Position Sizing When Markets Break feat. Rob Carver | Systematic Investor | Ep.386
Summary
Precious Metals: Extensive discussion of silver’s extreme volatility on “Freaky Friday,” liquidity dynamics versus gold, and how volatility-adjusted sizing and stops impact CTA outcomes.
Natural Gas: Highlighted massive weather-driven swings (60% up, then -26% in a day) underscoring why commodities are critical to trend-following universes for diversified opportunity.
Cryptocurrencies: Bitcoin’s sharp selloff featured prominently; many CTAs likely short via futures, with crypto also framed as an esoteric/alternative market with idiosyncratic trends and crisis constraints.
AI: Debate on the AI cycle and finance/AI lab convergence, including where AI/ML fits best in systematic workflows (e.g., execution), and skepticism about broad, unfocused applications.
Alphabet (GOOGL): Noted plans to roughly double AI-related capex to ~$185B, raising bubble concerns despite strong profitability; indicative of intensifying AI infrastructure spend.
Alternative Markets: Review of Man AHL’s paper showing alternative/esoteric markets deliver more idiosyncratic trend returns and enhance diversification versus traditional futures.
Portfolio Construction: In crises, traditional futures dominate crisis alpha; cash efficiency matters, pushing allocations toward highly liquid futures when investors may need withdrawals.
Macro Context: Discussion of the incoming Fed chair implications, cross-asset dislocations, and why systematic, diversified risk management can outperform discretionary calls in turbulent periods.
Transcript
Imagine [music] spending an hour with the world's greatest traders. Imagine learning from their experiences, their successes, and their [music] failures. Imagine no more. Welcome to Top Traders Unplugged, [music] the place where you can learn from the best hedge fund managers in the world, so you can take your manager due diligence or investment career to the next level. Before we begin today's conversation, [music] remember to keep two things in mind. All the discussion we will have about investment performance is about the past and past performance does not guarantee or even infer [music] anything about future performance. Also understand that there's a significant risk of financial loss with all [music] investment strategies and you need to request and understand the specific risks from the investment manager about their product before you make investment decisions. Here's your host, veteran [music] hedge fund manager Neil's Krup Larson. Welcome or welcome back to this week's edition of the systematic investor series with Rob Carver and I Neils Castro Llassen where each week we take the pulse of the global market through the lens of a rulesbased investor. Rob, it is uh wonderful to be back with you uh this week. First time in 2026. Hope you're doing well. How are things in the uh in the UK? >> Yeah, it's doesn't seem I mean 2026 is only a few weeks old, but it seems like a hell of a lot's happened. So, particularly in the market. So uh so yeah, it's all quite exciting. >> It is all all very exciting. And actually speaking of exciting, we have a great lineup of topics uh today um which I think people will really enjoy. U we're going to be tackling a couple of new papers, an article and uh a couple of questions that came in. So uh so this is all uh super exciting. Rob, as you know, I always I'm always curious to uh hear what's kind of been on your radar since we last spoke. um but not from the topics we're going to be talking about. But if there's something else that you found interesting um then uh >> let us know. >> Yeah. Well, I was I was hoping by now I would have seen the new Melania film, but um unfortunately I've not had that opportunity and it looks actually looking at the viewing figures worldwide, not many other people have as well. So, so hopefully I'll get a chance to review that next time I'm on. Um so I've saw an interesting um paper published by Alliance Bernstein. I don't want to talk about it today, but what made me um sort of impressed was the title of this paper. Um and I actually for reasons that were becoming obvious, I actually had to ask my son who did German at school how to pronounce this word in this title. So the title of the paper is about TPA total portfolio approach which we've discussed before. title of paper is portfolio design as um Gazamkensk the total portfolio approach and apparently a gazam is is to is something like a total piece of art and it's apparently it's a term coined by Vagnner in relation to his operas. So, uh, so yeah, that that intrigued me. But it it goes to show if you're working as a salesside analyst, you you desperately got to try and make your stuff interesting, right? And one way of doing that is coming up with a title that that makes people stop and think and go, "Well, that sounds weird at the very least." And it worked because I to be honest, I probably wouldn't even have glanced at that paper were it not for the the funky title. So, uh, so yeah, if you're interested in TPA and Vagner, then uh find that Alliance Bernstein paper and enjoy it. >> That is quite funny. I mean, since I have to come up with a lot of titles and headlines every week for the podcast episodes and the emails we send out, uh I'm always interested in in a catchy uh method of uh getting people to uh you know, listen or open what we uh what we share. I have to say I didn't expect a German word to do the trick, but I might have I might have to try it out. >> Well, there we go, Neils. There we go. All those long compound German words just waiting to be used in your your email subject lines. >> Clearly. Clearly. Okay. Well, let me tell you what caught my interest. And I know this first topic definitely caught your interest as well. And instead of putting it in the trend following section, I think we'll we'll discuss it now. Um maybe spend a little bit more time on it. And it is what happened uh in the last uh week or so. Well, actually the last uh few months, but it kind of uh really came to a head uh on the last day of January. >> Freaky Friday. I'm I've decided I'm going to christen it freaky Friday. >> Freaky Friday, especially if you're trading precious metals and especially if you're trading silver. And actually, I would say this week uh has been pretty uh interesting as well. So why don't you tell us a little bit about your silver experience and um maybe we can then talk a little bit about kind of the the good old days and and why silver is not gold uh in in some way. >> Yeah, I mean so um I obviously trade silver um it's a a futures contract. It's liquid so it would be on the table for me to trade. Now, I've talked about this before, but the way I trade um futures is a little bit complicated, and it's mainly to do with the fact I have a relatively small portfolio. So, I have a notional exposure to 250 futures contracts. Um many of those I couldn't trade anyway because of regulatory restrictions. Of course, they're not liquid or they're too expensive. Silver does not fall into that category. Um, and then I basically dynamically optimized my portfolio every day to match to get the best possible match I could do to a sort of abstract portfolio, which I could have if I had a notional value running into the hundreds of millions of dollars, which sadly I don't have. So that will mean that it often that I won't necessarily have a position in a given instrument even if I have quite a strong forecast on it. Um, now silver though I do have a position and I did have a position in I should say. Um and if I look at my history of exposure to that position, uh it is quite interesting. Um now again it's slightly complicated by the fact that that I have this dynamic optimization but broadly speaking you can sort of say that CTAs generally will have positions that are a function of two things. The strength of the trend and the volatility of the instrument you're trading. Um, now that's obviously a broad brush thing because of course some CTAs will take basically binary positions. So they'll, you know, they'll kind of go from fully short to fully long. Um, others will sort of do what I do, which is sort of continuous adjust adjusting forecast. So getting, you know, the moment the trend starts to look positive, they'll kind of increase their forecast and that would increase the position, all other things being equal. And of course they're not, which is what we'll get into in a second. Um and also of course some CTAs take volatility into account when they initially buy a position but then don't make any further adjustments to that position as volatility changes. Um so we can characterize the silver story over the last few months as a a positive bullish signal getting stronger. Um this is all up till last Thursday obviously because obviously on freaky Friday things change very dramatically. Um but at the same time the volatility also increasing. Um so my position broadly speaking and the position of anyone trading like me would be a function of those kind of two effects. If I could actually look at my position of of my exposure to silver I can see that um although I was trading it on and off last year um the kind of current position um I I bought some on the 27th of November and I bought some more on the 3rd of December. Um and that would have been down to the the forecast increasing um and then actually on the 31st of December I sold a contract and that would have been inevitably because the volatility had increased. So though the combination of those two effects often lead to these interesting effects um and then subsequently um on well actually not on freaky Friday itself because I have a daily rebalance. So I didn't do any trading in silver during Friday itself cuz my position was based on Thursday's close which was kind of normal if you like. Um but so on Monday, so that's a couple of days ago now. Um I actually then sold that um contract and that would have been again because volatility had increased. Um it's my trend signal is still long because um you know unless you're trading an insanely quick trend following system one day of negative returns even a day like Friday where silver dropped by the most in one day since and you know you you'll remember this Neils of course about >> it was decades at least >> decades ago many decades ago I think 40 45 46 years 47 years ago to exact. Um, so that's why I then I then sold sold out of that that silver position on that day. So that that's kind of interesting. Now a big question is did I you know useful question to ask is well did I actually make money doing this? And the answer is yes I did because although silver dropped precipitously on Friday, it only dropped back to the level where it had been a few weeks beforehand, a couple of weeks beforehand. Um so I can actually give you hard numbers. So, I bought I bought into my silver position um at levels of $53 and $59 uh per ounce, respectively. Um and then I sold at the end of December at $71. Um and I sold again at $81. So, just looking at those trades, you know, that was nice. That's a profitable trade. Um, of course, what that misses out is that in between the the end of December and the the freaky Friday, the price briefly went up to, you know, $110 $120 an ounce, I think it was. Um, so that that's kind of that's kind of my my kind of story with silver. Now, um obviously that would be different for someone who doesn't do the continuous volatility adjustment because what they would have done is seen the value of the silver in their portfolio kind of get bigger and bigger and bigger um and then not done anything about it and just hung on to that. Um and as a result would have basically would have seen a massive run up in value and then freaky Friday would have resulted in a quite substantial drop in value. Um and we've got the sort of monthly returns for January now for many CTAs and uh certain CTAs and many viewers this podcast will know who we're talking about who take very large um leveraged positions and have very high risk and do don't do this volatility adjustment. I think at one probably went from being about sort of 50% up for the month to being 25% up for the month. Now 25% up for the month is still amazing. Um, but the actual one day loss on on freaky Friday would have been, you know, very substantial indeed. >> I I think I can help you with the precision. Actually, think about being up 72% on the Thursday and ending the month up 27%. No names here of course, but it is it was it was a large move. I will say by the way there is I guess one other way that you could even though the you could say largely the trend is still up for silver in many respect. But of course if you were using a hard stop uh as one of your rules and that stop may be calculated from what has been the most recent high of some sort maybe with a V um stop uh linked to it. I guess you could have also been stopped out on Freaky Friday because the move was so big, but it really depends on your specific rule. >> It does. Yeah. And because I don't use um stop-loss rules. Yeah, you're absolutely right. Any any stop-loss I mean, I'd say that because um it depends again on how what you do with your stop-loss rules as your position runs up. So um when I if I was to run a system using stop losses which I don't as I said but I design it so that the the initial entry the stop would be set as a a a rat a multiple of the current volatility. Um now if you as the price runs up and the volatility increases if you then also increase your stop at the same time then the position you know having said that you know a 25% fall probably still would have breached any reasonable stop >> but importantly if you do do that if you do increase your stop width you also should reduce your position at the same time otherwise you're basically taking more and more risk effectively on the position. um if you don't increase your stop level as volatility rises then you you're you risk basically being shaken out of the position too early. Now on Friday that would have been fine because it would have been very nice if you'd I don't know you know how realistic would it have been to have executed at a tight stop on Friday. I don't know how how rapid the move was. Um but um it would have been you know you probably would have done better than somebody with a much looser stop. Um but you know that also meant you would have been shaken out on a much smaller move in the middle of a bigger trend which would been the wrong thing to do. So um you've got to be careful if you are using stocks how you adjust them as volatility changes just as I adjust positions as volatility changes. You you mentioned kind of you know if you had the ability to uh to uh trade on Friday um clearly it was a hugely volatile day. There's a not that many people I've I've heard from in terms of a logical explanation as to what happened. I've seen some speculations about a a large US investment bank causing all of this, but I don't want to speculate on air on that. But do do you have a sense or it's something you you monitor in some way how how liquid because I think maybe because silver and gold and platinum, they've all been going higher uh in the last uh many months and of course it's been great for trend followers. Maybe people have kind of maybe become a little bit complacent thinking there these are all the same to trade and they think oh they're all going to be as liquid as gold which is super liquid market um but but but they're not the same um in in my view and um and I so I don't know from your perspective in terms of liquidity and those kind of risks um if you have any thoughts on what happened. Um >> yeah, I mean obviously because I'm trading a relatively small amount of money, my kind of have a fairly broad liquidity filter. So I I would probably take positions in in um in instruments where a lot of big CTAs would either not bother or have a very small position indeed. >> Um so silver, you know, there's no no question of silver not falling into the liquid bucket for me. So that that wasn't an issue. But I I think the one kind of thing that does happen um in in markets that are not as liquid as you know the most liquid markets is that obviously if something runs up in price and you get a slight pullback well all the buyers are gone right everyone who wants to buy is bought. So there's no one sitting around waiting to buy off you. Um so you know people treat liquidity as a constant number as a constant number of volume of contracts they can expect to trade. But of course it isn't right and we we know that in market crisis liquidity disappears and that that happens everywhere. Uh and obviously the less liquid the market is to begin with the more likely that is to happen. Um, and I think there is a risk if you if you know, let's say something gets hot, right, and it runs up in price and lots of people start trading it, maybe you hadn't traded it before, and you look at the volume and the open interest numbers and you think, well, I can build up a pretty substantial position here because there's a lot of