The Derivative - Noodling on Ensembles, Trend, & Convexity with Newfound’s Corey Hoffstein

Episode Date: June 18, 2020

Our guest today is the Co-founder & CEO of Newfound Research, host of the popular podcast Flirting with Models, and FinTwit “influencer” – Corey Hoffstein. In this episode, we’re talking with Corey about the impact of “data”, the juvenile meaning behind Flirting with Models, 1000s of flavors of trend, hitting the gym, the founding of Newfound, suicides in Ithaca, sticking to your strategy through investor unrest, alts vs equities firm, the value of managed futures, implementing an ensemble approach, the mismatch between expected and market convexity, is trend actually dead,  applying AI and machine learning – in alts & the world, and Corey’s favorite Boston restaurants. Follow along with Corey on LinkedIn& Twitter. Check out Newfound Research’s website & blog, and listen to Corey on his podcast, Flirting with Models. And last but not least, don't forget to subscribe to The Derivative, and follow us on Twitter, 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

Thanks for listening to The Derivative. 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. 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. It's very, very coarsely tied to options theory, that when you are trend following, you are delta replicating a straddle. And so when markets go up, you get more long. As they go down, you get more short.

And that creates a very mechanical convexity. And just like options hedgers, the ability to perfectly replicate that delta is going to create slippage in your P&L. You're going to see whipsaw. But it is sort of by definition mechanically convex to those price movements. So that's one part of it,

which is you you know, the payoff profile should look like a parabola. Okay. Welcome back to another episode of The Derivative. I'm your host, Jeff Malek, and I think we're going to have fun with today's guest, whose skill set stretches across research, podcasting, investment strategy, and running an ETF. We're joined today by Corey Hofstein, co-founder and chief investment officer of Newfound Research and host of the popular investing podcast, Flirting with Models. Welcome, Corey. You can't say that name without me chuckling a little,

Flirting With Models. It is a pleasure to be here. Thank you. This is going to be fun. So how did that name come to be? That is one of the best pod names out there for sure. I think that one came to me in perhaps my more juvenile years. I ran a blog a long time ago that had the name Flirting With Models. And then when I started my firm, I think it just carried over to our firm blog. And then when it came to doing a podcast, it was just sort of brand at that point. I don't know. It's one of those, we do all sorts of quantitative stuff and we like to flirt with those quantitative models. And my wife thought it was funny. So I have permission, I guess. We have uh bastion balesta deepfield he was on the pod once telling a story about he was on the couch in his house listening to your pod and the wife like came over and saw the phone and like what do you what is this flirting with models what are you doing

it's like it's work it's work it's research i swear it's work well i i had my wife i was playing my podcast once listening to something on repeat from a guest and she walked in and she's like, who are these nerds? And I was like, Oh, I can't let her know this is my podcast. Nerd. My daughter made her listen to a few of ours and we get on a drive and she's like halfway through. She's like, can we read a book or do something different? Anything else? Anything else? One day you'll appreciate it. One day. So where in the world are you right now you're usually in la but you're usually in la outside of boston right now on cape cod okay and you're on cape cod nice what part uh and around the born area okay so katam it technically

sort of north falmouth area and uh and so normally in la what part of LA? So my wife and I actually sort of were bi-coastal for a very long time. I lived in Boston for seven, eight years. She was out in the LA area and we went back and forth, had a place in West Hollywood at one point. After we got married, she gave me the ultimatum that we were moving to LA. So I said, my ultimatum was fine. We got to be on the beach if you're going to drag me to LA, which I mean, drag to LA, who gets dragged to LA it's a nice place to be yeah so we ended up

choosing to live right on the Santa Monica Venice line so we can go north to Santa Monica where it's a little more put together we can go south to Venice where it's a little grungier a little more fun a little more artsy and that was uh I got a buddy out in Santa Monica. That was amazing to see the riots and all the action. We're recording this on June 3rd. So right in the middle of all this action. So in Santa Monica, which seems not the place you'd typically see that. I was not expecting that in Santa Monica, but there's a lot of action going on everywhere. I think there's a lot of places around the country where you're seeing that. So I can't be surprised by anything at this point, right? Yeah. Hopefully your, your house is still there when you get back.

Well, I don't expect my car to be there to be quite honest, but at this point it's a 10 or 12 year old car. So if it's on fire and I can collect some insurance that that might be worth more than the car at this point. So that wouldn't be a bad thing. Done. And I see you were a fellow uh upstate new york college peeps so i went to school union college in skaneo yeah i know union absolutely that's the only other people who know it are people from upstate yeah yeah rpi all those so i actually met my wife at rit i went there my freshman year. And then within three months knew I didn't want to be there anymore, transferred out and ended up at Cornell. Nice. And you're from Boston area originally? Yeah, originally from outside of Worcester, a really small town called Southboro.

Worcester, which is spelled like for people not from Boston, it's Worchester, right? Exactly. Well, that's like everything in this area. Nothing sounds like the way it's spelled. And so got out of Boston and went to, so RIT, was it Rochester Institute of Technology? Yeah, Rochester Institute of Technology. So I had this idea when I was younger that I was going to program video games for a living. And I actually, very early on, my father was an entrepreneur in the technology space, was always around computers. And for whatever reason, when I was 12 or 13, taught myself how to program.

And a lot of my original things that I was programming was I was using things like Q Basic to write really simple games. I just had this idea that I wanted to program video games. Ended up continuing to do that as a hobby all throughout high school, as you can imagine, very popular on a Friday night. And I decided I wanted to attend school somewhere where I could pursue computer science for my undergraduate. Went to RIT, very quickly realized this was not a career I actually wanted, that I did

not, I loved it as a hobby, but the idea of sitting behind a computer all day writing code was just not for me. But still wanted to finish my computer science degree, but ultimately felt like I had to pursue a different direction. So I ended up leaving RIT, going to Cornell, finishing my computer science degree. And while at Cornell, really fell in love with the concept in the world of quant investing. I don't think I fully realized the full scope of what everything was that was quant, but I knew I wanted to pursue my sort of MFE equivalent. So I applied to Carnegie Mellon's master's computational finance program,

which is one of the oldest MFE programs was fortunate to get accepted. MFE's master's of financial engineering. Exactly. Exactly. So most of those programs are going to be sort of multidisciplinary programs that are going to be a mix between statistics and probability, finance, computer science, and stochastic calculus is sort of its own coursework. A lot of it at the time when I went 2010, 2011 was still preparing you very heavily for sell side, where you would probably go either be a desk quant or sales and trading. A lot of it was

interest rate derivatives, credit derivatives. I think the demand has shifted to the buy side at this point. And so the coursework has changed subtly. But for the most part, it's all about pricing derivatives, understanding risk and that sort of thing. You're a little younger than I am. So I don't feel like that was around back. For sure, it wasn't around at Union College, but it seems this whole MFE concept is a little newer. It's definitely picked up. I think the original program started in the late 90s, mid 90s, so that they were there. But again, so I think Carnegie Mellon was one of the first. It was called the

