank of England expected interest rates to be.

And I call them whisker plots because there’s kind of a hard line with the actual interest rate. And at each point in time they show what the forecast was for the yield to do in the future. So, there’s kind of whiskers coming off the plot. And each whisker represents a forecast made at a point in the past.

And what typically happens is that, and it isn’t just central banks, to be fair, it’s the market generally are actually shocking at this. So, if we take, for example, the period of very low interest rates between 2009 to say, 2019, ’20 (depending on if you take Covid out essentially or not), you will find that consistently, for the whole time, people were expecting interest rates to go back up to 4%.

So, they were like, oh yeah, it’s 2009, interest rates are currently basically zero, but they’re going to go up to 4%. Four years later, it’s 2013, guess what, interest rates are still at 0% but expected to go up to 4%, and so on, and so forth. And we see a reverse effect happening. When rates are high, people expect interest rates to fall.

So, if we go back to say, I don’t know, 1999 when interest rates were quite high. So, I think T-bills were like 6%. People were expecting rates to go to 4%, 5%. So basically, it’s quite easy being an interest rate forecaster. You just say oh yes, 4%, 5% and your job’s done. It doesn’t matter where or what yields are, that’s what you always forecast.

So, bonds are basically, bond markets are a market where people think that interest rates are going to mean revert and they don’t, they don’t mean revert. They persist for long periods of time, very low or much higher than expected. So that’s bond markets.

And then, if we turn to equity markets, they do that sort of in a slightly different way essentially. Well, they look at analysts expectations for EPS growth and there are sort of a couple of things going here. One is the fact that analysts generally are over optimistic and there are institutional biases as to why that might be the case. And some of those biases went away after the, was it the Dodd Frank Act? I can’t remember anyway, but after the tech bubble the analysts became a little bit more honest. But that’s not the interesting thing here, even taking out the over-optimism.

What’s more interesting is that equity analysts are exactly the opposite. So, rather than expecting things to revert to a long term mean, like bond investors do, they’re the exact opposite. They think that if times are good they’re going to continue getting good. If times are bad, they’re going to continue getting bad. So, they’re much more emotional, equity investors, you might want to say.

Now that means that equity investors are expecting momentum and like more momentum than there really is. So, that’s kind of interesting.

So, this is a very interesting finding. And Antti goes into a lot more detail and tries to explain it, and so on, and so forth. But I want to return to my original introduction which is so, well, what does this mean for us as trend followers?

[00:59:32] Niels

Okay.

[00:59:33] Rob

And actually, if you look at the performance, by asset class of trend following across different asset classes, you do find indeed that trend following in bonds has done very well indeed. Trend following in equities has done much less well, in equity indices just to say. So, this is for CTAs or people like me just trading at the index level in futures. Individual equities are another story.

So, this finding does actually kind of support the idea that, well, one of the reasons why trend following works really well in bonds is because people in bonds just think mean reversion is going to happen and they’re wrong. It doesn’t.

Whereas equity investors essentially think momentum is going to happen and therefore it’s sort of ‘priced in’ and in fact overpriced in. So, in fact, they don’t see the crash coming, they don’t see the recovery coming. And so, there’s more mean reversion in equities.

So that might be one of the explanations as to why this asset class difference in performance exists. Now there are other explanations as well, like, for example, carry tends to be higher in bonds and that means that you will earn carry which creates a secular upward trend in bond prices.

And, of course, there are long-term effects, actually, in both equities and bonds. There’s been a secular downward trend in interest rates since know the last 50, 60 years, there’s been a secular upward trend in equity valuation. So, both of those things kind of give a tailwind to momentum in both cases anyway. But, that tailwind’s maybe been a little bit better in bonds. I don’t really know.

But yeah, I find it a very interesting paper. One of the things I do like thinking about is where does the money come from? So, I mean Antti has a good presentation that’s worth searching for where he says who’s on the other side when you’re trading, who’s on the other side? If you’re trading these common risk premia like value momentum, who are you trading with? He thinks that’s a useful question to ask, and I agree with him.