volume going through, there's a lot of open interest going through. Um, you build up that big position and then and then if the class, you know, if the price starts to move adversely against you, then everyone jams into it's like the the the cinema with only one fire door, right? everyone gets jammed trying to get through the exit at the same time and the liquidity you thought was there wasn't there and part of that's just that that can be driven by a few things right part of it could be driven by the fact that yeah a lot of these people are silver tourists who don't normally trade silver um and the you know what you really need to do is go back a year or two years and say well over a longer period of time what does liquidity look like um the other thing is the classic thing where if everyone's trading similar kind of strategy I mean if everyone is trend following um if everyone is doing vol adjustment um and that does you know don't forget that CTAs do vol adjustment but so do risk parity funds um and so do other kind of quant strategies um so anyone who sells on the hint of volatility spiking that the more people there are in a particular market like that then will magnify the move I mean the approximate cause of all of this supposedly was the fact that a a relatively sane person got elected to the to Fed chair next and we'll talk about that in a second I guess but as to the mechanics of why it affected silver more than gold and silver went down more in price than gold. I mean, we can speculate all day, but I think the main lesson here is is for me is risk management. Um, and the two key things about risk management are position sizing um and diversification. Um, so on Friday, I lost 3% of my total account value because I have a reasonably diversified portfolio. Um, if I was very highly concentrated in silver, that would have been a much bigger number. Um, and because I adjust my positions according to volatility as it changes, again, I was relatively protected from a large move, whereas, as we've mentioned, funds that don't do that and have much more leverage potentially are exposed to even bigger moves. And yeah, month over the month, still profitable. Great work, guys. But, you know, it scares the hell out of me, the possibility of that happening, I have to say. Yeah. Well, first of all, I mean, I think just to comment on that number, I think that that the number you mentioned was pretty among people with that kind of annual volatility that you run it at. I think 3% is is very decent and I think that's kind of what I saw on on the day from the people I can uh follow. Um, so I think the industry as a whole did really really well. Um and as you say of course the larger CTAs will be doing something similar to what you do meaning uh adjust the position along the way which means again just maybe to it reiterate for people because I remember during the uh cocoa debacle where prices also had a kind of a similar move as we saw in silver where it becomes kind of a parabolic and gold for that matter kind of a parabolic move. It's very easy to blame the quant funds uh to be behind the parabolic move, but I would hazard to to to say that most CTAs probably got into silver more than a year ago for sure. And the majority of the larger CTAs, they they're the ones who have been selling silver all the way through January, as you rightly said, maybe even through um December as volatility started to uh expand. We're just going to have to reduce our notional exposure. Um so we're actually kind of trying to to be the buffer here um and and limit these parabolic moves. if if anything I think that is important to say because most people will will think the opposite. Um, the other thing you touched on is risk management. And I just want to maybe share, and I I I I won't do this story justice in any way, shape, or form, but of course, a lot of people remember silver specifically from the story of the 1970s where the uh I think Texas oil billionaire Nelson Bunker Hunt. He became convinced that the US dollar uh would would fail. I mean, it's kind of almost like deja vu uh at the moment. Anyways, in his mind, silver was not real money. That was kind of or I I [snorts] should say silver was not just an investment for him. It was real money. So, he started to buy silver and he did it physically and he chartered, you know, several of these Boeing 707s and he flew that silver, as far as I understand, uh millions of ounces of silver to Switzerland to be stored in in vaults. Um and um and so again this for him was not about making a lot of money necessarily. It was about kind of controlling and making sure that the wealth he had built uh wouldn't disappear. Well here comes the problem. I mean he kept buying but he kept buying now for borrowed money. So the leverage of his position um became you know quite high. Um and very similar story as to what happened uh last week. Um as the volatility uh rose uh what happened in in 1979 price-wise was that silver went from $6 to about $50 an ounce. Um and that causes a problem for the exchange. So com started to change the rules on margins. We saw the same thing happening I think in January one or once or twice where margin got increased and at some point the people who have b who have made made money but they've made money from borrowed um or speculation from borrowed money to keep up with the margin calls at some point at some point they run out of of of money and so did he so um I think don't know the details but uh he ended up having you know a catastrophic uh experience experience. Um maybe he still ended up making a little bit of money but but not nearly as much as he had. And I think there was some agreement about paying back um debt. So risk management um and not letting positions get completely out of control by pyramiding and so on and so forth are all very important lessons. And even though this happened back in the 70s, as we can clearly see, they get repeated over and over again, just in different markets at different times. Um, but I think the beauty of all of this is the fact that many of the managers that we um we talk about, they've all been around for decades. We we've kind of know this. It's part of our DNA. We don't get emotionally attached to it. We manage our risk in a systematic, dispassionate way. And that's [snorts] super important. And regarding this idea that people should also remember that gold is not sorry silver is not the same as gold. I've heard stories or or information about the fact that the reason why this is important is that silver you don't mine very much silver directly. It's only like 30% of the production actually every year is specifically mined as silver. The rest comes from a byproduct of gold and copper. So it's maybe less controllable or the supply side maybe is a little bit uh harder to to um predict and therefore it can um maybe also be the reason why silver can be a much more difficult much more volatile market to trade compared to gold. But these are all very important points in my opinion. >> Absolutely. I think it's may also worth briefly mentioning I think that Friday and actually continuing to this week we're seeing a lot of dislocations across markets generally. So you know long short quant equity factors have gone to barmy. Um apparently AQR's um long short fund had both one of its worst days and also one of its best days in the last few days. So that's interesting. >> Um Bitcoin of course is selling off massively. Um there's been kind of disruptions in the particularly in the sort of stocks related to AI. So it's all kind of interesting and at least at the moment you know touchwood fingers crossed um you know my trend following portfolio seems to be coping pretty well with all of this. Um so so yeah it's an interesting time. >> Can I just mention I mean we talk about precious metals as if it was the only thing that moved. >> Yeah. >> March natural gas uh just to just to mention from the 9th of January to the 30th of January it um was up something like 60%. And then on the 2nd of February in one day it lost 26%. Similar move to more more or less to silver. And this is this is again this is weather related. This is due to the to the massive storm. So >> So we can't blame that we can't blame that one on the Fed share. >> Well, we can try but okay. Um we'll speak about him in a second. But >> what's interesting about it is, and this is why I think we've been advocating uh for so many years now on the podcast, and that is >> people need to realize how important it is to have the commodities as part of their trend following universe because this is where we sometimes see lots and lots of opportunities. Um and often, uh at times where there might be a riskoff event, we didn't see any riskoff event here in equities. they didn't really bother too much about what happened. Um, but they're just wonderful in terms of opportunities. Not always, for sure. Um, but at times, and this is why we like them. But speaking of what you said could be the reason, I don't know that it really was, frankly. Again, I hear this story in my back of my mind about a certain investment [snorts] bank that uh was um, you know, behind a lot of this. Um but yes, at the same time, we were told uh that we will have a new uh if if confirmed, we will have a new um Fed chair with the name of Kevin Walsh, >> which was a surprise to me because anyone who listened to the Christmas specials will know that my outrageous prediction for this year was that the new Fed chair would be Baron Trump. So, um fortunately, I didn't >> You're already out of the uh predictions. Oh my god. Yeah. Tough tough start to the year actually. Anyways, um tell me what you think about uh this. Not that our opinion is particularly uh insightful, I'm sure, but what what are your thoughts? >> Um I must say, so you know, I think I kind of wanted to sort of fess up that I'm not a fan of the current president of the United States, and this will come as a surprise to many people. Um and I I was of the opinion when he came into office a year ago that it would be very bad for markets. Um and mostly mostly I've been I admit I've been wrong. Um there was obviously the tariff tantrum um and which you know did something very silly and non-sistatic and solve all of my equities. So that was that was that's cost me a fair bit of relative performance sadly. Um but um generally speaking, markets have kind of shrugged off, you know, to an extent. And maybe that's because of, you know, Trump always chickens out the taco trade. Um maybe maybe no one really takes him seriously. Who knows? Um but um the I think there was an expectation that he was going to appoint someone to the Fed who would be basically in line with his wishes. So um now and there's sort of a view that um if you look at the spectrum of people who he could have appointed um then that this guy is on the kind of more sensible and sane end of that spectrum. I mean he does he does have some opinions that are less mainstream around things like QE and the bed balance sheet and so on and so forth. But in terms of just interest rate setting actually in you know some some people have said well actually this guy is quite hawkish um if you look at his you know look at some of his um decision- making over the years. Now a key question is you know given that Trump has installed or you know let's assume he gets confirmed and and I think he will do to be honest because I think the this you know the the Senate and the House will be just so relieved. they'll be like, "Yeah, okay. This, you know, this this guy's probably better than any other option we could get." So, they I think he will be confirmed. Um, and this will all happen before the midterms anyway. So, I think numerically the numbers are there. Um the question is once he's actually there given given that he's been appointed um you know by someone who demands loyalty whether if he then will he then basically do as he's expected to do which is just to kind of cut rates and keep the former real estate developer happy. Um or will he not do that and then have pressure put upon him like the current Fed chair has pressure put upon him? Um and if and then will he then crumble and say okay fine I'll do what you say or will he like the current Fed chairman has fight back and say no you know you've appointed me now I'm basically in you know independent I'm going to I'm going to do what I think is best. Um so yeah uh it's an interesting um it's interesting and I think the the true answer is no one really knows and this is true of most things related to the current president. No one really knows what's going to happen uh in May. I think it's May when Pal's terms up. No one really knows what's going to happen. So, um I think the the the reaction was if that you know I think that was quite a big reaction to be honest if if that really is approximate cause. Um and you know if you actually look at say the bond market where you could which you could argue is the most direct measure of this it was a roughly muted mood in the bond market. Um, so I haven't got the figures to hand, but I don't remember anyone's, you know, I don't remember the, you know, the 2-year bond kind of moving by sort of 50 basis points on the day, which is what you would have expected if this genuinely was like a a novel and interesting piece of information. But, uh, so I think everyone's relieved, got a bit carried away with their relief perhaps, and but as to whether this is good news or bad news for the bond market, for the economy generally, we'll have to wait and see. I I think you kind of touched on hit it on the nail here because or hit the nail on the head here because the reaction in the financial markets were very muted. So this thing about him being the cause of a silver selloff, I think it had much more to do with positioning and how people had been piling into this and they talk about the Chinese retail investor and all that stuff. I really don't think it had anything um to do with him really. But of course we always uh hear the in the news that there is some kind of cause you can point to. I think you're um I mean for people who don't know but um as far as I can tell uh his father-in-law is a massive financial backer of Trump. So obviously it may not be that surprising that he got the job. Having said that, let's not forget it was Trump who appointed Powell. So you never know as you say once they are in and they can't be >> something which he appears to have forgotten as well. >> Right. and and you know once they are in they really should concern themselves about their own legacy and not so much the legacy of others and then do the best possible job. So we don't know and then also let's not forget he is one of is it 12 votes uh in total I mean it's not like he can decide what interest rates are going to do. He has to uh compel or convince uh or force uh others to go in his direction. Now, there might be one or two that he's very easy to to convince. Um, Stean Moran, Steve Moran, and and and maybe another one, but >> but we we're still waiting for this Supreme Court decision as to whether Trump can effectively then pack the rest of the board with whoever he wants. Fire and high will. Yeah. >> Yeah. Yeah. All things up in the air, but but what you could say in general is of course that um because you mentioned this thing that Yeah. it a year ago you weren't too uh bullish about equities but I mean at the end of the day however high this market goes it could still all end you know pretty pear-shaped uh because of all these things that are bu building up to it it's always the timing of these things that impossible to to know but we'll see and I think as again as as we mentioned many times in in in what's going on with what's going on in the world right now being a systematic diversified uh investor um I think is so much better than trying to make any sense of this from a discretionary point of view. Uh even though those who get it right will probably get it really right, but you can really get it wrong as well. So >> yeah, well discretionary Rob sold all of his equities in April last year and systematic Rob is up since then is up, >> you know, 20 25%. So >> I think you should listen more to the systematic Rob. >> Clearly clearly. Anyways, talking about quote unquote the systematic Rob um before we move up to the trend following part and I know maybe we'll have time I'm not so sure to go back to this uh point about AI but what did hit my radar just uh this morning related to AI was just this headline I think in in Financial Times that Google said that it's planning to double its capex this year to as much as $185 billion to really bet huge on AI. Now, I know they just announced results and they made profits of something like 130 32 billion. So, I mean, they have a lot of money to spend, but still doubling the AI spend to almost $200 billion. I mean, this does sound like a bubble to me, but who knows? >> Yeah, we're both old enough to remember 99200, right? So, >> yes, we are, unfortunately. Anyways, let's uh talk about some trend following updates uh before