Computational Finance Program. No one else adopted that name. So no one has any idea what the degree means. But they all sort of chose this financial engineering nomenclature. And every program is a little different. I think, again, people are still trying to figure out what it means to get a degree in financial engineering. The basics are there. But what each school tends to focus on seems to be a little bit different. Yeah. And we just on just on a couple pods ago we had we're working with a group at a university of illinois and the ucla mfe um graduates through our university program and they're helping uh basically build a factor model for sports team valuations yep so it's super cool to like hear all these our world stuff in the investing world but it's for sports team valuations yep so it's super cool to like hear all these our world stuff in the investing

world but it's for sports team valuations and they were coming up with you know fans in the stadium doesn't really matter it's more of like the if you have a star player if you have this and television rights and all this stuff so i feel like that everything's just blurring between quant and non-quant right like it's going's going across all lines. I think you're finding that just the world of data is becoming very important everywhere. And the skills of being able to analyze and think thoughtfully about utilizing data, model it, use it to forecast, obviously very applicable in finance, but very applicable in everything else in the world. So the skills are becoming transferable. It's a question of, do you have the niche knowledge then to put it into work? For sure. The, uh, my son, I was toying around,

he was getting bored with his math at school. So I was trying to find a statistics, uh, but more of like data analysis course in LA actually is trying to put this into the schools and there's a free program out there. And I was, it was a little too much of a project for me to implement at home. But I liked the idea of like, cause they don't teach data analysis at grade school level for sure. And not even high school level mostly.

I feel like that's a skill people need moving forward for sure. I feel the same way about programming as a skill. I'm amazed that there isn't more emphasis on programming as a skill in high school statistics, understanding in high school. I mean, there's, there's people who are truly probability illiterate and understanding probability is just such a big part of life to me to not understand how things can play out. I think it gives you a more narrow view of what the world could be.

For sure. The And so moving on, I was hearing you on Resolve's happy hour the other day of who can lift more, Nassim Taleb or you? What's that all about? I don't know. I honestly don't know where that came from. I think I've developed this, um, a bit of a reputation for being a gym rat. It's funny on, on places like Twitter where I'm pretty active and there's a really big community. Um, a lot of people follow me for the research that I post and sort of the conversations, but I do post, you know, I'm really into fitness. So I love running. I love snowboarding.

I love going to the gym when the gyms were open. If I hit a PR, I would push it up there. And a bunch of friends in the industry are also really into that stuff. So we would share and congratulate each other. And there's some people who really hate it. A lot of people would say, Hey, I don't follow you to see you deadlifting in the gym, you know, and tune me out is my view. But I think, I think someone was sort of, I mean, Nassim is well known for talking about his deadlifting. So I think someone was having a little I mean, Nassim is well known for talking about his deadlifting. So I think someone was having a little bit of fun poking fun at himself.

But to me, so I used to do competitive lifting in high school in Florida. Yeah. So it was deadlift and bench press. And then I moved over to like Olympic clean and jerk and all that stuff. But it was fun for a while. But yeah, I'm a shadow of that former self. Well, the irony for me is I am amazingly weak for someone my size I'm like six four two hundred plus pounds I should be a lot

stronger and I am pathetically weak so a lot of people think I'm bragging when I'm showing my lifts and anyone who knows anything about lifting I mean you would just sit there and go like this is such a joke for someone that big so yeah I think I in, uh, I think it was my senior year at the competition. I lost to a, uh, a woman or a girl, I guess we were boys and girls in high school, but, uh, who out deadlifted me. So that was the end of my competitive lifting career. Those Olympic lifts take a lot of skill. It's amazing when, when you see someone quickness, yeah. Who really knows the technique, technique how how much less effort they have to put in with the correct technique i'm sure it's like all things in life but it is when you

see someone do it correctly it's like i didn't realize how poor i was at doing this until i saw someone who's truly skilled um yeah agreed and so when you you left college and went straight into newfound you founded it right out of college yeah so I have a bit of a weird story with Newfound. So the name Newfound is actually a lake that my family used to go visit in New Hampshire. And so when I was an undergrad, I knew I wanted to go into this field of quantitative finance. I did a couple internships, mostly around, and I didn't even realize it at the time, but I was really doing factor scoring, which I had no idea, but I was working with an SMA manager who was building a portfolio of single name equities in the dividend growth space. And I was identifying characteristics that were predictive

of future one, two, three year cross-sectional returns. Had no idea that was factor investing, but that's what I was doing. Um, and along with that, I, I ended up, um, working on a lot of models that were sort of more macroeconomic in nature, tilted more towards market risk. I just, I tend to tilt towards a more risk view of things than a return view of things. At the time I was also one of, I started two internships. The second internship was with my father's financial advisor. What I would do is interview managers that would come in. I remember in summer 2007, there was one manager who came in. It was a small cap value manager, ran a mutual fund. and I, before the meeting started, asked him what he thought of the market. And he just gave me the most bearish prognostication I'd ever heard. I mean, I was young, right? But I, but it was just so bearish. And it was in 07? It was in 07. So he was right. And I was really blown away. And I was like, so what are you going to do about this, man? And he was like, well, by mandate, I have to be 95% invested. And my job as a PM is to create the best

small cap value exposure I can, and look, at the end of the day, I don't even know who's invested in my fund. It's really up to the individual and their advisor to determine if the risk of my fund is right for them. I said, that's a pretty thoughtful answer. I get that. So after the meeting, I talked to my father's financial advisor, and I say, you know, sort of tell him what happened. I said, what do you think of that? And he said, I think that's crazy. He's supposed to be an expert in small cap value. If he doesn't think the risk is right, he shouldn't be investing the money. If he, right, if the outlook isn't good, how am I supposed to know if it's good time or not for small cap value? And I said, well, that's also a really good point. And all I

saw was two people pointing the finger at each other. And, you know, at the time I had my, pretty much my life savings invested in the market. And I said, Oh, no one's managing my risk here. And to me, risk was just preservation of capital. And so I went down this whole rabbit hole of, uh, thinking about building models that could help manage risk and ultimately ended up discovering trend following as being one of the core techniques there. Next summer, I end up getting introduced to another asset manager who sees what I'm working on, gets interested in licensing the model that I built, sort of these buy and sell trend following signals. I'm a broke college student planning on going to grad school. I say,

