The Derivative - Replicating Babies, Trend Following, Hedge Funds, and Warren Buffet with Corey Hoffstein

Episode Date: June 15, 2023

If a replicant in the 1980s sci-fi classic Blade Runner was a genetically engineered, bio-enhanced person with para-physical capabilities, "composed entirely of organic substance," created for slave labor – what does that make Corey Hoffstein and his takes on replicating?  On this episode of The Derivative, we're joined by the model flirting, possible replicant, Corey Hoffstein, who takes us through the intricacies of replication strategies, comparing different approaches and digging deep into the pros and cons of indices vs strategies vs replication. Learn about the challenges faced by replicators in the hedge fund industry, the importance of benchmarks, and the complexities of dispersion in managed futures.  Corey and Jeff provide insights into the factors that drive trend-following performance in different markets, explore the potential of alternative markets for managers, and delve into the replication of Warren Buffett's strategy, decoding its secrets. Plus, Corey discusses risk weighting, the role of market makers in the ETF ecosystem, and the integration of AI in various domains. This conversation goes beyond robots writing catchy descriptions. Get ready to become a replicant — SEND IT! Chapters: 00:00-02:04= Intro 02:05-04:13= Babies are easy 04:14-22:20= Efforts in Replication & Leveraged equity 22:21-33:14= When did CTAs, MFs, and Trend Following, become synonymous? Explaining Dispersion 33:15-41:52= Two ways to replicate a strategy, first up: Top Down 41:53-56:07= Replicating a Strategy: Bottom Up 56:08-01:04:29= The immaculate rebalance & not everything needs to be in an ETF 01:04:30-01:10:09= Chat GPT – Good, bad, or ugly? 01:10:10-01:19:21= Word association Follow along with Corey on Twitter @choffstein and for more information on Newfound Research visit thinknewfound.com. Also make sure to check out Corey's podcasts Flirting with Models & Pirates of Finance! Don't forget to subscribe to ⁠⁠⁠⁠⁠⁠⁠⁠⁠The Derivative⁠⁠⁠⁠⁠⁠⁠⁠⁠, follow us on Twitter at ⁠⁠⁠⁠⁠⁠⁠⁠⁠@rcmAlts⁠⁠⁠⁠⁠⁠⁠⁠⁠ and our host Jeff at ⁠⁠⁠⁠⁠⁠⁠⁠⁠@AttainCap2⁠⁠⁠⁠⁠⁠⁠⁠⁠, or ⁠⁠⁠⁠⁠⁠⁠⁠⁠LinkedIn⁠⁠⁠⁠⁠⁠⁠⁠⁠ , and ⁠⁠⁠⁠⁠⁠⁠⁠⁠Facebook⁠⁠⁠⁠⁠⁠⁠⁠⁠, and ⁠⁠⁠⁠⁠⁠⁠⁠⁠sign-up for our blog digest⁠⁠⁠⁠⁠⁠⁠⁠⁠. Disclaimer: This podcast is provided for informational purposes only and should not be relied upon as legal, business, or tax advice. All opinions expressed by podcast participants are solely their own opinions and do not necessarily reflect the opinions of RCM Alternatives, their affiliates, or companies featured. Due to industry regulations, participants on this podcast are instructed not to make specific trade recommendations, nor reference past or potential profits. And listeners are reminded that managed futures, commodity trading, and other alternative investments are complex and carry a risk of substantial losses. As such, they are not suitable for all investors. For more information, visit ⁠⁠⁠⁠⁠⁠⁠⁠⁠www.rcmalternatives.com/disclaimer

Welcome to the Derivative by RCM Alternatives, where we dive into what makes alternative investments go, analyze the strategies of unique hedge fund managers, and chat with interesting guests from across the investment world. Hello there. I've been walking with my son on some golf tournaments this week. It's been fun, but boy, are those kids good. 70s for 13 and 14 year olds. Wow.

You know who else is good? The pros on our panel out in Vegas that we got coming to you on next week's pod with Jim Carson, Zed Francis, and Luke Rubare edition on all things vol, including zero DTE, VIX, and where we go from here. On to this episode where we've got the replicant of Corey Hofstein speaking on replicating trend following. How all that works from the top down to bottom up approaches, why dispersion of managed futures results might not be as bad as it looks.

And I've got to say this, Corey is an A++ pod guest to go along with being an A++ pod host, which sort of annoys me that he's that good at everything, but hey, if you can get them on the pod, get him on a pod. That silky deep bass voice, the intelligent answers, the probing questions. He's a pros pro. Send it. This episode is brought to you by RCM's Guide to Trend Following White Paper. We talk about the conflation between managed futures, CTAs, and trend following names in

this pod. And we cover all that and more in the white paper, plus highlights on top managers, stats, definitions, all the good stuff for trend following. Go check it out at rcmalts.com today. And now back to the show. All right, everyone, we are here with Corey Hofstein. Corey, how are you? I'm doing great, Jeff. Thank you for having me. How are you doing? I'm great. We were just debating offline the origin of the lanai and if he's become a Floridian enough to use the word lanai instead of porch. I'm not there yet, but I also don't know whether I have a lanai. We really need to look this up. A carport is not a lanai instead of porch i'm not there yet but i also don't know whether i have a lanai we really

need to look this up right a carport is not a lanai right the uh so since you were on the pod last you've had a baby how's that all going i will tell you i was i was actually just texting my mother earlier saying i don't know what everyone's complaining about babies are easy now this is mind you after knock on two two and a half months of just my child would not sleep last night he slept for 12 hours it actually got like i i couldn't sleep because he was sleeping so well i was afraid something was wrong yeah it's uh mine had uh what did they call it group not group or uh whatever where they cry like every night from 4 to 8 p.m yeah which was awesome yeah we've we've got uh more recently like 6 to 7 p.m he sort of gets that

witching hour but for a while he he just i didn't know this about babies he was like a night grunter so he was asleep and he would sleep for hours but he would just make these grunting noises in the bassinet which made it impossible for my wife and I to sleep. So right in there. What's wrong? Yeah. And so, well, we had him sleeping next to the bed in the bassinet because he was young enough.

And we eventually just said, this is insane. So we started taking shifts. We'd be up for three hours and then the other person and then my in-laws, who are phenomenal, live near us. And they would come in and help do some shifts during the night. But knock on wood, hopefully he sort of figured it out you're strong enough for the co-sleeper i would do that and like reach over to grab the baby and hand it to my wife in the middle of the night and like throw out my shoulder yeah well i you know i won't say i was great my

side of the bed didn't have room for the co-sleeper. So it was always on my wife's side anyway. So. Oh, I wouldn't have even waken up then. She gave me a firm, a firm elbow and a kick to make sure I was up to help. All right. Uh, for all those people who didn't come here to have us talk about babies, I'll just offer my, what am I now? 12 and 14. So if you have any questions, shoot them at me. I appreciate it. The one now is the 12-year-old just turned and she's getting a phone, which I was heavy against, but everyone else has it. How is she supposed to take pictures when we're on vacation? Those are good questions. Great questions. But if Twitter and Facebook ruin her brain, I gonna be upset let's uh talk you had a thread which was great this is the

multiverse when you had a thread on resolves paper that we're coming onto a podcast to talk about so it's a paper turned into a twitter thread turned into a podcast which then we'll do a blog post about so that's how the multiverse works. Something I've been running into more and more, replication. DBMF came out screaming out of the block, so to speak. Was that two years ago?

Two years ago? I think a little over three years ago because they hit their three-year anniversary. That was a big deal. Everyone's talking about that. Now we started, I've started to hear more and more about, let's do this, let's do that, let's just replicate it. So I wanted to start with your kind of 30,000-foot view of like, why are we seeing more efforts at replication? Or are we, or is it just that I'm noticing it more?