we dive into a really, really interesting article. People should really look forward to this. Anyways, uh not surprising. My trend barometer is having a great time. It finished yesterday uh which was um Wednesday at 64. That's a super strong reading. And of course uh what it really tells you is that there is a lot of breath because the percentage or 64 means that there are 64 of the market that in the portfolio it tracks that are quote unquote trending. So it's it's about breath uh in the environment right now. So that's the important uh part and if we look at uh I mean we already alluded to the fact that January was a strong month for for for managers. Um February is off to a really good start as well. uh mainly this time driven by as far as I can tell uh the financial sectors like equities, a bit of currency, some of the fixed income markets even are supporting the um uh uh P&L this month and and also uh things like natural gas, a little bit of meats, some of the grains. So the breadth of the trend environment as far as I can tell um is is pretty good. Um, so before I um I dive into the numbers, uh, Rob, anything you want to add to kind of, um, where we stand and and and your from your vantage point um, overall? >> Yeah. Yeah. So, um, I actually don't think I've got a position in natural gas sadly. Um, >> well, the ball is huge. So, size would be a bit too big for my tiny account. So, yeah. And the one thing I would add, and I know not everyone trades it, but um Bitcoin obviously I think the market that's on everyone's lips this morning, at least on X Twitter, is is Bitcoin. >> Oh, really? >> Um and I think most CTAs would trading it would probably I mean I trade it just for the futures, just to say. >> Um most CTAs who trade it would probably be short. Um and that that's it's actually my biggest short position at the moment. So >> um that that will um that might be kind of doing some interesting things for some people as well I would say. Uh but definitely yeah know I I'm also um my my risk is a little bit higher than it has been for a while. So that kind of matches up with your trend barometer. Definitely. >> Yeah. Okay. Um cool. All right. Good stuff. Um let's run uh through the numbers quickly and then we'll dive into um the first article. Um so these are numbers are uh as of Tuesday because the Wednesday numbers have not been published at the time of we're recording today. Um but as I mentioned uh a positive start beat up 50 up 33 basis points up 5.38% so far this year. So very strong. Um sen index up 27 basis points up 5% so far this year. Uh sen trend up another half a percent in February up 5 and a quarter for the year. And uh the short-term traders index also up 13 basis points and up 2.37% so far this year. Uh the Msei world index on the other hand is down half a percent so far in February as of last night up 1.76 for the month. And the US aggregate bond index uh so far pretty flat in February and was pretty flat in January. So that means we're pretty flat for the year. And the S&P 500 uh down 81 basis points so far in February and [music] up 63 basis points so far in 2026. >> [music] >> Now, before we get to the [music] article, we actually got two questions in from uh Pedro who wrote in, and let me just run through those with you before we dive into it. Uh the first question, and this is a long one, so I'll probably probably butcher it a little bit. I'll do my best. Estimating realized book correlations from P&L. Rather than estimating pair-wise asset correlations, I've been thinking about measuring the daily correlation of my entire book's return using something like um a uh exponentially um way weight weight moving average. The logic is that this uh already factors in both asset correlations and signal correlations as expressed in realized P&L. I understand this adds noise, but the alternative assuming a high fixed high correlation like 7 doesn't match reality for my book. My average realized correlation is much lower. So assuming 0.7 means I consistently undersshoot my vault target. Is there a sensible middle ground here or a better way to calibrate this? interesting point because um also uh if you listen to which I'm sure you do Rob uh last week's conversation with Katie, we did actually talk about uh not so much correlations but certainly also V and how that really can have a huge impact um as we just talked about with silver in terms of uh your overall position size and and so on and so forth. >> Yeah. Um, I'm a little bit unclear about this question. So, I'm going to make some assumptions and and apologies Pedro if my assumptions are wrong and I'm answering the wrong question. But what I think you're talking about is imagine you're trading a bunch of instruments. So, you got a bunch of trend following systems, one for silver, one for gold, one for, you know, S&P 500 or whatever. And then what you need to do is essentially scale um kind of work out what the sort of total leverage is required on that to get to a given risk target. Now if all assuming that you're doing volatility targeting and everything I mean it's got the same volatility that means the only thing that's going to change your your risk outcome is the correlation of those things. So if they're per so if they're perfectly correlated, if you're targeting say a 15% risk target on each of them, you'll end up with a 15% risk target on the entire book because it's perfectly correlated. Now if your if your correlation is less than one, which is hopefully the case because we love diversification in in trend following space. We absolutely love it. We like we like nice low correlations. The lower that correlation is, the more likely it is that um you'll you'll undershoot your risk target. So instead of getting to 15% on your whole book, you'll only have 10 or perhaps five. And actually um the the kind of when I actually do this exercise and work out how much I need to increase my position size to compensate for the diversification effect across instrument, I get numbers of between four and five. So a 15% risk risk target would turn into just 3% risk if I didn't do this this scaling up. So the question Pedro is asking I think is how do we actually kind of calculate this number? Um now he's saying use a fixed correlation of 7. My gut feeling is that he's got that 7 from one of my books and it's a slight misunderstanding on his part because 7 is actually um roughly speaking the ratio between the correlation between the underlying instrument returns and the correlation between what happens if you trend follow the those instruments because trend following them redu because you're trend following them you're not going to get you're actually going to get a lower correlation on than if you were just holding them long any positions on each. So that's where I think the seven comes from. So in answer to the question, how should you actually estimate this correlation? Well, yes, it's a perfectly reasonable thing to do to look at the the P&L stream that comes from silver, the P&L stream comes from S&P 500 and then measure the correlation of those. And Pedro's exactly right that will basically be factoring in both um asset correlations and signal correlations. And yeah, you can use you can get can just use a rolling estimation window of about 6 months works quite well. or you know if you want a smoother correlation estimate you can use an exponential waiting of correlation that that's also fine. Um so so yeah I think the answer is yes. Um I'm not sure what he means by adds noise. Um maybe he means referring to the fact that if if you used a fixed because of the correlation estimates always changing that will result in some extra trading but to be honest that is as long as your correlation estimate is quite slow like 6 months is pretty good then that's going to be effectively irrelevant because your actual underlying trading process would be giving you a much much bigger effect. >> Yeah I think that's a a beautiful answer. He had one more follow-up question that is predictability. He says the research suggests correlation is weaker pu uh is is weakly persistent um but that the best forecast is often just the long run average with the main predictable component being volatility regime. Do you find it worth adjusting correlation assumptions based on V regime or is a constant conservative assumption sufficient? Um so th this is a little bit again I think there's a bit of confusion in the maybe in the way the question is phrased from PE's understanding. Now most of the academic research on on estimation is around the estimation of covariance matrices not correlation matrices. Um now a coariance matrix is is basically what you get if you combine a volatility and a correlation matrix together. Um because in most kind of financial optimization you it's the co coariance matrix that you use if you're doing you know your marovitz you invert your coariance matrix. Um, now what I found and and what the academic research finds as well is that the predictability of volatility and correlation is different >> and it's better to estimate them separately and then combine them together if that's what you want to do. So volatility for example is much more predictable than correlation and I think that's kind of what Pedro is saying and that's true that's correct. >> Um, but the other thing is that but correlation is still pretty predictable. Um, so if if you want some hard numbers, so off the top of my head, if you do a regression of next month's volatility on last month's volatility, you get an R squar of about.35. If you don't know what R squ is, don't worry, but 35 is pretty good. Okay, that's pretty good. Um, and for correlation, you get something more like 0.2 or 0.25, which isn't quite as good. But if you were to try and do to try and predict you know means which is the other moment of the distribution you would you'd get nothing. You just get noise. So that it's it's these things are relatively predictable. Um now the other thing about correlation is the as I said you probably want to use something like about a six month rolling estimate versus about a one month rolling estimate for V. So the the kind of correlation structure changes more slowly than the volatility structure changes. Um so what I personally do and what I recommend if is is to is to yes is to actually estimate these things separately. Um if you are using the the correlation and the volatilities for this scaling thing that we've talked about already well then the things you're actually playing with are should already be at the same target. So you don't actually need to estimate a volatility estimate. Um obviously you will still need your volatility estimate for position sizing as we've already discussed. Um but for actual the sort of riskmanagement kind of portfolio construction part the nice thing about doing things in a voltargeted way is you you don't need to estimate vols again you just assume that your volt targeting works which as a first order approximation is fine then you just focus on your correlations and then you can focus on estimating your correlations in the best possible way. So it's quite a technical it's quite a technical answer but it's quite a technical question and I think if you understand the question you'll understand the answer if if you don't then don't worry too much about it. >> Yeah. No perfect. Absolutely. Thank you for uh for doing that. Okay. Well as um we just talked about predictability and there is another thing that's very predictable and that is uh when we do review papers uh which we do very often on the podcast it's very predictable to say that it's most likely uh from either man or AQR. And yet again, here we are. Now, in the interest of time, I don't know if we will have time for both. Um, because I think the first one we're going to be doing is the man paper. It's a super good paper. Um, I'm going to try and remember to uh put a link to it. Uh, but this is a paper about not really about better models or faster signals. It's really about what markets managers to, you know, kind of choose to trade and therefore what kind of trend following uh they're running. But I'm going to let you uh take us into the paper and I'm going to follow along and see if I have any any thoughts along the way, but um I'm going to turn it over to you. >> No, it's a very good paper and it's actually almost maybe three papers in one to be honest. So there's an awful lot of content here. Um, now the it's been written by three people. And I have to say it's now been so long since I've been away from AHL that I don't know who know any of these people are. Um, I've never met any of them. But so that that also means I shouldn't be biased in loving this paper because it's not like it's been written by an old friend of mine who's going to buy me a beer in exchange for mentioning it. So be reassured. Um, so the I guess the the main thing that they sort of do in this paper is say well actually there are there are kind of different kinds of markets, right? And we've already discussed, you know, the fact that different markets have different levels of liquidity, for example. Um, and they they what they say is, well, we're going to divide um our market universe into three categories. Traditional markets, alternative markets, and alternative esoteric markets. Um, and I I guess although a lot of that is liquidity, um, so you know, then they've got this really nice graph. It's figure five if if you're following along at home. Um, which shows liquidity and and um on one axis and complexity on the other the other axis. So esoteric markets maybe less liquid but not necessarily. So one of their esoteric markets they've labeled as China. Mhm. >> Um I think they might mean futures markets in in China rather than the whole country. Uh another one is physical cryptocurrencies, many of which are relatively liquid of course. Um but those are quite high on the complexity axis. Um whereas on traditional markets they put cotton futures um which are they put low complexity because I that you know here I guess complexity means more like a future um but that they have actually on their graph those are less liquid than China and physical cryptocurrencies. So it's quite it's a bit richer than the the normal categorization um around liquidity or around you know are these futures not futures. They look at both. Um but then what they they do is say well the interesting thing about um these these different markets is kind of where the money comes from. Um now finance people love doing something called a factor decomposition. Basically what a factor decomposition does is say well given um you know um that I'm making some money from this portfolio can I identify um sort of where the key sources of those returns are from um and in trend following um one thing you can do is is to say well it looks like you know um if you do something like a PC principal analysis it looks like a lot of the return in my trend falling portfolio is actually coming from a kind of risk on riskoff factor which I'm trend following. So um you know if you look at 2008 for example um CTA's made a lot of money bonds went up equities went down that was a perfect example of that of that factor in action. Um and then you basically then what you can do is say well what's left over what I get left over that isn't explained by that factor return and that's they call that the idio syncratic return. Um so you could argue that that you know the returns we saw in January where as we've said bonds have gone nowhere equity slightly up you know finan February's looking a bit different maybe but financial markets generally didn't you know that kind of first principal component didn't contribute much well so maybe it was more something idiosyncratic around gold and silver you know or or something else or natural gas um so what they they they do is say well if we look at this this idea of decomposing into you The main things explaining a portfolio return and the idiosyncratic return. The key thing that's interesting about these alternative and esoteric markets is that a much bigger proportion of bare returns are idiosyncratic. They're not coming from these these these big factors. Um and in hard numbers um you know so roughly speaking in traditional CTAs about half the returns are from factors and about half of idiosyncratic actually it's probably more like 45 55 45% idiosyncratic 55 or so factors. Um but in these unusual markets those these alternative and esoteric markets it's more like 1/3 2/3. So only about one/ird is coming from the the main factors and 2/3 is idiosyncratic. So the main thing about alternative markets is they give you an exposure to weird stuff to weird sources of return. Now if you're doing any kind of portfolio construction, um you like weird stuff, okay? You like diversifications, we've discussed, you like idiosyncratic returns. If the only returns in your portfolio were coming from trend following a kind of equity risk on risk off return um you you know that that's going to be not very helpful right to you you want to be able to be trend