great, let's do it. He ends up offering me basis points. Didn't even know what a basis point was, but that's how we were going to do it. Yeah, sure. Whatever. Again, if I can get $500 to buy some beer, that would be fantastic. And ended up creating a company called Newfound Research because it was named after a late guy used to go visit and was very fond of, and we were providing research, figured the company would shut down in a year, uh, ended up going to grad school. And at that time, the manager that I has licensing the signals to the data ended up getting a sub advisory agreement and really starting to grow. And I saw this appetite for tactical ETF strategies out there in the market. And I said, you know, I can always come back to work at a bank on Wall Street, the opportunities to have a business that is now cashflow positive. And I see a need in the

marketplace and to try to make this my own, you don't always get that opportunity. And so after grad school, I said, I'm just going to try this entrepreneurial thing. And I've been very fortunate that it's worked out ever since. And what what timeframe is that? What year is this? Oh, nine ish? Yeah. So the company was founded in August 2008. Yeah. And then I got out of grad school in December 2010. Yeah. And so you see all those studies of people who graduated in that time period had a heck of a time getting jobs like there it is way higher than uh all else being held equal way higher than uh other people so my i remember at my college commencement in spring 2009 at cornell the one of the speakers said you know you should be really really proud

of yourselves 30 of you have jobs and i was like like, oof. 30% of us coming from an Ivy League school have jobs since spring 2009. Like, I can't imagine how the rest of the world is doing. So it was pretty eye-opening from the, again, like the, your little bubble that you live in, seeing like this is not a good economy. Isn't Cornell known, like the suicide capital of the college world? That bridge and the gorge. What is that all about? Yeah, yeah. Fix my light real quick.

Go ahead. I don't know if it's actually true or not. I mean, that's always been the rumor, but there are these beautiful, beautiful natural gorges in the Ithaca area. And basically, as it goes, Cornell is such a difficult school, especially the engineering college. And you have these gorges that are just such an easy way for a student who's suffering depression or having issues coping for them to take their own life.

And so it's when the means is there, I guess it makes it easier for it to actually happen. And so there's, I guess, just a higher suicide rate at Cornell, or at least that's at least what I heard going through Cornell. Yeah, the urban urban legend. Yeah, I never actually checked the numbers there. Every school thinks it's the hardest school. So for sure. And you guys had a heck of a basketball team right around that time, too, right? Didn't they win a few tournament games? Yeah, there were, if I remember correctly, and I wasn't really that into sports at the time. So

I think they were really well known for their hockey team that by the time I got there was sort of going downhill a little bit. The basketball team went on, I think a great run. It was either my junior or senior year. But again, it's a little weird because they played in the Ivy League. So it's sort of like getting to sort of March Madness from the Ivy League, I think is easier than other leagues, of course. But there's still some decent teams that they were competing against and did pretty well. And then their lacrosse team was quite good. My senior spring, I think they went all the way to the championship and were playing

Syracuse, which was the school my brother was going to. And so we had an absolute heck of a battle during that game. But yeah, I remember going to a game in the Syracuse Dome, a lacrosse game as a fan, and there were probably 20,000 people there. Yeah, it's I mean, it's a fun game. It's a fun game. And the Syracuse students love to go to those games. One of my best friends went to Syracuse and I would visit him all the time to go watch basketball or lacrosse. And it's just a blast. So we're going to switch gears a little bit and talk about the firm and your strategy. So usually we ask managers to give the elevator pitch of their

strategy, but I don't know if that really applies to you and what you guys are doing with a bunch of different strategies and doing a lot of bespoke work. So maybe give us the elevator pitch on the firm. The elevator pitch on the firm is that we believe that first and foremost, investors need to focus on risk. And so what we ultimately aim to do is provide risk managed strategies that allow them to meaningfully participate in equity market growth and try to avoid those severe and prolonged drawdowns that can be really adverse to their ability to achieve their financial goals. For us, it really does tie back to that ultimate, what do you need this money for? And how are you trying to use it? And thinking about how all this fits together. But for a lot of people who are near or entering

retirement, there's a tremendous amount of sequence risk that they face. And so that if we can help reduce volatility without necessarily creating as much reliance on fixed income as you've traditionally seen, we think that we can help them navigate that sequence risk problem without necessarily creating a huge fixed drag that comes in the lower expected returns that are related to fixed income. And so the way in which we do that is by trying to apply quantitative signals to asset allocation and tilt more towards equities when we think there are stronger signals that are supporting equity market growth and tilt away from equities when we see the opposite situation. And so there's several strategies inside of that? Yeah. So we have a number of different strategies that we offer in SMA and mutual fund format,

whether it's a tilt towards global equities, US equities, a sector-based framework, a factor-based framework. But the same overarching thesis is very similar. It's this protect and participate mandate that we're seeking to hit. I like that. And you get a lot of the, or I'll just say, you consider yourselves an alts firm or an equities firm or neither? You know, we probably fall more towards the alts than we do equities. But alts, in my mind, tend to truly try to focus on creating an idiosyncratic return

stream. Our view is we're trying to ultimately harvest the equity risk premium over the long run. But we're trying to recognize that there are certain market environments where we think we can create an edge via signals like trend following, because if we can remove ourselves from the equation through liquid markets, either through explicit hedges or actually just exiting the market entirely, and we can avoid those significant prolonged drawdowns that occur, then we can actually come out ahead in the full market cycle. And as naive as it sounds to say, for example, why do you think you can just use trend following to get out of equity markets?

We think that there are basically market structure pressures and liquidity needs that occur during bear markets that if you can be a nimble enough player and get out, you can tend to see market continuation. And so we think there's a lot of potential advantage in being nimble in your asset allocation in ways that a lot of other market participants can't be. Yeah. And I'd say what differentiates you, you guys from other alts firms I see is that concept of participation, right? So, and I've been banging this drum for a long time of, especially in the managed futures and trend following space, when they've been struggling for nine plus years,

I tell my son he's 11 that we've been in drawdown his whole life. You know, of what do they do? Like, do you stick to your knitting? As an investor, I'd love to have just pure premia. Like I'm hiring you just for this sole purpose, but they have families, they have kids to feed, they have, right? So I'm the best model. If you came to them as the only investment advisor on earth, they'd say, okay, you should mix some equities. You should mix this. So it boils down to all these asset management firms of, okay, do I give them the whole enchilada of what I think is best in a, in a portfolio or just the slice of premium that they need? Well, and it's an interesting question about how products are distributed in our industry as well,

which is there's very much the tilt towards, hey, focus on your premium. Don't give us the whole thing because we want to pick the best managers for each slice and we know how we want to build and optimize the portfolio. And so then as a manager, you are inherently restricted to providing a particular style, right? And they don't want style drift. But then as soon as your style is out of favor, the question is, well, what are you doing to fix this? Or we're leaving? And you say, well, you hired me for this purpose. If I fix it, then it becomes an issue of I can't get hired for this purpose anymore. So it's very, it puts you between a rock and a hard place. And it's an interesting question about how we think about, as an industry, product distribution and are we doing it really the right way?