Well, I will say, I think strong kudos to Andrew Beer here and his team with DBMF. He has been a longtime advocate of replication. And I think he's done a tremendous job in trying to educate the industry about where replication works and where it doesn't. If you go back and look at the literature, this was a really popular area of research for quants in the 2007 through 2012 area. And there were a lot of replication ideas and indices that got proposed, many of which fell flat on their face for potentially methodology reasons. And there was, I think the consensus of the research ended up being that,

yeah, you might be able to broadly use factors or market betas to capture some of these hedge fund concepts. But look, if they're creating true alpha, then you shouldn't be able to use the broad betas, right? It is the idiosyncratic return that really matters. I think Andrew presents this with a lot of nuance to say, there are categories by which you absolutely should not do replication. And there's categories. And I think the two he would really highlight are managed futures and long short equity, whereby it does seem like you can actually capture a really significant proportion of the total return using replication. And then if you can do that and dramatically reduce the fee, not just the flat rate, go from 2% to say 1%, but get rid of the carry, the 20 part of the 2 and 20, the incentive

fee, then whatever you missed in the alpha, you might be able to more than make up. And I think it's a really great story. I think the last point that I would look at is when most allocators look particularly at the managed future space, trying to make a decision about whether they want to allocate to managed futures, they're very often using an index. They're using something like the SOC Gen CTA index or Barclays Top 50 index, and then they'll do a manager search. And this happens to be a category of a pretty high dispersion. I think what Andrew's argument has long been is if you're using the index to make your allocation decision, why aren't you trying to get index-like returns? Why is that not good enough if that's what you built your allocation around. I don't know if we're seeing a huge rebirth in replication, but I think when it comes to 40-act products,

I think replication is an interesting approach that can be used for smaller teams to try to give beta-like exposure to these hedge fund categories. And then it's interesting, as you were talking through that, I'm like, yeah, 07 through 12 was all about risk premia. Basically, that was the same thing. They're just rebranding it now replication. David Collum, I would say risk premia is maybe a little bit different than replication, but the broad concept being a lot of these, what used to be fundamental discretionary ideas can be replicated with a systematic approach.

Risk premia was packaging that up from a quantitative investment strategy perspective and putting them in indices. I think replication is less about implementing the strategy necessarily and more trying to create the return stream. And there's a subtle difference there. And we can talk about the different ways in which replication works, but it's not about saying hedge funds do X, Y, Z, and we're going to replicate all that mechanically. It's hedge funds do X, Y, Z. And actually it turns out by using a combination of credit, stocks, bonds, and short vol, we can get a return profile that looks very similar from a big muscle movements perspective. But I think in managed futures in particular, they might be one in the same thing, or maybe

that's why replication. So like the strategy, you want the index and the strategy to kind of be the same thing. And so replication fits. But this is an area again, where there's a lot of conversation. So Andrew had a great, not to talk about another podcast on your podcast, but Andrew had a great conversation with Tim Pickering on Top Traders Unplugged recently. Tim is a long time CTA. He was one of the first people to actually put a CTA in an ETF almost a decade ago. And so he and Andrew had a debate because Tim is much more of a classical

CTA. Tim takes very strong off benchmark views. He happens to be very commodity focused. Whereas if you look at the broad SOC Gen CTA index, it's actually more equal risk weight between bonds, commodities, equities, and currencies. Andrew's saying if people are looking for the SOC Gen Index, they should just buy something that replicates it as closely as possible. What Tim is saying is they can look at the SOC Gen Index and still find value in off-benchmark weights in tracking error that's more valuable to the way they're building their portfolio, and that that active approach can add a tremendous amount of value for allocators. So it's a very open discussion.

And I think that's very healthy for the space. I heard that there was a bit of an odd debate. You thought it was a good debate? Well, I think they were talking past each other. I know both of them. I think it was one of those debates where it is difficult for two people who don't know each other to be thrown into a situation where they're debating virtually with a huge amount of context for the other person. And so I think it was a tough conversation.

I would say I think there were points, the two major points that they were disagreeing on. I think they were just largely talking past each other. I kind of hate that phrase, talking past each other. Right. What does that mean? It's like we were both ignoring what the other was saying. A little bit, or just not understanding what the other one was saying. Do you think replication is an American thing?

It seems to me there's something in there like, oh, we can just cheat this. We can do this better. Innovation. We're entrepreneurs. Why do we need to do the full thing? It just right like it's like a four work week before yeah yeah like uh let's do it i don't know if it's an american thing necessarily i don't have enough context for that what i do know is that the regulations here for the types of vehicles that

we have allow us to do things that aren't necessarily as easily done in Europe. In particular, putting commodities into an ETF or a mutual fund, often via a Cayman blocker, is a lot easier here than, say, trying to get commodities into a traded product in Europe. So there's a lot more complication around that in Europe, but there's a lot more... And again, I'm not an expert in the regulatory side in Europe. So you'd have to talk to someone who's done that dance. But my understanding is it's just less trivial than it is here. There's still plenty of hoops to jump through here, but it's much harder over there. And that might be one of the reasons you're not seeing the adoption. And then I will also say having

product both in the US and in Canada, different regions have different appetites for product. ESG is particularly strong in Europe, and there is very little appetite in the US right now for ESG, despite the fact that major index manufacturers have been trying to push it for a decade. And so you just find different regional appetites as well. And every third thing I read is about ESG, but it's not big. It's not big here, but you must read about it. Exactly. Well, they're doing everything they can to make it big. And then do you have any thought, we run our monthly asset class scoreboard where we show

managed futures versus hedge funds versus real estate. And just for ease of writing that post every month, we just use these ETFs. So the one we use for hedge funds is QAI, IQ Hedge Multi-Strategy Tracker ETF. So essentially a replication strategy. I don't know really actually what it does, but a lot of those I've seen on the hedge fund side haven't really worked. So then my brain goes, have they not worked? Or is that a true representation of what hedge funds have done? Yeah. So I think that's the interesting question potentially is if they have tracked their index perfectly, but the index hasn't done well,

would you say the product hasn't done well? Or has the product done exactly what it said it was going to do? Right. But it has a hedge fund where you're like, this is supposed to make 40% a year. What's going on? Yeah. I think the reality is when you look at broad hedge funds as a category, it's been a long time. I know exactly the product you're talking about, QAI. That was one of the first replication ETFs that came out. I think it's got well over a decade track record at this point. HFND is another one that came around. It's got 600. I was looking at it before this. That's 600 million in assets. Used to have a billion.

And if you look at the methodology, and again, it's been ages since I've looked at this. My recollection is that it's looking at a broad hedge fund index, and then it's using a regression-based approach to buy US equities, bonds, gold, basically major asset classes, and trying to capture the beta elements. And then with the idea being that when you look at hedge funds and aggregate, those beta elements are not moving incredibly fast. And so you can catch the turning points and capture the major muscle movements. I actually don't know what index they're trying to track or whether they've done a particularly good job of tracking it. but that does tell you the core concept, which is when you average all these hedge funds together, their alphas all cancel out.

You're left with a bunch of common betas, and hopefully we can use different betas to replicate. Trey Lockerbie, And then I think I've told you about this. A friend of mine here in Chicago runs a couple of funds. One's a venture capital index that's pure replication, has no actual exposure to venture capital. So it's basically a little more long NASDAQ, short some S&P to get the blue buds at her and get the real pop, real convexity of the NASDAQ. And then they're talking about doing one in the healthcare space and some other stuff. So it's like, how far can you push this? How far can you go to like, yeah, we're replicating healthcare costs with e-mini futures. Like what?