following other things as well getting itic returns from lots of different places. So what that means is is in a completely unconstrained portfolio if you do a standard portfolio optimization you're going to want to have quite a lot of these alternative and esoteric markets and of course this is uh one of the common themes of I think ever since I've been talking to you Neil's alternative markets which obviously is a you know has been a big thing for AHL um but also for for people like Floren Court for example was a rest and settle CTA that focus just on alternative markets and most big CTAs have have gone into alternative markets to a greater or lesser degree. Um so this is kind of something that that that's not a surprise um that that we already know about. Um and if they do this this kind of um sort of standard portfolio optimization, they find that um only about 30% of their portfolio needs to should be going into these traditional markets. Um which isn't a lot, right? And just to emphasize again, this is unconstrained. So this this assumes you've got no issues with liquidity and you can put as much into the weird stuff as you as you possibly want, which a big CTF of course can't do, right? Um because the they're they're just too big. And then they have um the rest of that portfolio, the 70% is in is in a mixture of alternative and you know alternative esoteric and and there's a bit of a breakdown and I can see that they put 5% into crypto for example. So the the crypto balls would would love that I'm sure. Now one question that again we we keep returning to is why should we invest in CTOs? Why should we invest in trend and following? And it it and it depend I think I've mentioned this before. It depends on whether you are looking at it as a standalone product or as something that goes alongside an existing 6040 or some other traditional kind of assets. All right. Now, if you're doing as a standalone product, you want, you know, to the first order approximation, you want a maximum sharp ratio. You'll do this kind of standard portfolio optimization. And so, as a standalone product, as a CTA without constraints around liquidity, what you would do is offer something that was 30% traditional, 70% alternatives, and that would be the best product for the investor to buy. But most people, of course, don't aren't doing that. Most people uh are putting a slice of trend following into their 60/40 portfolio. We'd like it to be a bigger slice of course, but you know that's life. Um and that means what they're looking for is an element of tail protection. Um and it's a pity we don't have Katie with us because of course she she wrote the book. Um and but um they're looking for for some kind of you know crisis alpha tower protection whatever you want to call it. So what um the these guys did was say, well, what if we um instead of looking at uh a portfolio optimization that basically just treats all states of the world the same, let's focus in on periods when equity markets fell because what we want to do is find the best portfolio for that scenario because that's when people really want to benefit from from what CTA performance is like. So they constructed their their their correlation matrix um differently um to to account for that and what they found um was that the traditional portfolio the traditional trend for things you trend for like you know >> so the liquid markets let's call >> the liquid well remembered Neil it's not just about liquidity it's about it's about so basically it's it's kind of it's futures traditional futures markets that a CTA would have had in it 30 years ago before people started doing all this weird stuff like trading German power and CDS and all this kind of stuff. Right. Exactly. So, it's things that are >> relatively liquid but also basically futures and therefore for a CT relatively simple. But more importantly, it's things that have a that whose returns are mostly being driven by this factor risk rather than this idiosyncratic risk. That's the key point about this article. That's the key innovation in how they categorize markets, I think. Um now it turns out that that actually what you want to do in the crisis situation is not have 30% traditional markets. You want to have 50% traditional markets and then obviously the the nontraditional stuff gets squeezed down to a smaller fraction. And the intuition behind that and which they explain is because it's that that total facto return that thing that's that you know that that thing that explains most of the risk of traditional markets that makes traditional markets look bad from a kind of generic optimization perspective because they haven't got much idiosyncratic risk. It turns out in a crisis that's what you want. Okay, that's what you want. And this kind of makes intuitive sense because in a crisis um you basically want to just make fairly big bets on going 2008 going short equities long bonds. You're not really going to need your your fancy um you know other weird markets there to to sort of help you. Um and of course there is a risk that many of those markets um might have issues around liquidity which is another thing that they they talk about next or around for example whether you can go short um because a lot of these things in a crisis you'd want to go short because a lot of them trade light risk on assets um because because um you know that things like well we talked about cryptocurrency you know spot cryptocurrency for example does that um so that that's kind of the the interesting thing now. Um they then take this a step further and say well all this time um you know um we've been kind of assuming that we're completely unconstrained. Um but actually another fantastic thing about about CTAs and or about anything that trades futures um is the fact that they're very cash efficient. So if I look at my own futures account um which has a a kind of a relatively high risk target for by institutional standards by you know not all some people are a bit crazy of course as we discussed but you know my my I'm probably running at about one and a half times the volume of say AHL is on most of their funds. Um I'm only using 30% of my cash's margin. So you know a typical CTA that number is 20%. And that's super cash efficient. Um and that that makes them very attractive from a structuring perspective. Um so the next thing they they did was say well let's say we want to um go a bit further and say well actually we want to kind of maintain this cash efficiency. Now your the non-futures markets obviously aren't as cash efficient. Um most of them have margin requirements that are higher than f well all of them have margin requirements that are higher than futures. Some of them even have 100% margin in the sense that you just you can't margin them. You have to just put put up cash essentially. Um and if they do that again it's not a surprise but but then what happens is is we we get even more into traditional markets. So we end up with 70% in traditional markets um and uh just just 30% um in the the alternative markets. Um and one final point to make is that when is cash efficiency most important is when we need cash. >> Yeah. >> When do we need cash? When the world's ending. So you know in 2008 um the fact that CTAs had plenty of cash, daily liquidity, no gating meant that many people use them as cash machines just you know. So we had this was a weird situation where although although you know CDS industry made a lot of money at the same time people were withdrawing it because it was a source of of ready cash that wasn't available elsewhere. Um, so that that that's yeah, so that's the that's the article. I mean, so there there's a lot of really interesting things there and like many like you know my many AHL articles, it's very well written, not technical, lots of nice graphs, very well presented and yeah, for me at least some really interesting innovations around a few different topics. >> Yeah. No, I couldn't agree more. People should definitely go and download it and and um and read it and as I said, I'll try to remember put in the link in the in the show notes. But I just want to kind of summarize uh all of this just to make sure people really get the the importance of this. Um and that is if you were just kind of trying to maximize and looking at your trend portfolio uh in isolation and you say oh what could be what you know what kind of market portfolio should I go for when I'm looking to maximize my my long-term sharp of of that um in all periods? Well, you would go for a portfolio that probably trades more markets. That's kind of what they they uh um conclude. However, as Rob rightly say, a lot of investors and certainly I would say most investors that I come across, they include trend following and CTAs specifically because they want to have something that really does well for them when equity markets are not doing well or when bond markets are not doing well. And in their work they find that then you need to find managers who are more focused on the traditional markets the more liquid markets the the classical futures markets we've been trading for the last 30 40 years um and go with one manager that that does that and then on top of that and I think I completely agree this is the interesting point they make um and that is if you then also are uh an investor where you might need to get some cash out in in terms of a crisis If your private equity fund is not giving you any money or drawing down or whatever comes during a crisis, then you need to be even more focused on managers that trades the the classical liquid futures markets. Um because the ca the the uh cash efficiency is important. Um and um and so I I thought it was really well done as well and um and to make that point um the way they did so that everybody kind of understand the difference between the choices we all have to make as managers. So it's not really all about kind of returns. It's also about you know um what's the best manager fit for me uh in terms of what my objective is with investing with the CTA. uh that should lead you down the path of identifying a smaller number of managers that might be relevant for for for for you. So really good uh paper kudos to the team uh for for writing that the next paper which we won't have time to do justice. So we'll we'll wait with that. Um but it's also a a super interesting uh paper. Um we'll see if we wait until Rob comes back or maybe we'll do it beforehand. But um it's something that is also super relevant uh in the current environment and what with what's going on in some of the um in some of the markets. I don't know if you want to spend 5 minutes, Rob, it's your uh it's your call >> uh on the Financial Times article you sent me uh this morning. I have not read it myself, so I don't know um what's in it, but but please do if you want to spend five, seven minutes on that. >> Yeah. So the the the article's on the FT alpha valal alphavville part the FT which is important because that's actually be in front of the pay wall so anyone can read it. Um and it's called the quant shop AI lab convergence. Um, and uh, it's it's kind of interesting because basically what they do in the article is they draw parallels between kind of finance quant and AI quant for one of a better word. Um, and their sort of thesis is that that basically these people are becoming more and more like each other. Now obviously there's always been a um a crossover between um the the sort of worlds of finance quant and you know software engineering more more generally and kind of um you know so and and and obviously a lot of places are using well let's let's be careful. So in finance, in systematic finance, we've always used something that I suppose you could wave your hands and call machine learning. Now machine learning to some people has to be very complicated, but um for example, if you're running something like a simple regression, which is a very simple a relatively simple way of of working out, for example, as we've just done, you know, how how much money you should give to a particular market. And you could use a regression to do that or you could use an optimization. Those are forms of machine learning. Um they're forms of what you might call supervised machine learning because the we're sort of telling the computer what to do and then it's giving us the numbers that come out. Um so that that's you know all the systematic finance is machine learning although some some you know some people say oh no that can't be machine learning because it's it's too simple but but you know then they can never tell me where the boundary is. So I say well no it must been machine learning. Um now then then you have something which you might call unsupervised machine learning and that kind of then blends into our AI and LLMs and neural networks and all this kind of stuff. Um and um the the sort of I think there's um certainly a lot of people trying to use AI uh and that goes for every industry of course not ours. Um I think pretty much every every fund um of a decent size will have some kind of AI thing going on. Um and uh how much of that there is is an open question. How much of it is just a marketing thing is just is an open question. How much of it kind of blue sky research like a you know let's buy a call option on this technology and you never know we might find something that no one else can do um is an an open question. I think the people who are doing it best are ones who are using it in very tightly defined areas. Um so things like for example execution research I think that's where it's potentially more interesting because that's where something like um well first of all you got a lot of data. Um so you you know we're building our kind of normal models with 50 odd years of daily data. Um and you know in execution research you've got tick data which is you know obviously much much richer much faster there's much more of it so it's much more suited to something like building any kind of nonlinear analysis which would include something like training an LLM on that data. Um, all that aside though, um, the article is quite interesting because rather than just making the obvious point that, you know, well, everything's AI now, you know, which is a slightly lazy argument. And I and I I do think they they do, and this is probably because they're talking their own book, they do slightly overstate, I think, the amount of of AI that that that sort of is going on. Um but they do kind of go further and say well if you actually look at the kind of workflow uh in quant finance and the workflow in our that's similar um you know if you look at the the sorts of people that are being hired um if you look at the technology stack um if you and even they even say well if you look at the the the way that kind of people are employed so you know we're used in we're used in finance to sort of non-competes and gardening leave and and things like that and and signing on bonuses. Well, they they've got that ini in AI now. And you know, you hear stories of people being offered kind of nine figure sums to move from from meta to from move to move from open AI to meta. So, so actually, you know, it many of the top AI researchers are making far more money than than any p quant finances, sadly, I have to say. Um, so so yeah, it's it's quite quite an interesting quite an interesting article. I am a skeptic of this stuff. I've said it before, I've said it again, but I still found it an interesting article. So, it must have been good. [laughter] >> Fair enough. Fair enough. Anyways, um we have come up to uh a little over the hour. So, we're going to wrap up uh our conversation today, Rob. This was uh fantastic. Thanks so much for spending all the time uh looking into this and and for answering the questions for Pedro. Uh I hope everybody found it useful. Um because I think actually our conversation today touches on on so many uh important points when it comes to CTAs and trend following and investing in general and all of that stuff. So if you want to show your appreciation for Rob and and all the other co-host uh please go to your podcast platform of choice um and leave a rating and review. Uh it really does help and uh we really appreciate uh the support we uh can get. Um if you have a question for next week where I will have two guests uh Andrew Beer will come back and he will be joined this time by Tom Roble from Sockgen. Um so you can send uh your questions uh as usual info@ toptraders onblog.com. I'll do my best to to uh get them in front of Andrew and Tom u when we speak. Um but for now for today uh 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 usual, take care of yourself uh and take care of each other. [music] >> 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 iTunes and subscribe [music] to 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. [music] And to ensure our show continues to grow, please leave us an honest rating and review in iTunes. It only takes a minute and it's the best way to show us you love the podcast. [music] We'll see you next time on Top Traders Unplugged.