Yeah, and I think you see a lot of the biggest of the big hedge fund firms aren't just a single fund. They've got flavor A, flavor B, levered this with equity. So they're doing this internally. They probably got lucky with their single model that raised enough assets that allowed them to spread out versus you got startup guys who, oh, my mandate's to stick to that strategy and they might not make it. Well, it's hard to do any one strategy well as a small firm if you're distracted by other strategies. But when you think about an asset management firm and how it generates its revenue, it's ultimately generating carry on its sort of

asset, dollar weighted asset exposure. So if you can be an asset management firm, that's large enough that you have, I've got an equity strategy over here and a, and a trend following strategy over there and a fixed income strategy. And altogether, I look like this great 60, 40 mix. Well, then you might be able to survive just about any sort of market cycle that comes at you. The BlackRock model, right? Exactly. There's so many strategies they'll never have a drawdown in. Exactly.

But to do that as a small startup asset manager, I think is almost, you got to hit that escape velocity with your one strategy first before you can start building that out. Totally agree. And circle back to your sort of elevator pitch and your focus on risk. So what does that mean to you, risk? So, right, there's reams of studies and papers of if you're looking at volatility, if you're looking at drawdown, duration of drawdown. So what do you view as, quote unquote, risk in that scenario? So I start with this higher framework where I just basically say investors really suffer one of two risks

at any given point. They have the risk of failing fast and the risk of failing slow. So in my framework, the risk of failing slow is basically you have far dated liabilities that unless you invest aggressively enough and try to harvest the risk premium, you could fail to ultimately meet them. So maybe you're a young investor who's saving for future retirement. Unless you save sufficiently and invest aggressively enough, you're ultimately going to lose to inflation over time. So that's the risk of failing slowly. You need to be aggressive. The risk of failing fast is the exact opposite, which is you are probably currently in withdrawal of your portfolio.

You've taken all your human capital and translated it into investment capital over time. You now want to live on that investment capital. And a sudden and severe drawdown creates permanent impairment. So that's the risk of failing slowly. What's really interesting thinking about this is... Go ahead. You flipped them, but yeah. Oh, sorry. The second one was quickly was risk of failing quickly. Sorry. Yes.

What's really interesting is this is very naturally like the progression of a young investor to an old investor, but you can fit an endowment or an institution in this framework by saying they're sort of right in the middle, right? An endowment has cashflow that they're taking out each and every year. So they do have a risk of failing fast. If they have too large a drawdown, it creates a very large issue for them and being able to create the cashflow and sustain the cashflow. But they also have hundred year mandates. Right. They want the money to be there for hundreds of years. And so they have the risk of failing slow. And so what you find is, well, risk of failing

slow, you need to be aggressive, risk of failing fast, you need to be conservative. Endowments find themselves somewhere in the middle and they find themselves in a balanced portfolio. We tend to think that risk of failing fast, risk of failing slow, excuse me, is just, you just got to bear risk. You just got to buy equities. If you can have some levered portfolio of diversified assets, even better. It's the risk of failing fast that we say, let's focus on that. A, because from a dollar weighted perspective, it's a bigger problem in the industry. And B, we can explicitly measure it as drawdown risk. So to come back to your original question, it's really that peak to trough drawdown risk,

because when it comes to investors starting to make withdrawals from their portfolio, those withdrawals that are normally fixed dollar tend to become a larger proportion of their capital as that drawdown gets deeper. And so for us trying to mitigate the size of that drawdown becomes really important. And so no nod to the drawdown duration there? Because you, right, if the flash crash, whenever that was, 2010, steep drawdown, but didn't even register on an end-of-day basis. So there's things, that's an extreme example, but you could argue that any asset class that snaps back quickly, I'm not as concerned about the drawdown. Counter-argument to that would be, well, you never know it's going to snap back that quickly in the future no you're absolutely spot on i mean there is there is certainly a uh both depth and timing aspect of it you could almost tie it up in one of the one of the metrics we really like to look at is the ulcer index right which is going to look at both the frequency and depth of drawdowns and ideally

what you're trying to do is reduce that ulcer index where you want to have less frequent and more shallow drawdowns. You want to continue to be making new equity curve highs more frequently. Yeah. And I think bringing it back to Managed Futures has had, right, their ulcer index has been painful. Yes. But not from a magnitude, but just the time. Right. How long you've been underwater. Absolutely. How long and how long can I just keep the death of right how long you've been underwater absolutely how long and how long can i just keep the death of a thousand cuts type of deal so let's get into that with uh trend following a little bit uh so it seems like you're still a big proponent of trend following, but I'm taking that you mean just on equities or as a kind of quote unquote managed futures global portfolio kind of look.

So I like both. Um, a lot of the, the, our research focus and the way we implement is explicitly on equities. My view there is simply that investors, most investors aren't working with a risk parity portfolio. So if't working with a risk parity portfolio. So if you started with a risk parity portfolio and everyone started there, I would say, then yes, we probably want to use managed futures as a way of managing that risk. But given that most investors we end up dealing with have something that's more like a 60-40, the majority of their risk is coming from equity. And so for us, we want to more explicitly focus on trend following on equities to try to hedge that very explicit source of risk in their portfolio.