Is that real or is it just an anomaly? There were some papers that came out a while ago about replicating EE and VC using levered small cap. You have to take the right sector tilts. I think the common idea here is whether private or public, whether hedge fund, are there common betas? Are there common driving factors that if we're looking at trying to capture not a particular VC or PE fund, but the industry as a whole, to all their bets sort of averaged together to look like a levered equity market beta. Ben Felix. I think, again, it's the same concept,

whether you're looking at a diversified basket of hedge funds and trying to replicate that, a diversified basket of managed futures managers, or trying to look at some of these private funds. Can you try to replicate private real estate investments with publicly traded REITs? Maybe. But I think that's a lot of what we're seeing nowadays. In fact, I think an ETF just launched recently, or maybe it's launching tomorrow, depending on when this is timestamped, that is someone who's trying to replicate privately traded REITs using publicly traded REITs. So you can't get out of the ETF? Yeah, exactly. We've replicated it. Your money is now locked up for three years. But that makes me think of everything's just equity exposure. If we can replicate all this stuff with different leveraged versions of different equity sectors, that really proves

the point that, hey, you think you're diversified, you think you're in this hedge fund strategy, that alt strategy, really all you have is some different levered version of some equity index. Yeah, I think the core question sort of becomes how constant is that versus how much does it change? So I went through this exercise years and years ago. I wrote a research note on my blog called Attack of the Clones. That was a clever Star Wars reference, but it was all about, yeah, there you go. It was all about trying to replicate long short equity managers. And what I found was that you could get a high degree of replication to a broad basket of equity long short managers using things like broad equity exposure to the US, international exposure, emerging market exposure. But you needed some other factors in there. You needed to be able to make growth versus value tilts, large cap versus

small cap tilts. And if you had the ability to make those tilts, what you found is you could replicate pretty closely, but those tilts were definitely time varying. So for example, what I found in just the replication was that the index was heavily tilted towards foreign markets in the 2000s, dramatically cut its net equity beta in 2008, and then went predominantly towards US equity markets post 2008. Now, if I just said, I'm going to replicate equity long short by just having a beta of 0.2 or 0.3 to global markets, it would be a much noisier fit than if I had these other micro levers that I could change over time based on how the index performance was changing. And do you think there's anything to like, that's too simple? You think investors are like,

I want the fancy, I want the big deck on everything that's going simple you think investors are like i want the fancy i want the big deck on everything that's going on instead of like hey i give you 0.2 beta to the the index enjoy it i think there is a certain degree of the index is fine but i there's particular managers that i think will outperform the index or for the particular allocation i want to choose a subset of these managers like I don't just want equity long short. I want healthcare long short because it fills out something in my asset allocation profile. Right to the conversation that Andrew Beer and Tim Pickering were having, Tim might say, look, managed futures is phenomenal, but if you're a stock bond investor, a commodity tilted managed futures product is better than a benchmark product

because it's more likely to have structural diversification versus what you already own. Yeah. Agree. Agree, disagree. I just think it's an interesting argument. I don't want to, again, make his argument for him. But the point there is, okay, how relevant is the benchmark? Maybe there is a good reason why you would choose an off-benchmark manager, and therefore replication isn't the approach you should take. And the thing of replication is kind of like you're throwing in the towel. It's like that stat 76% of drivers think they're above average drivers. Right. I think it's even higher than that.

Right. But the same with investors, investors right like 78 think they can beat the index if you're like i'll just do this replication strategy it's almost admitting like i know i can't beat it i know i'm not gonna beat the index let's just match it and the index for many people might be exactly what you want right um unless you like let's consider let's go back to managed futures how many people really have the capacity and wherewithal to understand managed futures managers and do the adequate amount of due diligence? Like they might be able to understand the category and they might be able to understand the historical return drivers. They might like the diversification profile,

but can they take that next step to say, I can choose the manager that I have a higher degree of confidence in? That's a different skill. That's not the same skill as being able to do a top-down asset allocation. Manager due diligence is a unique skill unto itself. And so I think we can have a little humility to say, I like the category. I like what it's doing. I like investing in equities, but that doesn't mean I can pick a good equity manager. It's not throwing in the towel to buy equity beta. It's one of the best long-term drivers of returns ever, right? Yeah. Yeah. So if I say I like managed futures as a category, is it throwing in the towel if I buy five managed

futures managers? Because I don't know which one to pick. Is that any more throwing in the towel than trying to replicate the index? Good rhetorical question. We'll leave it at that, possibly. Let's get into your thread, the thread on the paper on the Twitter. You started out by basically like, hey, look at all the dispersion to manage futures. That's a problem because you can't, or you don't maybe want to, per what you just said, you don't want to invest in all 30 of them. So let's do five of them or let's do two of them. But then you have this dispersion, you might pick the wrong two. So I have take a little issue with that dispersion chart.

I know it's not yours. Is that resolved? Put that together. Yeah, this isn't mine. So don't get mad at me for a chart that's not mine but i know what you're saying yeah well i just quickly and i am the guy who like okay let me look under the hood so i looked under the hood at all the ones in that chart and it's um it's got abby in there that has crable which is super short term pe that's a currency trader, Epistem, that's super high frequency.

It's got the alpha centric one, that's 50% mortgage-backed securities. It's got low cores, long, short commodities, no equities, no. So there's a lot of stuff in there. It's got Man AHL, who knows what they're doing these days. They have AI and all sorts of stuff packed under that. So it's a common problem in our space, right? Like managed futures, what does that really mean? And if you just quickly take that jump to that means trend following, yes, you're going to have a lot of dispersion in there, but you're also going to have a lot of dispersion from the trend following index. So- Well, I think the broader issue here is as an industry, CTA, managed futures and trend following all became synonymous

because historically they work, right? Managed futures was a category of strategies that was mostly being implemented by CTAs using trend following. More recently, that's diverged, particularly in the 2010s where trend following didn't work as well. You had a lot of folks start to introduce carry and other different types of strategies, seasonality, relative value trading, all of which are done via futures and are technically major future strategies. So I think if you look at the SOC Gen CTA index, I think it's still over 50% trend, but there's now a decent amount of airy and other stuff going on in there. Preston Pysh, M.D.: For sure. To me, that explains, now a decent amount of airy and other stuff going on in there right for sure um and so that to me that explains yeah most of the dispersion is explained by that well hey yeah it does was down

four percent because they do it does but let's explain but let's like just let's assume we're only talking trend following for a second right let's get rid of all the other variables. I mean, there is a huge degree of dispersion that it can occur just within trend following because of all the degrees of freedom of choice, right? So let's start with a couple of things we can do when we're building a trend following program. One, what's the universe we're using, right? The futures market, I've seen trend followers who do nine futures markets, and I've seen trend followers who do 90 plus. Two, what speed of trend are we looking at? Really fast trends all the way to intraday to, 25, 25, 25 commodities, currencies, equities, and rates on average. But you look at individual managers, they're nowhere near that. You get massive… Anyone who's just listening, there's a pirate walking around, a very handsome pirate walking around in the background there.

So you can get some very off benchmark weights for individual managers. Someone might be very heavy commodities. You see these managers who will say, we're only trading rates and commodities, no currencies and equities. Or you might be constrained in different ways. And then it's a question of how are they running their portfolio optimization? Are they doing risk parity weights? Are they doing mean variance optimization? Are they doing something more naive, more complex? All these degrees of freedom of choice lead to really big dispersion in the space. And the numbers show that. Yes, I get that chart that was put together in the space. And the numbers show that.