Position Sizing When Markets Break feat. Rob Carver | Systematic Investor | Ep.386
Summary
Transcript
Imagine [music] spending an hour with the world's greatest traders. Imagine learning from their experiences, their successes, and their [music] failures. Imagine no more. Welcome to Top Traders Unplugged, [music] the place where you can learn from the best hedge fund managers in the world, so you can take your manager due diligence or investment career to the next level. Before we begin today's conversation, [music] remember to keep two things in mind. All the discussion we will have about investment performance is about the past and past performance does not guarantee or even infer [music] anything about future performance. Also understand that there's a significant risk of financial loss with all [music] investment strategies and you need to request and understand the specific risks from the investment manager about their product before you make investment decisions. Here's your host, veteran [music] hedge fund manager Neil's Krup Larson. Welcome or welcome back to this week's edition of the systematic investor series with Rob Carver and I Neils Castro Llassen where each week we take the pulse of the global market through the lens of a rulesbased investor. Rob, it is uh wonderful to be back with you uh this week. First time in 2026. Hope you're doing well. How are things in the uh in the UK? >> Yeah, it's doesn't seem I mean 2026 is only a few weeks old, but it seems like a hell of a lot's happened. So, particularly in the market. So uh so yeah, it's all quite exciting. >> It is all all very exciting. And actually speaking of exciting, we have a great lineup of topics uh today um which I think people will really enjoy. U we're going to be tackling a couple of new papers, an article and uh a couple of questions that came in. So uh so this is all uh super exciting. Rob, as you know, I always I'm always curious to uh hear what's kind of been on your radar since we last spoke. um but not from the topics we're going to be talking about. But if there's something else that you found interesting um then uh >> let us know. >> Yeah. Well, I was I was hoping by now I would have seen the new Melania film, but um unfortunately I've not had that opportunity and it looks actually looking at the viewing figures worldwide, not many other people have as well. So, so hopefully I'll get a chance to review that next time I'm on. Um so I've saw an interesting um paper published by Alliance Bernstein. I don't want to talk about it today, but what made me um sort of impressed was the title of this paper. Um and I actually for reasons that were becoming obvious, I actually had to ask my son who did German at school how to pronounce this word in this title. So the title of the paper is about TPA total portfolio approach which we've discussed before. title of paper is portfolio design as um Gazamkensk the total portfolio approach and apparently a gazam is is to is something like a total piece of art and it's apparently it's a term coined by Vagnner in relation to his operas. So, uh, so yeah, that that intrigued me. But it it goes to show if you're working as a salesside analyst, you you desperately got to try and make your stuff interesting, right? And one way of doing that is coming up with a title that that makes people stop and think and go, "Well, that sounds weird at the very least." And it worked because I to be honest, I probably wouldn't even have glanced at that paper were it not for the the funky title. So, uh, so yeah, if you're interested in TPA and Vagner, then uh find that Alliance Bernstein paper and enjoy it. >> That is quite funny. I mean, since I have to come up with a lot of titles and headlines every week for the podcast episodes and the emails we send out, uh I'm always interested in in a catchy uh method of uh getting people to uh you know, listen or open what we uh what we share. I have to say I didn't expect a German word to do the trick, but I might have I might have to try it out. >> Well, there we go, Neils. There we go. All those long compound German words just waiting to be used in your your email subject lines. >> Clearly. Clearly. Okay. Well, let me tell you what caught my interest. And I know this first topic definitely caught your interest as well. And instead of putting it in the trend following section, I think we'll we'll discuss it now. Um maybe spend a little bit more time on it. And it is what happened uh in the last uh week or so. Well, actually the last uh few months, but it kind of uh really came to a head uh on the last day of January. >> Freaky Friday. I'm I've decided I'm going to christen it freaky Friday. >> Freaky Friday, especially if you're trading precious metals and especially if you're trading silver. And actually, I would say this week uh has been pretty uh interesting as well. So why don't you tell us a little bit about your silver experience and um maybe we can then talk a little bit about kind of the the good old days and and why silver is not gold uh in in some way. >> Yeah, I mean so um I obviously trade silver um it's a a futures contract. It's liquid so it would be on the table for me to trade. Now, I've talked about this before, but the way I trade um futures is a little bit complicated, and it's mainly to do with the fact I have a relatively small portfolio. So, I have a notional exposure to 250 futures contracts. Um many of those I couldn't trade anyway because of regulatory restrictions. Of course, they're not liquid or they're too expensive. Silver does not fall into that category. Um, and then I basically dynamically optimized my portfolio every day to match to get the best possible match I could do to a sort of abstract portfolio, which I could have if I had a notional value running into the hundreds of millions of dollars, which sadly I don't have. So that will mean that it often that I won't necessarily have a position in a given instrument even if I have quite a strong forecast on it. Um, now silver though I do have a position and I did have a position in I should say. Um and if I look at my history of exposure to that position, uh it is quite interesting. Um now again it's slightly complicated by the fact that that I have this dynamic optimization but broadly speaking you can sort of say that CTAs generally will have positions that are a function of two things. The strength of the trend and the volatility of the instrument you're trading. Um, now that's obviously a broad brush thing because of course some CTAs will take basically binary positions. So they'll, you know, they'll kind of go from fully short to fully long. Um, others will sort of do what I do, which is sort of continuous adjust adjusting forecast. So getting, you know, the moment the trend starts to look positive, they'll kind of increase their forecast and that would increase the position, all other things being equal. And of course they're not, which is what we'll get into in a second. Um and also of course some CTAs take volatility into account when they initially buy a position but then don't make any further adjustments to that position as volatility changes. Um so we can characterize the silver story over the last few months as a a positive bullish signal getting stronger. Um this is all up till last Thursday obviously because obviously on freaky Friday things change very dramatically. Um but at the same time the volatility also increasing. Um so my position broadly speaking and the position of anyone trading like me would be a function of those kind of two effects. If I could actually look at my position of of my exposure to silver I can see that um although I was trading it on and off last year um the kind of current position um I I bought some on the 27th of November and I bought some more on the 3rd of December. Um and that would have been down to the the forecast increasing um and then actually on the 31st of December I sold a contract and that would have been inevitably because the volatility had increased. So though the combination of those two effects often lead to these interesting effects um and then subsequently um on well actually not on freaky Friday itself because I have a daily rebalance. So I didn't do any trading in silver during Friday itself cuz my position was based on Thursday's close which was kind of normal if you like. Um but so on Monday, so that's a couple of days ago now. Um I actually then sold that um contract and that would have been again because volatility had increased. Um it's my trend signal is still long because um you know unless you're trading an insanely quick trend following system one day of negative returns even a day like Friday where silver dropped by the most in one day since and you know you you'll remember this Neils of course about >> it was decades at least >> decades ago many decades ago I think 40 45 46 years 47 years ago to exact. Um, so that's why I then I then sold sold out of that that silver position on that day. So that that's kind of interesting. Now a big question is did I you know useful question to ask is well did I actually make money doing this? And the answer is yes I did because although silver dropped precipitously on Friday, it only dropped back to the level where it had been a few weeks beforehand, a couple of weeks beforehand. Um so I can actually give you hard numbers. So, I bought I bought into my silver position um at levels of $53 and $59 uh per ounce, respectively. Um and then I sold at the end of December at $71. Um and I sold again at $81. So, just looking at those trades, you know, that was nice. That's a profitable trade. Um, of course, what that misses out is that in between the the end of December and the the freaky Friday, the price briefly went up to, you know, $110 $120 an ounce, I think it was. Um, so that that's kind of that's kind of my my kind of story with silver. Now, um obviously that would be different for someone who doesn't do the continuous volatility adjustment because what they would have done is seen the value of the silver in their portfolio kind of get bigger and bigger and bigger um and then not done anything about it and just hung on to that. Um and as a result would have basically would have seen a massive run up in value and then freaky Friday would have resulted in a quite substantial drop in value. Um and we've got the sort of monthly returns for January now for many CTAs and uh certain CTAs and many viewers this podcast will know who we're talking about who take very large um leveraged positions and have very high risk and do don't do this volatility adjustment. I think at one probably went from being about sort of 50% up for the month to being 25% up for the month. Now 25% up for the month is still amazing. Um, but the actual one day loss on on freaky Friday would have been, you know, very substantial indeed. >> I I think I can help you with the precision. Actually, think about being up 72% on the Thursday and ending the month up 27%. No names here of course, but it is it was it was a large move. I will say by the way there is I guess one other way that you could even though the you could say largely the trend is still up for silver in many respect. But of course if you were using a hard stop uh as one of your rules and that stop may be calculated from what has been the most recent high of some sort maybe with a V um stop uh linked to it. I guess you could have also been stopped out on Freaky Friday because the move was so big, but it really depends on your specific rule. >> It does. Yeah. And because I don't use um stop-loss rules. Yeah, you're absolutely right. Any any stop-loss I mean, I'd say that because um it depends again on how what you do with your stop-loss rules as your position runs up. So um when I if I was to run a system using stop losses which I don't as I said but I design it so that the the initial entry the stop would be set as a a a rat a multiple of the current volatility. Um now if you as the price runs up and the volatility increases if you then also increase your stop at the same time then the position you know having said that you know a 25% fall probably still would have breached any reasonable stop >> but importantly if you do do that if you do increase your stop width you also should reduce your position at the same time otherwise you're basically taking more and more risk effectively on the position. um if you don't increase your stop level as volatility rises then you you're you risk basically being shaken out of the position too early. Now on Friday that would have been fine because it would have been very nice if you'd I don't know you know how realistic would it have been to have executed at a tight stop on Friday. I don't know how how rapid the move was. Um but um it would have been you know you probably would have done better than somebody with a much looser stop. Um but you know that also meant you would have been shaken out on a much smaller move in the middle of a bigger trend which would been the wrong thing to do. So um you've got to be careful if you are using stocks how you adjust them as volatility changes just as I adjust positions as volatility changes. You you mentioned kind of you know if you had the ability to uh to uh trade on Friday um clearly it was a hugely volatile day. There's a not that many people I've I've heard from in terms of a logical explanation as to what happened. I've seen some speculations about a a large US investment bank causing all of this, but I don't want to speculate on air on that. But do do you have a sense or it's something you you monitor in some way how how liquid because I think maybe because silver and gold and platinum, they've all been going higher uh in the last uh many months and of course it's been great for trend followers. Maybe people have kind of maybe become a little bit complacent thinking there these are all the same to trade and they think oh they're all going to be as liquid as gold which is super liquid market um but but but they're not the same um in in my view and um and I so I don't know from your perspective in terms of liquidity and those kind of risks um if you have any thoughts on what happened. Um >> yeah, I mean obviously because I'm trading a relatively small amount of money, my kind of have a fairly broad liquidity filter. So I I would probably take positions in in um in instruments where a lot of big CTAs would either not bother or have a very small position indeed. >> Um so silver, you know, there's no no question of silver not falling into the liquid bucket for me. So that that wasn't an issue. But I I think the one kind of thing that does happen um in in markets that are not as liquid as you know the most liquid markets is that obviously if something runs up in price and you get a slight pullback well all the buyers are gone right everyone who wants to buy is bought. So there's no one sitting around waiting to buy off you. Um so you know people treat liquidity as a constant number as a constant number of volume of contracts they can expect to trade. But of course it isn't right and we we know that in market crisis liquidity disappears and that that happens everywhere. Uh and obviously the less liquid the market is to begin with the more likely that is to happen. Um, and I think there is a risk if you if you know, let's say something gets hot, right, and it runs up in price and lots of people start trading it, maybe you hadn't traded it before, and you look at the volume and the open interest numbers and you think, well, I can build up a pretty substantial position here because there's a lot of volume going through, there's a lot of open interest going through. Um, you build up that big position and then and then if the class, you know, if the price starts to move adversely against you, then everyone jams into it's like the the the cinema with only one fire door, right? everyone gets jammed trying to get through the exit at the same time and the liquidity you thought was there wasn't there and part of that's just that that can be driven by a few things right part of it could be driven by the fact that yeah a lot of these people are silver tourists who don't normally trade silver um and the you know what you really need to do is go back a year or two years and say well over a longer period of time what does liquidity look like um the other thing is the classic thing where if everyone's trading similar kind of strategy I mean if everyone is trend following um if everyone is doing vol adjustment um and that does you know don't forget that CTAs do vol adjustment but so do risk parity funds um and so do other kind of quant strategies um so anyone who sells on the hint of volatility spiking that the more people