Because what we're ultimately trying to do is create a negatively correlated return stream, whereas trend following for, or at least managed futures, diversified trend following for all the claims of crisis alpha, it's more of a decorrelated strategy. Yeah. And unpack what you mean real quickly on if they started at risk parity, just if their bond risk was equal to their equity risk, et cetera. Just even a truly more diversified core that is more, so your bond risk is equal to your equity risk. You have commodities in there. You might have currency exposure, whatever it is,

you have a strategy that actually more closely resembles what a managed future strategy would look like if all the trends were positive. Got it. And you can sort of then think of it as adding from the long side. Yeah, on the long side. And then if you were to add managed futures on top of it, it's very naturally a one for one connection of as this managed as a certain contract goes negative, it's becoming a hedge for my passive book. What we're ultimately trying to do is recognize investors very naturally have a long-only stock bond portfolio. And not only that, they have a long-only stock bond portfolio that they, through rebalancing, create a concave payoff profile, that they're constantly building

a mean reversionary trade into their portfolio. And so by doing trend following on equities, we're allowing them to inject not only a potential risk management system for the largest risk in their portfolio, but actually inject payoff diversification, that you're creating a convex return profile. And so you're creating diversification sort of in a, not in the what you're investing in, but how you're making those investment decisions. Yeah. I've always argued that you have, your career is sort of long equities as well, right? Like not perfectly tied to it, but you need a strong economy to get paid more money and all of that. Like you're everyone's short vol essentially in their careers and their

portfolios. So help me understand, like I'm sure you've done the research in the past and I haven't done it recently, but if you put every sector, energies, grains, currencies, fixed income, equities, trend following bucket, typically the equities is the worst performer there i don't know if that agrees with what you've seen but from what i've seen that typically the case gold is always really bad as well but yeah i think what you you would find is that bonds over the last 20 years 30 years have just been lights out one of the best sectors you could hit and then i think

you typically find that there are certain metals grains and that end up being streaky. Some are really good. Some are really bad. I can't find rhyme or reason. I think there's people I would talk to would probably argue it has to do with the participation of non-profit seeking actors in those markets, right? How much are people coming in to hedge? How badly do they need to hedge? And I haven't found a good way to really model that out, nor has anyone I've spoken to really found a good way to model that participation and tie it back to the profitability of trend. But yeah, I would say that if you look at- Guys, at standpoint, Eric Crittenton, he's got some research and work on

that yeah i actually just talked to him two days ago which is why i was top of mind he's a brilliant guy um but yeah i mean the the equity space i would say probably hasn't been the largest driver of return in managed futures but that's often because from a vol perspective gets squeezed way down relative to bonds and bonds have just had been lights out. I think it's still very interesting on a standalone basis, particular when you consider that if you really want to hedge against the equity risk in a passive book, it is going to be the most direct mapping. Again, everything else is you're basically taking a basis risk. You're saying, I'm hoping that trends emerge in gold. I'm hoping that there's a flight to safety premium in bonds that emerges

that can offset equity losses. But I think it's much safer to assume de-correlation rather than negative correlation with those trades. Yeah. I've been also hammering this for a while. And that's some of the sickness, quote unquote, that people are starting to have with managed futures of what i call there's too many ifs like oh it's going to cover your crisis if you didn't come into the crisis long equities if you didn't come in short bonds like all these ifs if palladium somehow rallies or like you know all these different path dependencies that people are just sick of i'm like like, hey, I just want it to rally when I need it to rally when the market's up. Well, going back to that sort of relative business trade, everyone is so equity-centric,

everyone is so 60-40 centric that I think if you were to step back and everyone were to wipe the slate clean and say, starting fresh, what would be the best portfolio we could build, there might be more emphasis on things like risk parity if it weren't so difficult to access from just a pure diversification argument. I think there might be more argument as to why managed futures should be a core component,

but we don't sort of live in that counterfactual world. We live in a world where everything is 60-40 and you have to deal with that relative trade and how it's being measured. And if you have five years of underperformance, it's just very hard to communicate that to an investor who maybe isn't as savvy about all the diversification benefits that it might have, could have, should have brought. Right. And there's also math to it too, right? Eventually the bleed at some point outweighs the gain like right say it's say managed futures is flat or down one percent for 20 years right is that bleed worth the worth what you saved in 2008 so at some point there there's a trade-off of like hey it wasn't worth what i gained maybe it was from an emotional bias standpoint and a ability to stick with my rest of my portfolio standpoint but from a pure

math there's definitely a point there. I've been paying insurance on my car. I've been paying insurance on my house and my apartment. I've never been robbed, never had an issue. Never had an issue with my car, never needed any of that insurance. I'm not going to go the resulting route and say that insurance wasn't worth it. So I understand what you're saying. And I think it is very hard to explain that to investors who often look at the way markets play out and see it as the only path

markets could have taken. But I think people who participate in markets day to day realize just the sheer randomness of day to day market action and how disconnected markets can be from economic fundamentals, how much they can be driven by the animal spirits, how much they can be driven by positioning of market participants in ways that I don't know if your average investor really understands. And so when I look at it and say, if I could buy managed futures and for all the potential positive qualities they have, if 10 years from now I still have a negative return, was it worth it? Well, there are so many ways in which the world could play out over 10 years.

You just have to think of it as the insurance cost. Yeah, and it's somewhat like athletes, right? A process versus outcome. Your process was sound. Everything you were doing was sound. The outcome didn't turn out like you wanted it to, but you have to separate the two. Absolutely. And so coming back to that, so just because equities is the worst performer of sectors inside of trend following, it doesn't mean it's a bad signal to use for risk on or risk off. It's kind of, or tilting one way or the other.

Absolutely. I think what's, what's particularly interesting is trying to isolate a given sector within, within a trend following mandates really hard because you have a lot of things going on, right? Not only are you talking about the actual trend signals you're looking at, but you're also looking at the relative sizing and the risk positioning. Often a lot of managed futures have a target risk profile. So what you end up finding is that when equity markets sell off, you end up in a position where your equity position within managed futures gets de-risked substantially because equity vol is going up. And so the ability for equities to actually hedge during that environment, even if you get the call correct, goes way down. Versus if you're purely just saying, I am going to one for one hedge long or short, I'm going to go long equities when there's a positive trend and short,

equal notional amount equities when there's a negative trend, you create that one for one hedge. I think there's substantial evidence that that can be a meaningful way to balance your risk in your equity book over the long run. And who is it? Is it Paul Tudor Jones with just, you'll never go broke with a 200-day moving average? Yeah, I think, I mean, I don't know if I agree you'll never go broke, you'll just go broke really, really slowly, right? And that's sort of the thing about trend following is you're going to take a lot of small losses, which can be really frustrating. I think trend following is really interesting as one of the few style premia and investment styles that I would say there's really two important aspects to consider. You know, other sort of equity long short, which I do a lot of investigation of,

you look at that and you say, it's hard to disentangle where this return is coming from, or why there should be a return. But I think a trend following, there's one aspect to it that I say trend following is a mechanically convex trading strategy. That I like to say it's very, very coarsely tied to options theory. That when you are trend following, you are delta replicating a straddle. And so when markets go up, you get more long. As they go down, you get more short. And that creates a very mechanical convexity. And just like options hedgers, the ability to perfectly