Yes, I get that chart that was put together in the paper maybe has more dispersion than just pure trend following. But when you look at academic research, this is a category that has some of the largest dispersion. It is a category where buying multiple managers has proven to dramatically decrease vol, dramatically decrease realized drawdown, dramatically decrease dispersion and terminal wealth over the long run, which is all a sign of just, again, large dispersion between managers because there's

so many ways in which they can do things differently. Yeah. My pushback on that would be like, if you spend the, I want to say 10 minutes, but let's call it 10 hours and dig into what they're doing. You don't have to have a PhD in due diligence, but just like you're saying of like, they trade our trends on just fixed income. Like, okay, I'm probably not getting the trend following profile I want out of that manager. So of course there's going to be bigger dispersion there. So if you really want that index performance and you focus in your search on the ones that give you that index performance, it's not as hard as... The replication seems to be fixing a problem that maybe doesn't

exist in that regard. Maybe, but all of them are getting rolled up to the broad index when people are allocating from a top-down perspective and they're looking at, I want the SOC Gen CTA index, these different managers are getting rolled up. Let's not even talk about hedge fund indices, go to Morningstar. They've got, I think, a managed futures category and a systematic trader category. The dispersion in there is incredibly large as well because they're just sort of catch-all categories. There's so many different ways in which these concepts can be implemented that it becomes really hard to come up with subcategories. Maybe not really hard, but they just don't want to do the work because there's so many

funds there. And so you say, oh, I'm allocating to this category. You choose any three managers at random, you're going to get very different performance. We could do a whole pod and there's like manager firm level alpha and choosing the right category, right? If you're in the wrong category, maybe no investor sees you. You're in this category. Absolutely. You're at the top of it and you might be doing the same strategy, but you're like, well, I'm the top performer in the option space. Don't think that people don't try to game that at Morningstar, for sure. Yeah, for sure. So moving on, let's get to the,

you mentioned it before, top-down, bottom-up. Explain some of that. Oh, I wanted one more thing on all these manager choices, like especially in that future space. Like I want to say, of course, this is doing that, that's doing that. But maybe if the public doesn't know or doesn't care to know, does it make a difference? So that's kind of to your point of like, it's still all in there. It's all rolling up in the CTA index. I would argue it's not rolling up in the trend index, right?

Because that's designed to fix that problem. Yeah, a little bit. Look, I mean, there are trend followers. I'm blanking on the name, but I was listening to a great episode with a trend follower. Their firm specializes in only doing alternative markets, right? So you're talking Turkish interest rates and you're talking inflation swaps. They don't hold any of the major markets. They do trend following, and it's all alternative markets. And their view is they're a completion manager, that there's really a lot. Their alpha as a firm is all the operational work it takes to actually trade these markets. That sort of thing is incredibly hard to package in a mutual fund or ETF. But that's going to have a huge amount of dispersion relative to the index. Perhaps it'll have a decent amount of alpha. I haven't looked at the numbers,

but it's an interesting concept. Again, it's going to look very off benchmark by definition, and maybe that's a good thing. Maybe that's a bad thing. When you start to think about putting these into a mutual fund or an ETF, you start to hit structural limitations of the vehicle choice itself that does narrow some of this dispersion. But even still, you can find a pretty decent degree of dispersion among managers. Even just this year, look how different managers responded to the bond rally in March. Were they using stops? How much did they have bonds? A lot? A little? All that stuff really created- Right. Did they cap it by market, by sector? Yeah, exactly. It creates a pretty strong level of dispersion.

Trey Lockerbie, MPH, While you brought up those unique markets, we had Sarah, I'm going to forget her last name here live, but used to be at AQR working on their trend, what did you call it? Unique markets or- Preston Pyshko, Yeah, alternative markets. Trey Lockerbie, Alternative markets trend, which led them into crypto. Now she's doing crypto for Coinbase asset management. But my question to her, which interested in your take, who wants that? Like, cool, I'm going to get this alternative.

Maybe it performs well, but how much am I going to allocate it to where it moves the needle on my main trend? Like it's almost not in there for a reason. It's not going to take your trend was down 10% one year. It's not going to take it to up 10%. Maybe it takes it from down 10% to down 9%. Well, I think there's two questions I would ask, and I've never done due diligence on these managers, so I don't know how they would respond. My first question would be, how much capacity do these different markets have? If I'm trading Chinese Apple futures, how much can I start to allocate

before I dominate that market? But I think I would answer on their behalf, which is, you say trend is up 10% or down 10%. Is this really going to be up 10%? It might be. These markets might be so different that you're not expecting it to behave like the benchmark. And if you can make it a large part of your allocation before they have to close the fund for capacity reasons, that can be a really valuable diversifier as an allocator. Now, I did listen to an interview with this manager, and they said, actually, surprisingly, even though it's all alternative markets, they actually catch a lot of the major standard market movements because correlations are correlations. There's only so many global macro muscle movements that are happening at any given time. These things bleed into each other.

And so even though they tried to stay alternative, they did track broad trend exposure at large. Yeah. No, I would say more saying more of if the main trends down 10 and this thing can be up a hundred, but how much are you realistically going to allocate it, allocate to it where it's a meaningful portion of that's going to take your negative 10 to plus 10 on the portfolio level, right? I mean, I feel like you, I would argue diversification is always valuable, right? No matter how much you have in it it whether it's a small amount or a large amount it's just a question of comfort with that manager so if you can get the diversification you should almost always add it so let's talk the paper the thread they took two approaches top down and bottom down let Let's start at the top.

See what I did? Yeah. So this is a question about the replication methodology, right? Yeah. How do we replicate? So there's a lot of ways in which you could replicate a strategy. I often find talking about hedge fund replication, like managed research replication, be a little bit more confusing than it needs to be. So I'm going to talk about replicating Warren Buffett, because I find that that's actually the easier way to explain it. Put on a fat suit. Yeah.

So let's say we wanted to replicate Warren Buffett's returns, the returns of Berkshire Hathaway. We didn't actually want to buy Berkshire Hathaway for whatever reason. We just wanted to replicate it. There's two ways that come to mind in which you could do this. There's more than two ways you can do replication, but there's sort of two major ways are what we call top-down and bottom-up. of stocks, which stocks and how much we should allocate to them that would give us a return profile that is as close as possible to the return profile of Berkshire. And so you can use all sorts of mathematical techniques to try to suss this out. But the salient point here is that we don't

actually care how Warren Buffett is picking stocks. All we care about is trying to find a portfolio that gives me returns that are super close to his portfolio in the returns. In theory, we would hope it actually backs out exactly what he's holding. But we know that Berkshire, for example, has private holdings that we can't get. And so this mathematical regression-based approach might actually identify some proxies. We can't buy some of his private holdings, but there are public holdings that end up creating a price return stream that fills the gap. And so you get this sort of approach that says, let me try to figure out what he's holding and how much he's holding at any given time to replicate his returns as

exactly as possible without caring how he picks stocks. The problem with this approach, in theory, is that if he suddenly, say, sells all of his Geico or all of his Apple, and I'm looking at the last five years of Berkshire Hathaway returns to figure out what portfolio to own, I'm going to miss that turning point. I'm assuming that he's holding that portfolio constant. That may not be true. And so the question is, well, how much data can I really use? How often is he going to be shifting his portfolio? Warren Buffett may not shift his portfolio very quickly. Trend followers do. So if I'm trying to replicate the trend following index, it's the same concept. I'm going to look at the index's returns. I'm going to try to figure out which futures markets to hold, how much long, how much short in each. That gives me the closest portfolio to replicate the returns over the recent

history. But I can't use that much history because trends can turn on a dime. So maybe I can only look at the last 20, 30, 40 days of returns. And then the core assumption is that portfolio I'm going to hold for the next day or maybe the next week. And then I'm going to have to do all the analysis again to make sure that that replicating portfolio hasn't changed dramatically to no longer fit the most recent returns. So that's the top-down approach. Which I was confused about in trend following space. I thought they're looking at, hey, we can replicate a 60 market portfolio with these nine markets, but then we're going to trend follow those nine markets with our own model. You're saying, no, that's bottom up.