there are in a particular market like that then will magnify the move I mean the approximate cause of all of this supposedly was the fact that a a relatively sane person got elected to the to Fed chair next and we'll talk about that in a second I guess but as to the mechanics of why it affected silver more than gold and silver went down more in price than gold. I mean, we can speculate all day, but I think the main lesson here is is for me is risk management. Um, and the two key things about risk management are position sizing um and diversification. Um, so on Friday, I lost 3% of my total account value because I have a reasonably diversified portfolio. Um, if I was very highly concentrated in silver, that would have been a much bigger number. Um, and because I adjust my positions according to volatility as it changes, again, I was relatively protected from a large move, whereas, as we've mentioned, funds that don't do that and have much more leverage potentially are exposed to even bigger moves. And yeah, month over the month, still profitable. Great work, guys. But, you know, it scares the hell out of me, the possibility of that happening, I have to say. Yeah. Well, first of all, I mean, I think just to comment on that number, I think that that the number you mentioned was pretty among people with that kind of annual volatility that you run it at. I think 3% is is very decent and I think that's kind of what I saw on on the day from the people I can uh follow. Um, so I think the industry as a whole did really really well. Um and as you say of course the larger CTAs will be doing something similar to what you do meaning uh adjust the position along the way which means again just maybe to it reiterate for people because I remember during the uh cocoa debacle where prices also had a kind of a similar move as we saw in silver where it becomes kind of a parabolic and gold for that matter kind of a parabolic move. It's very easy to blame the quant funds uh to be behind the parabolic move, but I would hazard to to to say that most CTAs probably got into silver more than a year ago for sure. And the majority of the larger CTAs, they they're the ones who have been selling silver all the way through January, as you rightly said, maybe even through um December as volatility started to uh expand. We're just going to have to reduce our notional exposure. Um so we're actually kind of trying to to be the buffer here um and and limit these parabolic moves. if if anything I think that is important to say because most people will will think the opposite. Um, the other thing you touched on is risk management. And I just want to maybe share, and I I I I won't do this story justice in any way, shape, or form, but of course, a lot of people remember silver specifically from the story of the 1970s where the uh I think Texas oil billionaire Nelson Bunker Hunt. He became convinced that the US dollar uh would would fail. I mean, it's kind of almost like deja vu uh at the moment. Anyways, in his mind, silver was not real money. That was kind of or I I [snorts] should say silver was not just an investment for him. It was real money. So, he started to buy silver and he did it physically and he chartered, you know, several of these Boeing 707s and he flew that silver, as far as I understand, uh millions of ounces of silver to Switzerland to be stored in in vaults. Um and um and so again this for him was not about making a lot of money necessarily. It was about kind of controlling and making sure that the wealth he had built uh wouldn't disappear. Well here comes the problem. I mean he kept buying but he kept buying now for borrowed money. So the leverage of his position um became you know quite high. Um and very similar story as to what happened uh last week. Um as the volatility uh rose uh what happened in in 1979 price-wise was that silver went from $6 to about $50 an ounce. Um and that causes a problem for the exchange. So com started to change the rules on margins. We saw the same thing happening I think in January one or once or twice where margin got increased and at some point the people who have b who have made made money but they've made money from borrowed um or speculation from borrowed money to keep up with the margin calls at some point at some point they run out of of of money and so did he so um I think don't know the details but uh he ended up having you know a catastrophic uh experience experience. Um maybe he still ended up making a little bit of money but but not nearly as much as he had. And I think there was some agreement about paying back um debt. So risk management um and not letting positions get completely out of control by pyramiding and so on and so forth are all very important lessons. And even though this happened back in the 70s, as we can clearly see, they get repeated over and over again, just in different markets at different times. Um, but I think the beauty of all of this is the fact that many of the managers that we um we talk about, they've all been around for decades. We we've kind of know this. It's part of our DNA. We don't get emotionally attached to it. We manage our risk in a systematic, dispassionate way. And that's [snorts] super important. And regarding this idea that people should also remember that gold is not sorry silver is not the same as gold. I've heard stories or or information about the fact that the reason why this is important is that silver you don't mine very much silver directly. It's only like 30% of the production actually every year is specifically mined as silver. The rest comes from a byproduct of gold and copper. So it's maybe less controllable or the supply side maybe is a little bit uh harder to to um predict and therefore it can um maybe also be the reason why silver can be a much more difficult much more volatile market to trade compared to gold. But these are all very important points in my opinion. >> Absolutely. I think it's may also worth briefly mentioning I think that Friday and actually continuing to this week we're seeing a lot of dislocations across markets generally. So you know long short quant equity factors have gone to barmy. Um apparently AQR's um long short fund had both one of its worst days and also one of its best days in the last few days. So that's interesting. >> Um Bitcoin of course is selling off massively. Um there's been kind of disruptions in the particularly in the sort of stocks related to AI. So it's all kind of interesting and at least at the moment you know touchwood fingers crossed um you know my trend following portfolio seems to be coping pretty well with all of this. Um so so yeah it's an interesting time. >> Can I just mention I mean we talk about precious metals as if it was the only thing that moved. >> Yeah. >> March natural gas uh just to just to mention from the 9th of January to the 30th of January it um was up something like 60%. And then on the 2nd of February in one day it lost 26%. Similar move to more more or less to silver. And this is this is again this is weather related. This is due to the to the massive storm. So >> So we can't blame that we can't blame that one on the Fed share. >> Well, we can try but okay. Um we'll speak about him in a second. But >> what's interesting about it is, and this is why I think we've been advocating uh for so many years now on the podcast, and that is >> people need to realize how important it is to have the commodities as part of their trend following universe because this is where we sometimes see lots and lots of opportunities. Um and often, uh at times where there might be a riskoff event, we didn't see any riskoff event here in equities. they didn't really bother too much about what happened. Um, but they're just wonderful in terms of opportunities. Not always, for sure. Um, but at times, and this is why we like them. But speaking of what you said could be the reason, I don't know that it really was, frankly. Again, I hear this story in my back of my mind about a certain investment [snorts] bank that uh was um, you know, behind a lot of this. Um but yes, at the same time, we were told uh that we will have a new uh if if confirmed, we will have a new um Fed chair with the name of Kevin Walsh, >> which was a surprise to me because anyone who listened to the Christmas specials will know that my outrageous prediction for this year was that the new Fed chair would be Baron Trump. So, um fortunately, I didn't >> You're already out of the uh predictions. Oh my god. Yeah. Tough tough start to the year actually. Anyways, um tell me what you think about uh this. Not that our opinion is particularly uh insightful, I'm sure, but what what are your thoughts? >> Um I must say, so you know, I think I kind of wanted to sort of fess up that I'm not a fan of the current president of the United States, and this will come as a surprise to many people. Um and I I was of the opinion when he came into office a year ago that it would be very bad for markets. Um and mostly mostly I've been I admit I've been wrong. Um there was obviously the tariff tantrum um and which you know did something very silly and non-sistatic and solve all of my equities. So that was that was that's cost me a fair bit of relative performance sadly. Um but um generally speaking, markets have kind of shrugged off, you know, to an extent. And maybe that's because of, you know, Trump always chickens out the taco trade. Um maybe maybe no one really takes him seriously. Who knows? Um but um the I think there was an expectation that he was going to appoint someone to the Fed who would be basically in line with his wishes. So um now and there's sort of a view that um if you look at the spectrum of people who he could have appointed um then that this guy is on the kind of more sensible and sane end of that spectrum. I mean he does he does have some opinions that are less mainstream around things like QE and the bed balance sheet and so on and so forth. But in terms of just interest rate setting actually in you know some some people have said well actually this guy is quite hawkish um if you look at his you know look at some of his um decision- making over the years. Now a key question is you know given that Trump has installed or you know let's assume he gets confirmed and and I think he will do to be honest because I think the this you know the the Senate and the House will be just so relieved. they'll be like, "Yeah, okay. This, you know, this this guy's probably better than any other option we could get." So, they I think he will be confirmed. Um, and this will all happen before the midterms anyway. So, I think numerically the numbers are there. Um the question is once he's actually there given given that he's been appointed um you know by someone who demands loyalty whether if he then will he then basically do as he's expected to do which is just to kind of cut rates and keep the former real estate developer happy. Um or will he not do that and then have pressure put upon him like the current Fed chair has pressure put upon him? Um and if and then will he then crumble and say okay fine I'll do what you say or will he like the current Fed chairman has fight back and say no you know you've appointed me now I'm basically in you know independent I'm going to I'm going to do what I think is best. Um so yeah uh it's an interesting um it's interesting and I think the the true answer is no one really knows and this is true of most things related to the current president. No one really knows what's going to happen uh in May. I think it's May when Pal's terms up. No one really knows what's going to happen. So, um I think the the the reaction was if that you know I think that was quite a big reaction to be honest if if that really is approximate cause. Um and you know if you actually look at say the bond market where you could which you could argue is the most direct measure of this it was a roughly muted mood in the bond market. Um, so I haven't got the figures to hand, but I don't remember anyone's, you know, I don't remember the, you know, the 2-year bond kind of moving by sort of 50 basis points on the day, which is what you would have expected if this genuinely was like a a novel and interesting piece of information. But, uh, so I think everyone's relieved, got a bit carried away with their relief perhaps, and but as to whether this is good news or bad news for the bond market, for the economy generally, we'll have to wait and see. I I think you kind of touched on hit it on the nail here because or hit the nail on the head here because the reaction in the financial markets were very muted. So this thing about him being the cause of a silver selloff, I think it had much more to do with positioning and how people had been piling into this and they talk about the Chinese retail investor and all that stuff. I really don't think it had anything um to do with him really. But of course we always uh hear the in the news that there is some kind of cause you can point to. I think you're um I mean for people who don't know but um as far as I can tell uh his father-in-law is a massive financial backer of Trump. So obviously it may not be that surprising that he got the job. Having said that, let's not forget it was Trump who appointed Powell. So you never know as you say once they are in and they can't be >> something which he appears to have forgotten as well. >> Right. and and you know once they are in they really should concern themselves about their own legacy and not so much the legacy of others and then do the best possible job. So we don't know and then also let's not forget he is one of is it 12 votes uh in total I mean it's not like he can decide what interest rates are going to do. He has to uh compel or convince uh or force uh others to go in his direction. Now, there might be one or two that he's very easy to to convince. Um, Stean Moran, Steve Moran, and and and maybe another one, but >> but we we're still waiting for this Supreme Court decision as to whether Trump can effectively then pack the rest of the board with whoever he wants. Fire and high will. Yeah. >> Yeah. Yeah. All things up in the air, but but what you could say in general is of course that um because you mentioned this thing that Yeah. it a year ago you weren't too uh bullish about equities but I mean at the end of the day however high this market goes it could still all end you know pretty pear-shaped uh because of all these things that are bu building up to it it's always the timing of these things that impossible to to know but we'll see and I think as again as as we mentioned many times in in in what's going on with what's going on in the world right now being a systematic diversified uh investor um I think is so much better than trying to make any sense of this from a discretionary point of view. Uh even though those who get it right will probably get it really right, but you can really get it wrong as well. So >> yeah, well discretionary Rob sold all of his equities in April last year and systematic Rob is up since then is up, >> you know, 20 25%. So >> I think you should listen more to the systematic Rob. >> Clearly clearly. Anyways, talking about quote unquote the systematic Rob um before we move up to the trend following part and I know maybe we'll have time I'm not so sure to go back to this uh point about AI but what did hit my radar just uh this morning related to AI was just this headline I think in in Financial Times that Google said that it's planning to double its capex this year to as much as $185 billion to really bet huge on AI. Now, I know they just announced results and they made profits of something like 130 32 billion. So, I mean, they have a lot of money to spend, but still doubling the AI spend to almost $200 billion. I mean, this does sound like a bubble to me, but who knows? >> Yeah, we're both old enough to remember 99200, right? So, >> yes, we are, unfortunately. Anyways, let's uh talk about some trend following updates uh before we dive into a really, really interesting article. People should really look forward to this. Anyways, uh not surprising. My trend barometer is having a great time. It finished yesterday uh which was um Wednesday at 64. That's a super strong reading. And of course uh what it really tells you is that there is a lot of breath because the percentage or 64 means that there are 64 of the market that in the portfolio it tracks that are quote unquote trending. So it's it's about breath uh in the environment right now. So that's the important uh part and if we look at uh I mean we already alluded to the fact that January was a strong month for for for managers. Um February is off to a really good start as well. uh mainly this time driven by as far as I can