replicate that delta is going to create slippage in your P&L. You're going to see whipsaw. But it is sort of by definition mechanically convex to those price movements. So that's one part of it, which is, you know, the payoff profile should look like a parabola. How deep those losses are or how much noise there is really depends on your implementation. The second part then is, okay, if I, why would I expect this mechanical convexity to necessarily create a premium over the long run? And that's sort of the premium question related to trend. And I think that goes into, okay, are there non-profit seeking actors in the markets you're operating in? Or are there

situations where you are going to exploit market autocorrelation due to potential liquidity reasons, right? A market environment like 2008, where you may see there are forced fire sales that create autocorrelation in the market, that if you can be a nimble enough participant, you can get out, there's positive autocorrelation, trends continue, and you're able to avoid those continued sell-off and forced de-risking and margin calls and all that sort of stuff. So I think what's really interesting to me, at least about trend is I say, look, even if you don't believe the second part, that this can create a positive premium, I think there's still a really strong argument why this

can be interesting from a portfolio diversification standpoint, because of that mechanical convexity aspect of it. Right. Which I always call that of like, it's not magic. They're not guessing there's going to be a big downtrend in oil in 2018 or 14, whenever that was. They're getting into every single move and they're losing tons because a lot of them are false breakouts, but they're getting into almost every single move so they can guarantee to be in the big move that happens.

Absolutely. So if you're saying it's mechanical. We're talking all this trend following and it seems, and we mentioned the 200-day moving average, right? So if I just do a simple trend following model with the 200-day moving average, I could be looking great or I could be in a lot of trouble. So you've done a lot of work on saying, Hey, let's focus more on the signal by building an ensemble approach instead of just figuring out, okay, we're using 213 days for whatever test. So if we take a step back and say, okay, I believe in this trend

following thing. I think at the very least, I'm going to believe in this mechanical convexity. How do we measure a trend? And my buddy, Meb Faber wrote a very popular paper where he used a 10 month moving average. Gary Antonacci wrote a very popular paper in a book where he used 12 month total returns. AQR wrote some, a lot of papers with 12 month total returns and all of these models and their specifications seem to have a lot of efficacy when tested historically. You look at MEB's 10-month moving average and you apply it to a whole bunch of different markets, and it seems to do a really good job of meaningfully participating and cutting the drawdowns,

similar to the 12-month, similar to the 200-day. And so the question becomes, which one do I pick? Yeah. And the problem becomes, what do you do when your 200-day moving average signal is positive, but your 12-month total return signal is negative? How do you sort of justify choosing one model over another? And so a lot of the research we did was take a step back and say, well, why would we make a choice? If we think that all of these models have some sort of positive signal, that there's

efficacy to all of these approaches and all of these model specifications, why would I choose the 200-day over the 199-day? Is there something magical? Is there something special? And what we ultimately found was the answer was no. And so very much like we might think about diversifying a portfolio by buying a large number of stocks to get market beta and get rid of idiosyncratic equity risk, we might think about buying a large number of trend following systems that are all slightly

differently specified that together give us a more diversified ensemble of trend following systems that are all slightly differently specified that together give us a more diversified ensemble of, of trend following as a style and get rid of those idiosyncratic, uh, model specification risks. I love it. And so, but there's a point of diminishing returns there. So do you have 500 different timeframes or five different timeframes? Like that's where it gets hairy. It's definitely, yeah. I mean, I guess the question would be what you need to think about. And it looks very much sort of when you map this mathematically, it looks very much like the volatility reduction of adding equities to your portfolio, right? You have one stock and we've all sort of seen those in those intro to finance courses. One stock in your portfolio, there's a lot of vol. Then you add two and it starts dropping.

And then it keeps going until about 30, 40 stocks. And then after that, it's just very marginal vol reduction benefit. Now, when you do this with trend following systems, here's what's really interesting. It doesn't actually affect volatility all that much, which is a little, it does a little. I'm not going to get into the math reasons of that, but not the same way you see it with equities. What it actually affects is what we would call your dispersion in terminal wealth. So if you were to do your back test and say, this is a totally valid trend following system. I'm going to do my 200 day moving average, how growth of a dollar over time. And then I'm going to do another trend following system. It's 10-month growth of a dollar over time. What you find

is that there is this dispersion in what that dollar grew to. And the more and more systems you add, so if you say, instead of just using one system, I'm going to have five systems. All of that cluster of five systems, the dispersion in terminal wealth is going to go down. If you compare a whole bunch of systems that are 30 systems, you're going to end up with way less dispersion again. It's almost like a virtual fund of funds at the end of the day is what we're building. What you see is this reduction in dispersion of terminal wealth. And so to your point, how many systems do you need? Well, it's tough to say because when you start looking at what creates diversification in a trend following model, it's both the model itself.

So the way in which you're defining a trend and the sort of look back horizon, the speed that you're looking back. And so what's interesting about trend following signals is a lot of them are actually very close mathematical cousins. So when you look at prior 12 month returns, for example, all that's really saying is I'm going to equally weight the prior 12 months of daily log returns. When you do a price minus moving average, what you're actually ultimately end up doing with that model is you're just reweighting those prior returns. So all these interesting ways of trying to measure a trend really are just interesting ways of reweighting these prior returns that we're looking at.

And so what you find is, you know, pure total return is sort of this equal linear weight across all prior weights. Price minus moving average model is very front weighted. And a dual moving average crossover is very back weighted. So you can get some diversification in terms of, well, where am I weighting my views as to which prices are important? And then the question is, okay, do I want a fast model? Do I want a slow model, intermediate model? And all of those bring other diversification benefits. So I would say to you that a portfolio comprised of a couple short, couple medium,

couple long of different trend models is going to be a lot more diversified than 500 intermediate term with the same model type. So it's hard to sort of answer how much is enough because it comes down to which types of models are you including in the specifications that you're including. Yeah. And I feel a lot of the biggest trend followers in the world are definitely doing this internally, right? And they'll probably tilt more short or more long based on their kind of philosophy and their investor preferences, but they're definitely combining the models. So to that point of these guys are so big and if you're going to have all

those models, so you're just using it as a filter, right? For the most part, if you were actually trading it, it seems to me that you'd run into a granularity problem really quickly. So you don't actually, so you think of it as massive fund of virtual fund of funds, right? I mean, from a quant perspective, there's nothing that prohibits me from having 30,000 different model specifications, right?