But top-down is like, hey, we're going to just, we know over the last 40 days, a lot of this return came from being long 30 years and some 10 years. That's what we're going to hold to replicate it. And I have no trend signal whatsoever. Yeah. And you have no trend signal. In fact, you don't even know. You don't care that these are trend following signals.

They could go from trend following to using carry signals. And all you're trying to do is replicate the returns. Now, there are things you can do as a manager to try to dial in certain profiles. If you do know their trend following, you can come up with some trend following factors that you can try to use in your regression. There's ways to be smart and clever here to build a more robust system. But just generically speaking, the top-down approach is to say, I don't really care how the return profile is being generated. All I want to do is find the portfolio that replicates the return profile as closely as possible and assume I'm going to hold that going forward over some short period of time. What's kind of cool here about this is,

let's say the SOC Gen CTA index or the SOC Gen Trend Index is what we're trying to replicate. Maybe I, like DBMF doesn't trade 60 markets. DBMF trades, I think, 13 markets, 14 markets. It doesn't trade Japanese government bonds, for example, which are a pretty common element of most CTAs. But what you might find is while it doesn't trade JGBs, there might be some mixture of US interest rate movements and yen dollar movements that come pretty darn close to explaining JGB movements. And so you might find that it can find sort of a basis trade. By weighting these different futures, you might only need nine or 13 futures to explain the big muscle movements of the SOC Gen Trend Index. Which comes back to my other point, then why bother with those

alternative markets at all? Now you're talking things like cotton and silver. Well, if you're trying to track the index, you might not need to, right? So common sense here is that the index itself is made up of very large players. Very large players are going to be capacity constrained, which means the majority of their dollars are getting allocated to the most liquid markets. So if you then say, I'm now averaging all these major players together, those small markets become an even smaller part of the index. You need all those managers to be trading those small markets in the exact same direction for that small market to really

influence the index. So you would say, okay, I can replicate the major muscle movement of the index with just a handful of 10, 12 futures markets. That really explains almost all of it with the caveat of if you can pick an individual manager who's really good, they might be able to generate a tremendous amount of off-index alpha from those small markets. But again, that doesn't really matter at the index level. But it's almost like the very definition of trend following. I'm going to risk a little bit to make a lot in those markets. They're almost, by definition, additive. Right. Yeah. Again,'s weird over the long term.

Generally speaking, you don't want to forego diversification when you can get diversification. But to your point way earlier, but maybe if that saves you a 20% incentive fee, maybe it's worth it to get rid of those. How much alpha can they generate in that? A, what's the probability

of you picking that manager? B, is the amount of alpha that that manager can generate going to exceed the higher fee versus the replication strategy? Okay. And yes, there may be a large number of managers who can do that. So that's top-down. And do you think, real quick, do you think top-down approaches, because surely people do try and do that with Buffett and other hedge funds, right? Of like, we're into these, what do they call them, hedge fund hotels of the most popular names and whatnot. Yeah.

Like, do you think there's like a self-fulfilling action there of like, people are trying to get into the same names, trying to replicate, and then maybe they get out too late? Maybe. I mean, there's an inherent lag to the process because you're just saying, let me replicate the recent results. You could argue that there's actually a self-fulfilling aspect to it of if enough people are trying to replicate, they actually are chasing the managers and creating that positive price pressure to the names. It's hard to suss that out, particularly with like 13F types of strategies. But you definitely see that in the hedge fund security selection side. There are 13F strategies

that try to go through and find the high conviction names of hedge fund managers and then replicate those in a basket. All right, bottom up. So bottom up, if top down is about we don't care how Warren Buffett is actually picking stocks, bottom up says, actually, we really care. We want to figure out what he's doing. Warren Buffett, we believe, is picking high quality value stocks and then levering them 1.6 times. So what we're going to do is we're going to come up with our own process that picks high quality value stocks and levers them 1.6 times. And we're going to try to pick a process that gives us results as close as possible to Berkshire's historical track record. And then going forward, we're going to kind of ignore his track record and just keep running that process, right? We're going to every day say, okay, which high quality value stock should we buy today and lever up 1.6 times. And we're basically using his historical returns to inform

and our knowledge of how we pick stocks to inform a stock selection process. What does that mean for trend following? If we're trying to replicate the trend following index, let's now cross that chasm. Well, it might mean what we do is we build a different set of trend following indices. We might do short-term trend following, intermediate-term trend following, slower trend following. We might do it on commodities and rates and currencies and all these different markets. And then what we're going to do is we're going to try to find the weights of those different systems that when taken together, broadly replicate the index. So we might find, for example, and this

is something that the paper finds, is that something like NatGas seems to be traded a lot faster by CTAs than something like the FTSE 100, where the FTSE looks like it gets traded mechanically a lot slower. NatGas seems like when you look at the contributing returns to the index, it gets traded a lot faster. You put all that together and what you're left with is actually a set of trading strategies and how much you should weight them. So it might say, hey, look, you need to run the 60-day moving average system on nat gas every day. You need to run a 220-day moving average system on FTSE futures, figure out whether that signal is positive or negative, and then trade it. And so what you're left with is a strategy that is generally an ensemble of different signals, different weights for different futures

contracts that you are running every single day. And that strategy is in and of itself a trend following manager in many ways. You've created a new trend strategy that has been designed and tuned specifically to try to replicate the way the broad index works or looks historically. Preston Pyshko So that's interesting. You think that the standard there would be to create a model per market? Because a standard trend follower says, I don't want to do that. I don't want to overfit. And I'm going to have one strategy for all the markets. There are a lot of ways to do this.

So this is where there's art and science into how do you do the replication. All I would define the bottom up as being is you're trying to create your own trend following strategy where you're selecting parameters and awaiting methodology and all of that, such that what results is a trend following strategy that looks a heck of a lot like the index. What the paper does that Resolve wrote is it basically creates... So there's 27 futures markets that they look at. They create trend signals from five days all the way out to 260 trading days for every market. And then when you take 27 by all those different signals, they've got... I'm trying to remember the number of actual trading strategies they use. I mean, it's hundreds that you could potentially select from, if not more. And then what they do is they run a long-term regression and say, what weight of all these different markets stably gets us a profile that looks a lot like the SOC trend index? Trying to maximize the diversification here to take into account diversification as much as we can while minimizing that tracking error.

We don't want to overfit. We don't want to just rely on some super isolated fit of a market. So what you find, at least when you replicate that, is you get a spread. For the most part, the weights seem to be in the sort of 150 to 220 day area. It's sort of these nine month trends, but right. I said, Nat gas leans a little, uh, faster footsie leans a little slower. Um, that faster might be okay. It's 60 days to 200 days versus, you know, the footsie is something like 150 days to 260. Like There's still a range of parameters that are getting used to try to create a diversified ensemble to replicate the SOC Gen index. Nick Neumanis, And that's super interesting to me because

I don't know many trend followers who might trade NatGas on five-day look back and everything else in the portfolio 200. So it's interesting, they're not looking at that performance of nat gas on that look back on its own price data it's just how does it fit with the index does it make the index does it make our tracking closer or further from the index okay how does it combine with everything else exactly and so what you find actually is when you look at the data, they didn't choose 5, 10, 15. I think 20-day was the fastest parameter that got chosen for any model. And the weight on that 20-day was like less than 1%. It's a very small proportion of the risk. Vast, vast, vast majority of the model risk is in that 150 to 220 day range.