tell uh the financial sectors like equities, a bit of currency, some of the fixed income markets even are supporting the um uh uh P&L this month and and also uh things like natural gas, a little bit of meats, some of the grains. So the breadth of the trend environment as far as I can tell um is is pretty good. Um, so before I um I dive into the numbers, uh, Rob, anything you want to add to kind of, um, where we stand and and and your from your vantage point um, overall? >> Yeah. Yeah. So, um, I actually don't think I've got a position in natural gas sadly. Um, >> well, the ball is huge. So, size would be a bit too big for my tiny account. So, yeah. And the one thing I would add, and I know not everyone trades it, but um Bitcoin obviously I think the market that's on everyone's lips this morning, at least on X Twitter, is is Bitcoin. >> Oh, really? >> Um and I think most CTAs would trading it would probably I mean I trade it just for the futures, just to say. >> Um most CTAs who trade it would probably be short. Um and that that's it's actually my biggest short position at the moment. So >> um that that will um that might be kind of doing some interesting things for some people as well I would say. Uh but definitely yeah know I I'm also um my my risk is a little bit higher than it has been for a while. So that kind of matches up with your trend barometer. Definitely. >> Yeah. Okay. Um cool. All right. Good stuff. Um let's run uh through the numbers quickly and then we'll dive into um the first article. Um so these are numbers are uh as of Tuesday because the Wednesday numbers have not been published at the time of we're recording today. Um but as I mentioned uh a positive start beat up 50 up 33 basis points up 5.38% so far this year. So very strong. Um sen index up 27 basis points up 5% so far this year. Uh sen trend up another half a percent in February up 5 and a quarter for the year. And uh the short-term traders index also up 13 basis points and up 2.37% so far this year. Uh the Msei world index on the other hand is down half a percent so far in February as of last night up 1.76 for the month. And the US aggregate bond index uh so far pretty flat in February and was pretty flat in January. So that means we're pretty flat for the year. And the S&P 500 uh down 81 basis points so far in February and [music] up 63 basis points so far in 2026. >> [music] >> Now, before we get to the [music] article, we actually got two questions in from uh Pedro who wrote in, and let me just run through those with you before we dive into it. Uh the first question, and this is a long one, so I'll probably probably butcher it a little bit. I'll do my best. Estimating realized book correlations from P&L. Rather than estimating pair-wise asset correlations, I've been thinking about measuring the daily correlation of my entire book's return using something like um a uh exponentially um way weight weight moving average. The logic is that this uh already factors in both asset correlations and signal correlations as expressed in realized P&L. I understand this adds noise, but the alternative assuming a high fixed high correlation like 7 doesn't match reality for my book. My average realized correlation is much lower. So assuming 0.7 means I consistently undersshoot my vault target. Is there a sensible middle ground here or a better way to calibrate this? interesting point because um also uh if you listen to which I'm sure you do Rob uh last week's conversation with Katie, we did actually talk about uh not so much correlations but certainly also V and how that really can have a huge impact um as we just talked about with silver in terms of uh your overall position size and and so on and so forth. >> Yeah. Um, I'm a little bit unclear about this question. So, I'm going to make some assumptions and and apologies Pedro if my assumptions are wrong and I'm answering the wrong question. But what I think you're talking about is imagine you're trading a bunch of instruments. So, you got a bunch of trend following systems, one for silver, one for gold, one for, you know, S&P 500 or whatever. And then what you need to do is essentially scale um kind of work out what the sort of total leverage is required on that to get to a given risk target. Now if all assuming that you're doing volatility targeting and everything I mean it's got the same volatility that means the only thing that's going to change your your risk outcome is the correlation of those things. So if they're per so if they're perfectly correlated, if you're targeting say a 15% risk target on each of them, you'll end up with a 15% risk target on the entire book because it's perfectly correlated. Now if your if your correlation is less than one, which is hopefully the case because we love diversification in in trend following space. We absolutely love it. We like we like nice low correlations. The lower that correlation is, the more likely it is that um you'll you'll undershoot your risk target. So instead of getting to 15% on your whole book, you'll only have 10 or perhaps five. And actually um the the kind of when I actually do this exercise and work out how much I need to increase my position size to compensate for the diversification effect across instrument, I get numbers of between four and five. So a 15% risk risk target would turn into just 3% risk if I didn't do this this scaling up. So the question Pedro is asking I think is how do we actually kind of calculate this number? Um now he's saying use a fixed correlation of 7. My gut feeling is that he's got that 7 from one of my books and it's a slight misunderstanding on his part because 7 is actually um roughly speaking the ratio between the correlation between the underlying instrument returns and the correlation between what happens if you trend follow the those instruments because trend following them redu because you're trend following them you're not going to get you're actually going to get a lower correlation on than if you were just holding them long any positions on each. So that's where I think the seven comes from. So in answer to the question, how should you actually estimate this correlation? Well, yes, it's a perfectly reasonable thing to do to look at the the P&L stream that comes from silver, the P&L stream comes from S&P 500 and then measure the correlation of those. And Pedro's exactly right that will basically be factoring in both um asset correlations and signal correlations. And yeah, you can use you can get can just use a rolling estimation window of about 6 months works quite well. or you know if you want a smoother correlation estimate you can use an exponential waiting of correlation that that's also fine. Um so so yeah I think the answer is yes. Um I'm not sure what he means by adds noise. Um maybe he means referring to the fact that if if you used a fixed because of the correlation estimates always changing that will result in some extra trading but to be honest that is as long as your correlation estimate is quite slow like 6 months is pretty good then that's going to be effectively irrelevant because your actual underlying trading process would be giving you a much much bigger effect. >> Yeah I think that's a a beautiful answer. He had one more follow-up question that is predictability. He says the research suggests correlation is weaker pu uh is is weakly persistent um but that the best forecast is often just the long run average with the main predictable component being volatility regime. Do you find it worth adjusting correlation assumptions based on V regime or is a constant conservative assumption sufficient? Um so th this is a little bit again I think there's a bit of confusion in the maybe in the way the question is phrased from PE's understanding. Now most of the academic research on on estimation is around the estimation of covariance matrices not correlation matrices. Um now a coariance matrix is is basically what you get if you combine a volatility and a correlation matrix together. Um because in most kind of financial optimization you it's the co coariance matrix that you use if you're doing you know your marovitz you invert your coariance matrix. Um, now what I found and and what the academic research finds as well is that the predictability of volatility and correlation is different >> and it's better to estimate them separately and then combine them together if that's what you want to do. So volatility for example is much more predictable than correlation and I think that's kind of what Pedro is saying and that's true that's correct. >> Um, but the other thing is that but correlation is still pretty predictable. Um, so if if you want some hard numbers, so off the top of my head, if you do a regression of next month's volatility on last month's volatility, you get an R squar of about.35. If you don't know what R squ is, don't worry, but 35 is pretty good. Okay, that's pretty good. Um, and for correlation, you get something more like 0.2 or 0.25, which isn't quite as good. But if you were to try and do to try and predict you know means which is the other moment of the distribution you would you'd get nothing. You just get noise. So that it's it's these things are relatively predictable. Um now the other thing about correlation is the as I said you probably want to use something like about a six month rolling estimate versus about a one month rolling estimate for V. So the the kind of correlation structure changes more slowly than the volatility structure changes. Um so what I personally do and what I recommend if is is to is to yes is to actually estimate these things separately. Um if you are using the the correlation and the volatilities for this scaling thing that we've talked about already well then the things you're actually playing with are should already be at the same target. So you don't actually need to estimate a volatility estimate. Um obviously you will still need your volatility estimate for position sizing as we've already discussed. Um but for actual the sort of riskmanagement kind of portfolio construction part the nice thing about doing things in a voltargeted way is you you don't need to estimate vols again you just assume that your volt targeting works which as a first order approximation is fine then you just focus on your correlations and then you can focus on estimating your correlations in the best possible way. So it's quite a technical it's quite a technical answer but it's quite a technical question and I think if you understand the question you'll understand the answer if if you don't then don't worry too much about it. >> Yeah. No perfect. Absolutely. Thank you for uh for doing that. Okay. Well as um we just talked about predictability and there is another thing that's very predictable and that is uh when we do review papers uh which we do very often on the podcast it's very predictable to say that it's most likely uh from either man or AQR. And yet again, here we are. Now, in the interest of time, I don't know if we will have time for both. Um, because I think the first one we're going to be doing is the man paper. It's a super good paper. Um, I'm going to try and remember to uh put a link to it. Uh, but this is a paper about not really about better models or faster signals. It's really about what markets managers to, you know, kind of choose to trade and therefore what kind of trend following uh they're running. But I'm going to let you uh take us into the paper and I'm going to follow along and see if I have any any thoughts along the way, but um I'm going to turn it over to you. >> No, it's a very good paper and it's actually almost maybe three papers in one to be honest. So there's an awful lot of content here. Um, now the it's been written by three people. And I have to say it's now been so long since I've been away from AHL that I don't know who know any of these people are. Um, I've never met any of them. But so that that also means I shouldn't be biased in loving this paper because it's not like it's been written by an old friend of mine who's going to buy me a beer in exchange for mentioning it. So be reassured. Um, so the I guess the the main thing that they sort of do in this paper is say well actually there are there are kind of different kinds of markets, right? And we've already discussed, you know, the fact that different markets have different levels of liquidity, for example. Um, and they they what they say is, well, we're going to divide um our market universe into three categories. Traditional markets, alternative markets, and alternative esoteric markets. Um, and I I guess although a lot of that is liquidity, um, so you know, then they've got this really nice graph. It's figure five if if you're following along at home. Um, which shows liquidity and and um on one axis and complexity on the other the other axis. So esoteric markets maybe less liquid but not necessarily. So one of their esoteric markets they've labeled as China. Mhm. >> Um I think they might mean futures markets in in China rather than the whole country. Uh another one is physical cryptocurrencies, many of which are relatively liquid of course. Um but those are quite high on the complexity axis. Um whereas on traditional markets they put cotton futures um which are they put low complexity because I that you know here I guess complexity means more like a future um but that they have actually on their graph those are less liquid than China and physical cryptocurrencies. So it's quite it's a bit richer than the the normal categorization um around liquidity or around you know are these futures not futures. They look at both. Um but then what they they do is say well the interesting thing about um these these different markets is kind of where the money comes from. Um now finance people love doing something called a factor decomposition. Basically what a factor decomposition does is say well given um you know um that I'm making some money from this portfolio can I identify um sort of where the key sources of those returns are from um and in trend following um one thing you can do is is to say well it looks like you know um if you do something like a PC principal analysis it looks like a lot of the return in my trend falling portfolio is actually coming from a kind of risk on riskoff factor which I'm trend following. So um you know if you look at 2008 for example um CTA's made a lot of money bonds went up equities went down that was a perfect example of that of that factor in action. Um and then you basically then what you can do is say well what's left over what I get left over that isn't explained by that factor return and that's they call that the idio syncratic return. Um so you could argue that that you know the returns we saw in January where as we've said bonds have gone nowhere equity slightly up you know finan February's looking a bit different maybe but financial markets generally didn't you know that kind of first principal component didn't contribute much well so maybe it was more something idiosyncratic around gold and silver you know or or something else or natural gas um so what they they they do is say well if we look at this this idea of decomposing into you The main things explaining a portfolio return and the idiosyncratic return. The key thing that's interesting about these alternative and esoteric markets is that a much bigger proportion of bare returns are idiosyncratic. They're not coming from these these these big factors. Um and in hard numbers um you know so roughly speaking in traditional CTAs about half the returns are from factors and about half of idiosyncratic actually it's probably more like 45 55 45% idiosyncratic 55 or so factors. Um but in these unusual markets those these alternative and esoteric markets it's more like 1/3 2/3. So only about one/ird is coming from the the main factors and 2/3 is idiosyncratic. So the main thing about alternative markets is they give you an exposure to weird stuff to weird sources of return. Now if you're doing any kind of portfolio construction, um you like weird stuff, okay? You like diversifications, we've discussed, you like idiosyncratic returns. If the only returns in your portfolio were coming from trend following a kind of equity risk on risk off return um you you know that that's going to be not very helpful right to you you want to be able to be trend following other things as well getting itic returns from lots of different places. So what that means is is in a completely unconstrained portfolio if you do a standard portfolio optimization you're going to want to have quite a lot of these alternative and esoteric markets and of course this is uh one of the common themes of I think ever since I've been talking to you Neil's