I could have a 199 day moving average, a 200 day moving average or 201 day. I mean, there's literally from a quantitative computational perspective, very easy to do. From an actual implementation perspective, I'm not treating those as all individual models, right? That I'm going to go trade my contracts in each model because you're going to have massive netting effects. And so what you want to do is roll up all those positions together and let the different models net out. And what you end up getting is almost,

you can almost think of it as like a confidence dial that when all Voting machine right instead of having a binary light switch of a single signal You're getting something that's gonna look a lot more like a dimmer switch. That's gonna be very much based upon this idea of Model confidence when all of the models are signaling the same direction, you're going to get a much stronger signal. So from that perspective, you're not necessarily getting all these, these micro trades that are happening. Still, you are going to want to be cost and commission conscious, right? If you're running 30,000 models and one of those models keeps flicking on and off, you don't want to be making a basis point, single contract trade left and right. You're going to, you know, from an actual practical

implementation perspective, you're going to want to look at things like, okay, how much turnover does this actually create? What's the cost of the turnover? What's my true model portfolio? What's my tracking error to that true model portfolio. And you're going to run some sort of, sort of cost benefit optimization there as to when you should be making these trades and are they of substantial size that it actually is going to be impactful for your P&L net of commission. Yeah, I'm looking at it more from, say I've got 20 models and it's max confidence, all 20 are long NASDAQ or whatever, and I need 20 NASDAQ futures in my portfolio. So I would need a starting balance big enough to allow that max position. So I would actually think about it as, again, this virtual fund of funds. If I was saying,

I only ever want to have one contract, and it doesn't really work with just one contract, but bear with me. Let's say we could have partial contracts. If I just wanted to have one contract, I would be allocating 1 20th of that contract to each of the systems. Yeah. And so if all of them are on, I'm just buying one contract. If half of them are on, I'm buying half of the contract. So you're never you're not going to say I need 20 times the capital to run 20 times these systems. Again, you can almost think of yourself as a fund of funds. You're taking your money and divvying it up equally across these different models. Agreed. So we had a question on Twitter for you. I don't know when we reached out, but

all right, before I go to that question, I just want to ask you, so the, in this ensemble approach, you're considering it just for the trend following signals, or do you consider it across all the portfolio and all the strategies that you're doing so this is where you get into some really nuanced mathematics my view my view is that you should do the ensemble you should build your full portfolio so for each sort of specification model decision look back and let's just keep it a trend to keep life simple you would build what that portfolio would look like. And then at the very end, you would have, let's say we had 20 different models. We would end up with 20 different portfolios that we would

average together. And the reason behind doing it that way, rather than averaging the signals is if you only average the signals together, all you're doing is creating one new signal. You're not actually getting the benefits of the ensemble, particularly if your portfolio construction has nonlinear elements to it. So if you're running some sort of, let me pass my trend following signals into a mean variance optimization, that's going to result in nonlinear output. If you're combining all of those signals prior to the optimization, you're just basically creating one new signal versus what you really want to do to maximize the benefits of the diversification is do it after that sort of nonlinear step. So for folks who are familiar maybe with some more obscure mathematical

stuff, this is the basic idea here is Jensen's inequality, which is basically that the expectation of a function is not equal to the function applied on the expectation. And it's this weird idea of concavity and convexity, but it's a really important and subtle nuance here that I think a lot of people get wrong when they try to introduce diversification of models. They start at the top, they get all these trend signals, they might get a three month, six month, 12 month return, and then they average them together. And all they've done is created one signal, they haven't actually really benefited from diversification.

So it's like a play on the sum is greater than the parts of the sum of the parts is greater than the sum. Yes, close to that. Yeah. Close to that. The, um, and so, but I was going out. So on the portfolio level, on each of your strategies, though, if you're doing fixed income, are you running an ensemble of fixed income approaches as well? Yeah. Yeah. So, uh, if we are like a tactical fixed income mandate that we manage, we actually have five tactical fixed income mandate that we manage, we actually have

five sleeves within that mandate. So we take a sort of style driven approach. There's a value sleeve, there's a momentum sleeve, a carry sleeve. Within that carry sleeve, we have a number of different models in which we use to measure carry. And so what we would do is for each of those models and that measure of carry, we would create a portfolio. And then at the end, when we go to build that sleeve, we would average all those portfolios together. Got it. So the signal and what's, what consideration is there given for correlation and max drawdown across the signals, things of that nature,

or do they stand alone and just this is a good standalone signal. We're going to add it to the ensemble. So that's a really, really good question. I would say you need to be thoughtful either. You need to be thoughtful at the top and say, I'm going to only pick signals that I know are going to create diversification, right? So to your point, if I said, I'm going to have one short-term signal, a hundred intermediate-term signals, and one long-term signal, and I'm going to flow it all through and come up with, you know, I don't know, 102 portfolios and average them together, well, I'm still very, very weighted on intermediate. So to your point, you either need to, at the top, choose an array of signals that are appropriately diversified so that

when you just equal weight average at the bottom, you're okay. Or at the bottom, you need to do some sort of weighted average to take into account that diversification aspect. Yeah. And I call this concept of fundamental versus statistical correlation. So, and we see this with investors all the time on our platform, they grab three or four hedge funds, they run the correlations, they're not correlated. And then something happens and they lose money. And they're saying, why did I lose money? I'm like, well, because you pick four short vol flavors, you know, and one guy, so that's what I call fundamental non-correlation. I want to look at what their return drivers are. You didn't understand what the latent risk factor was there that didn't show up in the returns. Exactly. Right. It's in there. It just hasn't shown up yet.

And so that's an important concept. So you guys are doing that. So you explained the two methods, but you're doing which method? You're looking at it from... So we typically do it at the top level, just because it's a little bit easier than trying to say, let me flow to the bottom and reweight these. We are very careful at picking the signals at the top such that we can just equally weight them at the bottom. And that equal weighting is based on the risk, which is what you mentioned before, the drawdown component, or is it? No. So again, so let's see like a very, very simple example, just to give a concrete drive home here. Let's just say all we're going to do is be long or short the S&P 500. And so if we're above the 200 day, for example, we'll be long. If we're below the 200 day, we'll be short.