And then what would you say this makes, right? Often a pan on managed futures on trend following is like, oh, you just make all your money from interest rate moves. Like, why do I need all those other markets? If I just trend follow some, you know, the 30 year bond, 10 year notes, and maybe some JGB, like you said, I can get most of the performance. I think some of that negativity also comes from like, oh, you just get the tailwind of holding the cash in T-bills. But ignoring that for a second of like, okay, does this all just boil down to, well, yeah, of course, I just got the core of what drives trend following performance, which is the bonds. Yeah, I'm not 100% sure that's always true.

I mean, the bonds are important. I think what you do find historically, and who knows how things will go going forward, what you find historically is like equities haven't been a big contributor. I think that's predominantly not because the signals haven't been good. But what you find is that when the signals on equities turn off, that's often when vol rises dramatically in equities. Or you have that strong negative correlation between returns and volatility with inequities, which means those position sizes in equities often get crushed pretty significantly. You don't see the same thing necessarily in commodities where you can actually have crashes up, right? And same with bonds. You can see crashes up in bonds. You don't tend to see equities crash up.

It's just a very different dynamic. Currencies as well. So I think, if anything, the argument I often hear is why even bother having equities? Again, I would say, yeah, it takes a part of the asset allocation away from rates and currencies and commodities that you could be allocated to otherwise. But there is an argument of diversification potential, and you don't know what the future necessarily holds. There could be market regimes in which equities could be a much more positive contributor. I don't think the evidence is there, at least as far as I've seen that it's always rates. I mean, yes, there is the tailwind from holding cash. But what I would argue is that that tailwind exists academically

in every asset class. We always look at asset classes as their excess return, right? That's why we talk about the equity risk premium. It's an excess return above cash, bond risk premium, excess return above cash. Managed futures, when I look at the performance managed futures, I don't include cash. I take the cash out because that cash return is just a risk-free rate that exists as the base return for evaluating every asset class. So I think that's a bad argument, in my opinion. I won't say there are many arguments that are bad, but I think that one's just fundamentally bad. So the question of, are all the returns coming from trend following and rates, I haven't seen it. I think if you isolate trend following in

currencies, isolate trend following in commodities, isolate trend following in rates, they have all been long-term return drivers. We had a good blog post maybe a year ago now when Silver had a move, and we ran it on the socked and trend indicator it's like the last 15 were losers over maybe 15 years or more right it's like why would anyone in their right mind have that in their portfolio and then this one was a six sigma outlier or whatever and like boom yeah you get that massive move um and how do you think about like now i've created this model it's tracking that but in order to avoid idiosyncratic risk i've like added my own idiosyncratic risk right i've like

created my own problem by trying to avoid those problems yeah so this is this is the key right so this is this is a part of the conversation that that tim pickering and andrew beer were really getting after each other and that talk is unplugged. Right. Andrew's saying the whole point of doing replication is to avoid single manager risk. And Tim is saying, all you've done is created your own single manager risk. You're a single manager. I think the point that, again, I know you don't like the phrase they were talking past each other. But I think the core point of what Andrew was trying to say is, if we do a good job, our dispersion from the index, and by definition,

you're not replicating perfectly, there will be tracking error. That tracking error should be substantially lower than the expected tracking error of picking any one manager at random. So yes, you are absolutely introducing model risk. That model risk is absolutely a form of single manager risk. It's a question of if your dispersion around the index is call it, your tracking error is 300 basis points a year, but as a category, the dispersion around the index is usually 800 basis points a year. But as a category, the dispersion around the index is usually 800 basis points a year. You could argue you have substantially less single manager risk. You have cut down on the single manager risk because of the way you've designed your program to try to track the index. So yes, absolutely. Whether you do top down, bottom up, there is a degree of model risk that is its

own unique single manager risk. It's just a question of if you can do a good enough job to keep that substantially lower than the risk people might realize otherwise in selecting one or even a small handful of managers to allocate to. And then my last thoughts on replication, we'll move on. Why don't the big pensions, institutional investors just do this on their own in-house? Why pay these high fees and get these managers just replicate it? Why don't they replicate or why don't they run their own strategies? Why don't they run their own replication? Well, some of them do run their own strategies.

Yeah, absolutely. Teachers. Yeah. So that's out there. I don't know whether they run their own strategies. Yeah, absolutely. Teachers, yeah. So that's out there. I don't know whether they run their own replication or not. Again, I think it's a question of... There's probably a whole lot to that. I'm like, well, we'll get fired if we run our own thing in-house and whatnot.

Yeah, there's definitely principal agent risks there that should be considered. There's the operational burden of actually doing this stuff like is it in their best interest to be spending time and money to like try to run a trend following strategy and then if you're gonna do it do you really want to try to replicate or do you think you can find a particular edge if you're going to go through or or build a program that is designed very specifically to take into consideration the other things that are unique to your allocator profile. Again, not many people have the ability to say, this is what our allocation profile looks like.

Let's generate a strategy to specifically address that because we're not taking outside money. Most of us who are managing money have to assume the money is coming from a variety of sources who have a variety of different utility profiles and wants and needs. So you end up somewhere with a more generic concept rather than a hyper-specific one. So there's a variety of reasons. I would be surprised if you don't start to see more replication get picked up, particularly as DBMF grows. I don't know whether just from a behavioral perspective, it'll ever truly replace

the majority of hedge funds. And I would hope it doesn't, right? Because in a weird circular way, it needs those hedge funds to be successful at what they do for it to continue to replicate the index. It wants those hedge funds to create a good index exposure. Right. It's like Mike Green's passive. If we just get all these big replicators, what are they replicating? Yeah, exactly. Exactly. So you need the hedge funds. What I would say is that what I would expect to start happening is smaller institutions that can't get access to some of these hedge funds might start using replicators. And the larger institutions, while they're doing a manager search, might use the replicators to fill the liquidity bucket. You see this pretty often with credit.

Yeah. Or they might be looking for a high yield manager. And in the interim, they'll use something like an HYG just as an index proxy. I could see them doing a manager search and manage futures, but not wanting the cash to sit idle. And so they might use a index replicator. All right, I got five unrelated, not five, three that we'll go quick on. The rapid fire session? No, before that, just your quick thoughts on rebalancing real quick. Do you want to give us your 30,000 foot view on rebalancing, rebalance timing luck? Do you have an elevator pitch for it? You got another three hours?