alternative markets which obviously is a you know has been a big thing for AHL um but also for for people like Floren Court for example was a rest and settle CTA that focus just on alternative markets and most big CTAs have have gone into alternative markets to a greater or lesser degree. Um so this is kind of something that that that's not a surprise um that that we already know about. Um and if they do this this kind of um sort of standard portfolio optimization, they find that um only about 30% of their portfolio needs to should be going into these traditional markets. Um which isn't a lot, right? And just to emphasize again, this is unconstrained. So this this assumes you've got no issues with liquidity and you can put as much into the weird stuff as you as you possibly want, which a big CTF of course can't do, right? Um because the they're they're just too big. And then they have um the rest of that portfolio, the 70% is in is in a mixture of alternative and you know alternative esoteric and and there's a bit of a breakdown and I can see that they put 5% into crypto for example. So the the crypto balls would would love that I'm sure. Now one question that again we we keep returning to is why should we invest in CTOs? Why should we invest in trend and following? And it it and it depend I think I've mentioned this before. It depends on whether you are looking at it as a standalone product or as something that goes alongside an existing 6040 or some other traditional kind of assets. All right. Now, if you're doing as a standalone product, you want, you know, to the first order approximation, you want a maximum sharp ratio. You'll do this kind of standard portfolio optimization. And so, as a standalone product, as a CTA without constraints around liquidity, what you would do is offer something that was 30% traditional, 70% alternatives, and that would be the best product for the investor to buy. But most people, of course, don't aren't doing that. Most people uh are putting a slice of trend following into their 60/40 portfolio. We'd like it to be a bigger slice of course, but you know that's life. Um and that means what they're looking for is an element of tail protection. Um and it's a pity we don't have Katie with us because of course she she wrote the book. Um and but um they're looking for for some kind of you know crisis alpha tower protection whatever you want to call it. So what um the these guys did was say, well, what if we um instead of looking at uh a portfolio optimization that basically just treats all states of the world the same, let's focus in on periods when equity markets fell because what we want to do is find the best portfolio for that scenario because that's when people really want to benefit from from what CTA performance is like. So they constructed their their their correlation matrix um differently um to to account for that and what they found um was that the traditional portfolio the traditional trend for things you trend for like you know >> so the liquid markets let's call >> the liquid well remembered Neil it's not just about liquidity it's about it's about so basically it's it's kind of it's futures traditional futures markets that a CTA would have had in it 30 years ago before people started doing all this weird stuff like trading German power and CDS and all this kind of stuff. Right. Exactly. So, it's things that are >> relatively liquid but also basically futures and therefore for a CT relatively simple. But more importantly, it's things that have a that whose returns are mostly being driven by this factor risk rather than this idiosyncratic risk. That's the key point about this article. That's the key innovation in how they categorize markets, I think. Um now it turns out that that actually what you want to do in the crisis situation is not have 30% traditional markets. You want to have 50% traditional markets and then obviously the the nontraditional stuff gets squeezed down to a smaller fraction. And the intuition behind that and which they explain is because it's that that total facto return that thing that's that you know that that thing that explains most of the risk of traditional markets that makes traditional markets look bad from a kind of generic optimization perspective because they haven't got much idiosyncratic risk. It turns out in a crisis that's what you want. Okay, that's what you want. And this kind of makes intuitive sense because in a crisis um you basically want to just make fairly big bets on going 2008 going short equities long bonds. You're not really going to need your your fancy um you know other weird markets there to to sort of help you. Um and of course there is a risk that many of those markets um might have issues around liquidity which is another thing that they they talk about next or around for example whether you can go short um because a lot of these things in a crisis you'd want to go short because a lot of them trade light risk on assets um because because um you know that things like well we talked about cryptocurrency you know spot cryptocurrency for example does that um so that that's kind of the the interesting thing now. Um they then take this a step further and say well all this time um you know um we've been kind of assuming that we're completely unconstrained. Um but actually another fantastic thing about about CTAs and or about anything that trades futures um is the fact that they're very cash efficient. So if I look at my own futures account um which has a a kind of a relatively high risk target for by institutional standards by you know not all some people are a bit crazy of course as we discussed but you know my my I'm probably running at about one and a half times the volume of say AHL is on most of their funds. Um I'm only using 30% of my cash's margin. So you know a typical CTA that number is 20%. And that's super cash efficient. Um and that that makes them very attractive from a structuring perspective. Um so the next thing they they did was say well let's say we want to um go a bit further and say well actually we want to kind of maintain this cash efficiency. Now your the non-futures markets obviously aren't as cash efficient. Um most of them have margin requirements that are higher than f well all of them have margin requirements that are higher than futures. Some of them even have 100% margin in the sense that you just you can't margin them. You have to just put put up cash essentially. Um and if they do that again it's not a surprise but but then what happens is is we we get even more into traditional markets. So we end up with 70% in traditional markets um and uh just just 30% um in the the alternative markets. Um and one final point to make is that when is cash efficiency most important is when we need cash. >> Yeah. >> When do we need cash? When the world's ending. So you know in 2008 um the fact that CTAs had plenty of cash, daily liquidity, no gating meant that many people use them as cash machines just you know. So we had this was a weird situation where although although you know CDS industry made a lot of money at the same time people were withdrawing it because it was a source of of ready cash that wasn't available elsewhere. Um, so that that that's yeah, so that's the that's the article. I mean, so there there's a lot of really interesting things there and like many like you know my many AHL articles, it's very well written, not technical, lots of nice graphs, very well presented and yeah, for me at least some really interesting innovations around a few different topics. >> Yeah. No, I couldn't agree more. People should definitely go and download it and and um and read it and as I said, I'll try to remember put in the link in the in the show notes. But I just want to kind of summarize uh all of this just to make sure people really get the the importance of this. Um and that is if you were just kind of trying to maximize and looking at your trend portfolio uh in isolation and you say oh what could be what you know what kind of market portfolio should I go for when I'm looking to maximize my my long-term sharp of of that um in all periods? Well, you would go for a portfolio that probably trades more markets. That's kind of what they they uh um conclude. However, as Rob rightly say, a lot of investors and certainly I would say most investors that I come across, they include trend following and CTAs specifically because they want to have something that really does well for them when equity markets are not doing well or when bond markets are not doing well. And in their work they find that then you need to find managers who are more focused on the traditional markets the more liquid markets the the classical futures markets we've been trading for the last 30 40 years um and go with one manager that that does that and then on top of that and I think I completely agree this is the interesting point they make um and that is if you then also are uh an investor where you might need to get some cash out in in terms of a crisis If your private equity fund is not giving you any money or drawing down or whatever comes during a crisis, then you need to be even more focused on managers that trades the the classical liquid futures markets. Um because the ca the the uh cash efficiency is important. Um and um and so I I thought it was really well done as well and um and to make that point um the way they did so that everybody kind of understand the difference between the choices we all have to make as managers. So it's not really all about kind of returns. It's also about you know um what's the best manager fit for me uh in terms of what my objective is with investing with the CTA. uh that should lead you down the path of identifying a smaller number of managers that might be relevant for for for for you. So really good uh paper kudos to the team uh for for writing that the next paper which we won't have time to do justice. So we'll we'll wait with that. Um but it's also a a super interesting uh paper. Um we'll see if we wait until Rob comes back or maybe we'll do it beforehand. But um it's something that is also super relevant uh in the current environment and what with what's going on in some of the um in some of the markets. I don't know if you want to spend 5 minutes, Rob, it's your uh it's your call >> uh on the Financial Times article you sent me uh this morning. I have not read it myself, so I don't know um what's in it, but but please do if you want to spend five, seven minutes on that. >> Yeah. So the the the article's on the FT alpha valal alphavville part the FT which is important because that's actually be in front of the pay wall so anyone can read it. Um and it's called the quant shop AI lab convergence. Um, and uh, it's it's kind of interesting because basically what they do in the article is they draw parallels between kind of finance quant and AI quant for one of a better word. Um, and their sort of thesis is that that basically these people are becoming more and more like each other. Now obviously there's always been a um a crossover between um the the sort of worlds of finance quant and you know software engineering more more generally and kind of um you know so and and and obviously a lot of places are using well let's let's be careful. So in finance, in systematic finance, we've always used something that I suppose you could wave your hands and call machine learning. Now machine learning to some people has to be very complicated, but um for example, if you're running something like a simple regression, which is a very simple a relatively simple way of of working out, for example, as we've just done, you know, how how much money you should give to a particular market. And you could use a regression to do that or you could use an optimization. Those are forms of machine learning. Um they're forms of what you might call supervised machine learning because the we're sort of telling the computer what to do and then it's giving us the numbers that come out. Um so that that's you know all the systematic finance is machine learning although some some you know some people say oh no that can't be machine learning because it's it's too simple but but you know then they can never tell me where the boundary is. So I say well no it must been machine learning. Um now then then you have something which you might call unsupervised machine learning and that kind of then blends into our AI and LLMs and neural networks and all this kind of stuff. Um and um the the sort of I think there's um certainly a lot of people trying to use AI uh and that goes for every industry of course not ours. Um I think pretty much every every fund um of a decent size will have some kind of AI thing going on. Um and uh how much of that there is is an open question. How much of it is just a marketing thing is just is an open question. How much of it kind of blue sky research like a you know let's buy a call option on this technology and you never know we might find something that no one else can do um is an an open question. I think the people who are doing it best are ones who are using it in very tightly defined areas. Um so things like for example execution research I think that's where it's potentially more interesting because that's where something like um well first of all you got a lot of data. Um so you you know we're building our kind of normal models with 50 odd years of daily data. Um and you know in execution research you've got tick data which is you know obviously much much richer much faster there's much more of it so it's much more suited to something like building any kind of nonlinear analysis which would include something like training an LLM on that data. Um, all that aside though, um, the article is quite interesting because rather than just making the obvious point that, you know, well, everything's AI now, you know, which is a slightly lazy argument. And I and I I do think they they do, and this is probably because they're talking their own book, they do slightly overstate, I think, the amount of of AI that that that sort of is going on. Um but they do kind of go further and say well if you actually look at the kind of workflow uh in quant finance and the workflow in our that's similar um you know if you look at the the sorts of people that are being hired um if you look at the technology stack um if you and even they even say well if you look at the the the way that kind of people are employed so you know we're used in we're used in finance to sort of non-competes and gardening leave and and things like that and and signing on bonuses. Well, they they've got that ini in AI now. And you know, you hear stories of people being offered kind of nine figure sums to move from from meta to from move to move from open AI to meta. So, so actually, you know, it many of the top AI researchers are making far more money than than any p quant finances, sadly, I have to say. Um, so so yeah, it's it's quite quite an interesting quite an interesting article. I am a skeptic of this stuff. I've said it before, I've said it again, but I still found it an interesting article. So, it must have been good. [laughter] >> Fair enough. Fair enough. Anyways, um we have come up to uh a little over the hour. So, we're going to wrap up uh our conversation today, Rob. This was uh fantastic. Thanks so much for spending all the time uh looking into this and and for answering the questions for Pedro. Uh I hope everybody found it useful. Um because I think actually our conversation today touches on on so many uh important points when it comes to CTAs and trend following and investing in general and all of that stuff. So if you want to show your appreciation for Rob and and all the other co-host uh please go to your podcast platform of choice um and leave a rating and review. Uh it really does help and uh we really appreciate uh the support we uh can get. Um if you have a question for next week where I will have two guests uh Andrew Beer will come back and he will be joined this time by Tom Roble from Sockgen. Um so you can send uh your questions uh as usual info@ toptraders onblog.com. I'll do my best to to uh get them in front of Andrew and Tom u when we speak. Um but for now for today uh 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 usual, take care of yourself uh and take care of each other. [music] >> 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 iTunes and subscribe [music] to 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. [music] And to ensure our show continues to grow, please leave us an honest rating and review in iTunes. It only takes a minute and it's the best way to show us you love the podcast. [music] We'll see you next time on Top Traders Unplugged.