[01:01:27] Niels

Yeah, and I really appreciate you doing that because also I had written down exactly the same, before we recorded today, and saying well this is exactly probably what we see in our own returns, that the bond sectors have done are so much better over time. And even though a lot of people, I think, would say well the equities have gone up for the last 15 years, how can you not have made money in equities? But they forget all the V shaped torture we go through and all of that. So anyways, very good.

Now, before we leave, I think we should give a little bit of time to one of our favorite content creators, namely Cliff Asness himself, because he wrote a letter. Not a letter, a paper that I may have touched on briefly a couple of weeks ago, I’m not entirely sure. But it’s one of these where he goes back and criticizes his own writing, many years ago, because it is an argument.

And the argument is, I should say, what if you miss the best days of a market? Meaning you lost out on whatever day we saw recently in the S&P where it went up by 9.5%, and so on, and so forth. And, of course, he says there are lots of big institutional firms that make this argument to maybe keep investors in their funds.

And frankly it’s also something that I’ve seen in our industry. But let’s keep that aside. I’d like to discuss that separately. But maybe you could briefly describe what his new thinking is, if any.

[01:03:07] Rob

Yeah, I mean, I don’t think it’s new thinking. I think he basically was pointing out that it’s a bit of a silly argument to make because, of course, you can very easily traverse it and say, well, what if you could avoid the worst X days in the market?

And because equity returns are fat tailed, actually, in both cases that would be a good thing to do. Actually though, because equity returns are also negatively skewed, and the worst days are much worse than the best days are good.

If you did have a choice between avoiding the worst days or missing the best days, you would choose avoiding the worst days even if it meant missing the best days as well, because those worst days are really bad, much greater than the returns you get from the best days.

So, that was the thesis he had. He wrote the first paper in 1999, and obviously it’s a quarter of a century on, and we’ve gone through the tech bubble since then, and we’ve had some very nasty days indeed on top of those that we’d already seen before that. So yeah, I mean, one of the nice things is for him to say, well, generally speaking, it’s bad to try and time the market, particularly (and this may surprise people because I’m a trader, right, which means, personally, I’m trying to time the market) because there’s a difference between timing the market in the sense of saying, well, over the last three months the market has been trending down, therefore, I think I should no longer be owning the market and go short it, for a bit, until it starts to trend back up again.

There’s a difference in doing that and saying, right, it’s Tuesday 11th of November, I think tomorrow, Wednesday the 12th, the market will go down. Therefore, I’m going to close every single position I have and sit on cash for a day. No one does that. No one does that. I mean, there are people who have a forecasting horizon of a single day, but that’s not how they trade. They don’t trade like that. They’re trading all the time. They’re not selectively choosing to miss good or bad days. It’s just an absurd argument.

So, one thing he does that’s quite nice is to do some simulations. I’m a big fan of simulations. I love simulations because they allow you to kind of extrapolate from the real data in a much bigger sense and say, well, there’s a bunch of alternative history.

If you are going to go around avoiding bad and good days, well, let’s be more realistic and say that actually, it’s happening on a random basis and you can’t actually forecast what effect that will have on your return distribution. That’s, for me, a better way of doing it because it ultimately reinforces the argument that it’s a really bad idea to just close your positions for a day and then open up next day. Because, essentially, what it does is increase the dispersion of your returns over time. And that’s why his statistical work shows. Trading costs, taxes, there are all kinds of reasons why this is a really, really bad idea.

Yeah, it’s a silly, whole, oh, you know, missing the best days of the market is a marketing thing. And it’s a bit pathetic, frankly. Like if you’re a long-only equity investment fund. Sorry Niels, I’ve been diplomatic for most of the podcasts. I’m turning the diplomacy off now. You know, I’ve had enough. If you were running a long-only equity investment fund and you were trying to persuade people of the benefits of long-term investing, well, just point to the long-term track record and say, yes, equities haven’t gone up for a long period of time.

You know, past performance is no guide to the future. But you know, I’m happy with the thesis that if you have increased risk, you’ll get higher returns. Equities are quite risky therefore, if you’re investing for the long-term, buy equities.

Why do you then need to turn around and say oh yeah, you should buy equities because if you miss the best few days in the market, you know, you’ll be in trouble. Just be honest and say, yeah, long-term returns are good, most people can’t time the market for the better or the worse.