Yeah. No, I want it in like half a paragraph. Half a paragraph. Rebalance timing luck is one of the largest sources of uncompensated risk in your portfolio that you know nothing about. But here's a real pitch. When you rebalance, it turns out it matters a lot. It defines the opportunity set that you are seeing, which I think makes sense. So people who are rebalancing once a year in something like a value strategy, turns out if you rebalance at another point in the year, you could have incredibly different performance. This impacts the indices we're benchmarking to, it impacts

the managers we're evaluating. And just the same way we diversify across holdings and process, we should be willing to diversify across when we rebalance. Which is as simple as dollar cost averaging fixes that? Yeah, basically. So the real easy fix is to say, let's say, again, I think it's sort of easy to use stocks as an example. If you're a value manager and you're, for the Russell 1000 value, rebalances once a year, well, what it should probably do is rebalance one 12th of its portfolio every month. And in doing so, it takes the emphasis off of the when there's a very famous case of this called the immaculate rebalance research affiliates, March, 2009. They just so happened to rebalance in March. That was a totally arbitrary

random decision. Ended up creating about 1,500 basis points of excess return above their benchmark versus if they had chosen to rebalance in September, they would have underperformed their benchmark. Which led to like 8 billion in assets. Yeah. I mean, you have to ask, how would history be different if they had underperformed their benchmark in 2009 versus absolutely smashing it? It's a very different path that they would have taken as a firm, I would argue. To their credit, they recognized that they got lucky. I don't know if they'll admit they got lucky, but they inherently recognized it and moved to this staggered rebalancing approach where once a quarter, they rebalanced one fourth of their portfolio. Did you coin that term or that was coined elsewhere? The immaculate? No, that one, I don't know who coined that term. It's one of these weird things that in certain

circles is incredibly well known. And then there's other, for many other folks, it's just something that was lost to history. And what are your thoughts? I argue with our pirate friend who just walked through of like, rebalancing isn't always great, right? If you're rebalancing into a loser, say Bitcoin since 20, whatever, since it's high as you are losing more than you otherwise would have. Yeah. I actually just argued with our pirate friend about this last week. All right. So there's two things I'll say. Look, if you are rebalancing into something that's trending, it's going to hurt you. If you're rebalancing into something at the right time of its mean reversion cycle, it'll help you. So there's the dynamics of what you're rebalancing into.

I think that's hard to predict. What I will say at the aggregate level, right? You become a timer, right? You're inherently trying to time. What I'll say at the aggregate level, right, is compound annualized growth rate is approximately equal to your arithmetic expected return minus your variance drag. If I can rebalance and keep my expected return the same, what rebalancing does is it hopefully keeps minimizing that variance-driven component. Because let's say I don't rebalance. Let's say I have

stocks and bonds, and I don't rebalance for 30 years. Well, what happens? My portfolio, as stocks go up more than bonds, will become more and more stocks. And at the limit, the variance will then approach the variance of an entirely stock-driven portfolio. And the expected return might go up. But what you might find is the compound growth rate actually goes down. The expected return might not compensate you for the variance drag. Versus if you can rebalance back to the profile that maximizes the compound growth rate, that has nothing to do with whether the assets are trending or not. What you're trying to do is maximize the expectation of growth.

And so to me, rebalancing is much more around making sure we're diversified in a way that is maximizing our opportunity for compound growth. My second quick one was the ETFs versus mutual funds. You said on a panel in Vegas, not everything needs to go into an ETF. Don't use my words against me, Jeff. Do you have a quick explainer on that? Because now it seems like everything we just talked about, like, well, you could just replicate it and put it in an ETF. Yeah. Strategy and structure are different things, right? And an ETF and a mutual fund are fundamentally different things. For those who aren't super well versed, the way to think about it is an ETF is really just a mutual fund that trades on an exchange. But that last element

is super important because it means the market maker community is really intimately involved, and it requires a high degree of transparency into what you're doing. So there are certain strategies or certain things you might want to hold that would make it hard to put it in an ETF. So there's a well-known CTA that I tried to hire as a sub-advisor for one of their strategies where they implement some shorter term mean version concepts, five-day type strategies. And after doing the legwork with them, they said, look, we can't take this and put it, we can put it in a mutual fund, but we can't put it in an ETF because we think we don't want to disclose these trades, that transparency element is a problem.

And some of the things we trade, we don't think the market makers are going to feel comfortable hedging. So for example, one of the things you'll see is some of the CTA-based ETFs really focus on making sure the futures they're trading are European and American listed. You don't see Asian listed futures in those ETFs. And that's because it's really hard for the market makers to try to hedge a product when the futures aren't trading. When they're asleep. You don't need that in the mutual fund because the mutual fund only needs to strike NAB once a day. You're not trading an intraday. And so the types of products that can get wrapped in ETFs have to consider the entire ecosystem in which against each other and the market maker is matching us. No, right? Like the market maker is actually buying and selling from us. Yeah. I mean, right. So he's wearing the risks until he can

offload that risk. Yeah. And a lot of this comes down to how much liquidity is on the order book, how much liquidity will the market maker actually ingest? A lot of standing liquidity has more to do with how much is the average volume trading. The market makers, they have capacity constraints. If your ETF is trading $500,000, it's not like they're going to leave $20 million just sitting on the order book. They're going to have substantially less. They don't want to tie up capital. But the real measure is what's the bid-ask spread? The tighter that is, it means the market maker is more comfortable immediately hedging whatever they take. Because yes, in theory, in an ETF that's trading a ton, you might get players matching off against each other. But the reality is almost every time you're probably selling to

a market maker, who's then maybe turning around and immediately flipping it to another market participant, or if they're acquiring inventory, they're going to have to hedge it. Get your thoughts on AI while we have you in chat, GPT, and we'll start with good, bad, indifferent for society as a whole? This is an area that I can only speculate. I think it's really cool. I think there's some things about it that are really cool that have nothing to do with the potential productivity lift that we can all get. I think we're already seeing

ways in which it can get integrated into music and help people write things i think there's some things that are incredibly scary like we're already hearing about scam calls where voices are being replicated i said to my wife do you ever get a call that i'm i'm on the other line on the phone saying i've been kidnapped just hang up. My voice is too out there. Assume it's fake. They'll start with the people at podcasts and around. Yeah. I mean, come up with a code word for sure. One of the things that I am kind of endlessly fascinated with though, is the idea that all of these things are being trained up to a point in time, which means once it's trained, there's this snapshot that's occurred of whether the training is right or wrong. It

sort of incorporates all the bias within what it's learned up to that point, which means to me, in 100, 200, 300 years, there might be the opportunity to go back and start to query that data set at a point in time to understand the bias as it truly existed then versus the way it's been written in history books, right? And I think it's a very cool concept of being able to encapsulate everything known up to a point and be able to interact with that data and query it to really understand what did people understand? What was the bias at the time that was ingested within the data? And how did that evolve over multi-hundred year periods that's a cool insight right like it's basically i can talk to this medieval times lord and be like hey what's it like versus the way the

history books were rewritten right um and what do you think does it do anything for you modeling portfolio type work like that's what's weird in my space because it's like it's like been here in our quant world for a long time yeah i mean machine learning techniques all right are just really a lot of them are just sophisticated statistical techniques i don't see the ai of chat gpt being particularly useful especially if it's not fully online right yeah you need something that's constantly ingesting the new data if it's trained up to last december that doesn't really help me. It's cool for me to think about like, one, your kid probably might have this always on tutor,

right? That has infinite patience, infinite empathy can I don't quite understand what you're saying. Okay, tell it to me like I'm Dora the Explorer, whatever, like, right, you can switch it all up. And it just the interactive teaching, i think is a fascinating element though i'm sure you've seen the same things i've seen which are um these hallucinations i guess is what they're calling them like they that it's literally making up sources so you have to learn a new skill will be identifying don't hallucinate right ai that is lying to you it's not intentionally lying to you but like you know that that famous study not the famous study the famous case where a lawyer tried to use it to write a brief and then it was citing briefs that didn't actually exist yeah and they

threw it out and the judge asked him about it and then he tried to use it to make the briefs that didn't exist i mean just i mean compounding integrity issues here obviously stop digging the hole right but the point is like i heard about this really cool um homework that a professor gave their students that said i actually want you to use chat gpt to write an essay about something you know a ton about and then show me all the ways in which chat gpt is wrong oh that's good that's a good teacher there and then then you think RAs will start using it? Like, hey, call our chat GPT line to talk about your portfolio or whatnot. Like instead of,

like just imagine the scale you can get. Like, I know you're worried I'm going to give you a call, but it's not really them calling. It's just because all you say on those calls is a bunch of, yeah, it'll be fine long term. So I'm chatting with Dave Natattig about this from Betify. Because I think his title is like futurist, basically. Yeah, that might be his exact title.

He thinks about this stuff all the time. have a survey among advisors to understand what part of their job they think is best replaced by chat GPT or other sort of AI. Is it the conversation with clients? Is it responding to emails? Is it dealing with some compliance stuff? what part of their business can actually be enhanced? I would wager that they, most advisors would feel like that human element can't be replaced. You need to talk, when someone needs to talk to someone,

they need to talk to someone, right? But if someone's emailing in asking about an account paperwork thing, is that something that can be automatically replied by chat gpt 99.9 of the time maybe that 0.1 of the time it gets it wrong though well that sounds like a big risk to me yeah you know lost that client and where and where's the how do you talk about being a fiduciary in some of these cases yeah who's the fiduciary who's the fiduciary well

for sure i think it'll be this email right there's going to be new footers of like this email was ai assisted or something right all right something new gonna try with you oh the guinea pig the guinea pig of some uh i don't know what he caught word association yeah i'm gonna give you some words this is some psychological stuff here this is right then we're gonna do rorschach test this is a verbal podcast rorschach test um which i had to take as a young child of divorced parents i actually had to go to a psychologist and i was messing with the lady i'm like uh that's my dad beating me up that's my mom locking me in the car. And finally, after like six questions, the lady's like, oh, I see what

you're doing. Stop it. And then I'm like, fine. It all looks basically like spaghetti to me. I'm hungry. Anyway, that was my personal Rorschach. So word association game. You ready? Let's do it. All right. Jason Buck. Handsome. Is he still there? Can you hear me? No. Yeah. Tampa. Tampa. Hot. Boston. Brick.

Brick. Interesting. We're on a location basis here. Cayman. Swimming. Rick. Interesting. We're on a location basis here. Cayman. Cayman. Swimming. Did you ever switch to the Cayman pronunciation? I gotta tell you, it took me forever to try to figure out how to pronounce it.

The way I tried to get it, listening to the radio, is it sounds like caveman. So it's Cayman. Cayman. Cayman. And no, I never got there uh podcasting exhausting oh this is my favorite the next one clients uh clients clients

this is a tough one i don't i'm trying to think of a single word. I have amazing clients, by and large, who ask me really thoughtful questions. So the word I want to use is something like, it's about the feedback they give me. They push me to look and research into new areas. It's not inspiring, but it's something along those lines. Inspireful Compliance Necessary evil Good NFTs

Underrated Still I was thinking you might go burn marks but still underrated I think the majority of the space is complete garbage. I think there is some really interesting technology within the NFT space that gets overlooked because it's all people are going. It's just a bunch of annoying monkey JPEGs.

But I think there's some really cool stuff that can happen there. I almost bought a five thousand dollar lightsaber jpeg back in the day i might have to go look at that it's probably five dollars now probably very cheap now yeah um gold complex bmw BMW uh falling apart is what mine is I've got a 15 year old beamer that I love and it has just followed me everywhere and it is falling apart at the moment I saw it when we did our illegal rum transaction that's right on the side of the highway I had to do a lot of work to it recently and the mechanic just kept calling me he's like like, well, I don't know if you know you have this problem. I was like, I didn't know it explicitly,

but yes, just give me the laundry list at the end. What do I have? Diapers. Diapers. Positive trend. Expensive. Rucksack.

Shoulder pain. I can't believe you're posting that stuff. We're out at like 4 a.m. with the rucksack. Yeah. It's unbelievable. It was my last. It definitely was my last time. I mean, this was one of those. I knew a kid was on the horizon.

It was going to be the last time I could really pursue, at least in the short term, like a real leaning in physical challenge. And so it was this, look, it was like 28 mile ruck march through the Hills of Pennsylvania. I really wanted to push myself. So yeah, I think I spent four or five months training for it. It was, um, and then I haven't picked up the backpack again and i don't ever want to how many pounds uh ended up being just around 80 pounds oh you're a good one to ask this did you see mark zuckerberg's stats on his uh the marine test what do they call that though what i gotta ask what what do you think is happening from a PR perspective here? There has been a strong change in Mark's PR team pushing his jujitsu.

He's watching those pictures of Bezos. I want to be jacked and cool. He's like, oh, that's right. I could do something with all this money to prolong my life forever. I think that's probably it. They were like, oh, he was two minutes off the world record.

Yeah. Last one, rum. Underrated. I think rum is massively underrated. Actually, another word for that might be dispersion. That is a category of massive dispersion because there's really no rules when it comes to rum. But I think you can find some incredibly high quality rums for incredibly cheap relative to bourbon and scotch. And people who are bourbon drinkers, there's certain rums I would steer them towards.

Scotch drinkers, there's certain rums I would steer them towards. If you're like me and you just naturally have a sweet tooth like there's other rums i just i just think they're really underrated um by the way i want to change my answer on jason he is incredibly handsome i thought he was just there and could hear me potentially though i realize you got headphones on uh jason is this isn't a single word jason is the most interesting person i've ever met potentially for sure uh you know i don't like i'm not going through every single person I've ever met, potentially. For sure. I'm not going through every single person I've ever met, but Jason is hands down one of the most interesting people

I've ever met, if you can actually get him to talk about himself. Yes, that's the key. I keep trying to break him because I give him clothes every time. So I'm like, here's a t-shirt, here's this, but he's known for only living and having a single bag of clothes.

Lives out of a single bag somehow pulls it off incredibly stylish has lived all over the Americas North and South America multiple lives within entrepreneurship you know from from high highs to low lows and somehow just knows a little bit about everything in a very annoying way. Yeah, I said that. He's like, well, I know about this 99% more than everyone else in the world, but nowhere close to the people who are in that top 1%. I'm like, yes, but when you know that about 99% of the stuff, that's not true. I'll be like, I know this obscure fact about aussie rules football and

he'll be like well let me tell you where that came from going back to the 1800s i'm like how do you know this yeah what too well ran yeah um all right well i'm looking forward to your rum success do you think it'll be like the next right all these people made a billion dollars off their tequila branding you think that's coming next with celebrities no i think what's more likely is that i will lose a lot of money yeah trying to launch rum yeah i mean uh this is this is listen i am not ryan reynolds i am not uh the rock i do not have the following that is required to launch an actual liquor brand as i keep trying to explain to jason but god damn it if we won't try right, man, this has been fun. Thank you for having me. Everyone go follow Corey, Twitter,

listen to flirting with models podcasts and listen to the pirates of finance. What else, what else you got? That's it. That's it. How'd I do? That's it. Appreciate it as always, my friend. All right. Good talking to you. Okay, that's it for the pod. Thanks to Corey. Thanks to RCM for supporting.

Thanks to Jeff Berger for producing. We'll see you next week with Jem, Zed, and Luke. Peace. You've been listening to The Derivative. Links from this episode will be in the episode description of this channel. Follow us on Twitter at RCM Alts and visit our website to read our blog or subscribe to our newsletter at rcmalts.com. If you liked our show, introduce a friend and show them how to subscribe. And be sure to leave comments. We'd love to hear from you.

This podcast is provided for informational purposes only and should not be relied upon as legal, business, investment, or tax advice. All opinions expressed by podcast participants are solely their own opinions and do not necessarily reflect the opinions of RCM Alternatives, their affiliates, or companies featured. Due to industry regulations, participants on this podcast are instructed not to make specific trade recommendations, nor reference past or potential profits. And listeners are reminded that managed futures, commodity trading and other alternative investments are complex and carry a risk of substantial losses. As such, they are not suitable for all investors.