[01:07:20] Niels

So, let me broaden this out a little bit before we wrap up. And that is turning it to our own industry, where a lot of people would still ask, when is a good time to invest in trend following? And I also meet clients and prospective clients who will also have an opinion about when not to invest in trend following because they think, oh, the next three months or the next six months is not going to be good for trend following.

So, one answer to that of course would be, you know, you just stay invested because we don’t know when the next big month was going to come and you don’t want to miss that. And we also know… Well, I don’t say we could also know, but from memory, and this is maybe just a mathematical thing, some of the best returns actually come after your worst returns period. And we’re certainly hoping that will be the case this year as well. But what are your thoughts on the argument when it comes to an active strategy like trend following?

[01:08:23] Rob

Yeah, I mean, so obviously it’s a subtly… Well, not subtly. It’s a bit of a different question because we’re not talking about missing the worst end days in trend following, we’re sort of saying can we miss,

you know, a month or two, or even that’s a bit silly, I think. If you could put it in advance that 2024 and (the way things are going) 2025, were bad years for trend following, you know, would you just sit out the market for a bit or lower your allocation, you know, is that possible?

So, for that to be the case (I’ll put my nerdy hat on again for the last time in the episode) you’d need to have a negative autocorrelation of returns at the appropriate time frequency. So, in other words, you would need to have a meaningful autocorrelation returns at the appropriate time frequency. Now, if that autocorrelation was positive, that would mean basically you should trend follow trend following.

If trend following had a good year and there was a positive autocorrelation between annual returns and trend following, then if there’s been a good year, you should invest more or invest for the first time because it’s going to be followed by another good year. Conversely, if the autocorrelation was negative, that would mean that good years tend to be followed by bad years, as you say that. And kind of, from a sort of narrative perspective, that feels like what happens.

Now, I’m not the only person who’s looked at this. Winton looked at this a few years ago and also, there’s a sort of subtle methodological thing to say, which is that the effect of performance fees, so performance fees actually make your returns more negatively autocorrelated because you only take performance fees out of when you’re doing well. So, that means the good returns aren’t as good and the bad returns aren’t as bad because you’re not paying bad returns when you’re in a drawdown. So, that does introduce some bias. And you can Google the Winton paper at your leisure.

I looked at the same thing without the effect of fees. So ,what I actually found was that, for momentum specifically, if I looked at, say, annual returns, I did find there was a bit of weak positive autocorrelation. In other words, good years tended to be followed by good years. That autocorrelation went away when I looked at longer time periods.

So, if it’s been a bad couple of years, you’ll probably get a slightly better couple of years. But the thing to really, really emphasize here is that these effects are incredibly weak, incredibly weak. And you would really be crazy to try and use them in any kind of momentum market timing strategy because they would definitely get eaten up by fees or costs. So yeah, I think the lesson is market timing is difficult. And, you know, strategy timing is also difficult. And I know that Cliff Asness agrees with me on that. So, there we go.

[01:11:24] Niels

I do too, because I listen to the same podcast as you did. Anyways, good stuff.

This was great, Rob. I’m very, very grateful for all the efforts you put into reviewing this and answering questions. And hopefully everyone listening got some value from this. If you want to say a big thank you to Rob for doing this, head over to your favorite podcast platform, leave a rating and review. We do appreciate them, we do read them, so we would be grateful for that.

And next week we have a new co-host, Otto van Hemert, the Director of Core Strategies at Man AHL. He’ll be covering lots of new ground with us, so this will be very, very interesting indeed. Someone that Rob knows, of course, and if you want to ask a question, you can do that as well. You should email them as soon as possible to info@toptradersunplugged.com and I’ll do my best to tee it up for Otto.

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

[01:12:33] Ending

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

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

And remember, all the discussion that we have about investment performance is about the past, and past performance does not guarantee or even infer anything about future performance. Also, understand that there’s a significant risk of financial loss with all investment strategies, and you need to request and understand the specific risks from the investment manager about their products before you make investment decisions. Thanks for spending some of your valuable time with us and we’ll see you on the next episode of the Systematic Investor.