We might come up with a thousand different ways of measuring that trend. So maybe it's prior 12 month returns, maybe it's 10 month moving returns, maybe it's 10 month moving average, maybe it's 13 week moving average over the 34 week moving average. And each of those would basically 70,000 minute bars or something. Exactly. Variance bars, tick bars, volume bars. I mean, there's so many different ways in which this can be done. And so what we're going to do is we're going to try to at the top level, ensure that we have the, you know, equal weight diversification of different categories. And then each of those would basically say on or off for the S&P 500. And then what we would do is at the very end for

all of those on or off, basically one or negative one signals, we would average them together to get our final position. Got it. So I'm going to come back to this Twitter question here. So when it comes to portfolio construction, how does at see Hofstein, is it Hofstein or Hofstein? Well, the long story there, but it's Hofstein. Is it Hofstein or Hofstein? Well, the long story there, but it's Hofstein. Hofstein. All right. How does, that's your Twitter handle. How do you look at the trade off between trend speed, convexity and allocation size? Oh yes. I saw this question. This is a real softball from my buddy, Andrew Miller. All right. Speed, convexity and allocation size, right?

Yeah. Yeah. So. So there's a really interesting part. So we talked about this idea of trend following being mechanically convex. And what we find when we look at different trend followers is that you have fast trend followers, intermediate term trend followers, and slower term trend followers. And there's a quant by the name of Archer Sepp who introduced me to this concept. It was really very obvious when he showed it to me, but something I had never really thought before, which was the period over which you see convexity emerge is inherently tied to the model speed. So for example, if you have a very, very fast trend following system, something that's looking

at prior one month returns, for example, is going to change very, very fast trend following system, something that's looking at prior one month returns, for example, it's going to change very, very quickly. We would expect to see convexity when we compare that model's returns versus it's the underlying asset over say a two week to six week period. It's got to be a period sort of centered around the speed of the model. You're not going to see convexity when you look at year versus year returns. Conversely, you've got a very slow model. It's running like a 200-day moving average or a 12-month return. You're not going to see convexity day-to-day, week-to-week, month-to-month. You should not expect that model to perform well in something like a corona crisis, right? It's a slow model. You should not expect that model to perform well in something like a Corona crisis, right? It's a

slow model. You should not expect to see convexity when measured month versus month returns. But if you start comparing six month returns against six month returns or 12 month returns versus 12 month returns, that's where you start to see convexity emerge. I think that's the mathematical terms applied to what the investors are feeling of like, where was the crisis period before? What they're really saying is, why was there a mismatch between the convexity I expected and the convexity I experienced in the market? Yes, absolutely. And again, what you tend to find is those very slow moving models, maybe don't have as much bleed necessarily, right? Because you can almost tie that to how quickly am I turning over my options or how far

in the money versus out of the money or the options I'm looking at that I'm trying to delta hedge. And so that very, very fast model is like I'm looking at placing a straddle that's probably close to at the money and I'm rolling over every single month. And so it just becomes very, very expensive and very, very choppy as to you're getting in and out and in and out versus your longer term slower models. You know, when you look at what actual straddle you're replicating, that straddle tends to be deep in the money at any given time. If I'm looking at prior 12 month returns as my signal, I'm basically looking at where the market was 12 months ago and saying, that's the straddle I'm trying to replicate. And at any given time, that can be pretty in the money in either direction. And so I'm not likely to make any changes in the short term. So you're not necessarily going to get

as whipsawed, but your expectation is also the signal is not going to change as fast. So you wouldn't expect to see that convexity emerge over a very short period. So there's this inherent trade-off between speed and convexity that I don't think investors give a lot of thought to, particularly as it relates to the type of crisis they're trying to protect against. And particularly as the more retail-y they get, right? If they're just buying the AQR managed futures mutual fund and thinking it's going to perform when, you know, in a down 4% month in stocks or something. Right. Exactly. Exactly. Even something like Q4 2018, you know, the question is, okay, you had a three month sell off, but you're looking at a trend following system. Most trend following mutual funds are going to be in that intermediate to longer term horizon. You can't look at a three months and expect convexity. You

need to look at six, nine, if not 12 month rolling periods. And I think, again, these systems are more designed for 2008 or 2000 to 2002 than they are for a March 2020. And then the big question is whether we ever see that type of move again, or whether it's been fed, put it out or, or whatever reasons, you know, people holler trend following is dead. There's too much money in it. There's this, there's that. So my opinion is it can't be dead because as we said, it's mechanical. But what are your thoughts on that? Are those days over? Well, this is a tough one. Um, I mean, I, so I'll start with by saying one of the reasons I like

trend is because of that, that split between mechanical convexity and the premium is the premium gone. I don't know. It's hard to measure crowdedness in strategies like this, especially where trend following is such a broad category that trying to figure out, I mean, the dispersion in returns, and you probably know this better than anyone else, the dispersion in manager returns is so wide. It's very hard sometimes to say that they're even in the same category. Like Crable put in the managed futures category and they have 6,000 short-term intraday trading models that they're doing. So really nothing to do with trend following. Right. And so it becomes tough to say, okay, is it really getting overwhelmed? I think Eric Crittenden would say, who we were

talking about earlier, Eric would say, well, look, this market is still dominated by hedgers and maybe they come in and out and maybe they're harder to sort of distinguish now because they're using banks to execute their flow. But at the end of the day, so long as they want to hedge, then I should earn a premium by taking the opposite of their trade. So if prices are high and they want a short, then if I go long, I should be earning a premium. So as prices go up and I follow that price up, then I'm earning a premium for selling them that insurance. And so trend following is inherently the insurance strategy in much of the commodity complex. I think it's a really interesting argument. I think we do have to consider though the question of crowdedness and are there better ways in which we can measure it. But I still think that

that mechanical convexity aspect of trend following makes it a really interesting diversification play from the way portfolios are traditionally constructed, which are just inherently mean or versionary driven. That's all rebalancing is you're selling your winner to buy your loser. And so if you're, that is all you're doing, there may be benefits in including a system that does the exact opposite. Agreed. And my argument always to that is like, do you, this move we had in crude oil, right? We went from 50 to negative 37, even ignore the negative part. Even if we went to 15, like that's the kind of move these things catch. And that's the kind, do you think moving forward that that move is eliminated from possibility? Like it seems impossible. I sort of look at it as saying what could cause auto-correlation?

There are a couple of things to me that can create auto-correlation in markets. The first is that the market just doesn't sort of see the fractal nature of economic solvency, right? And the situation we're in today is it just the market is not adequately pricing the way in which businesses are dominoes to each other and that we might end up in a situation where business is default and that causes other businesses to default. And so you get this slow rolling recession that becomes sort of economic gravity to the market. The other way, which I think we saw was more of a 2008 is more

of a solvency driven crisis. And I think a solvency driven crisis is one of those that remains, excuse me, a liquidity driven crisis, credit, credit and liquidity driven crisis. It remains one of those that it's, it's out there. I mean, that's exactly what partially happened in March as you started seeing massive margin calls, firms went under, positions got liquidated, hedgers were having to put on massive hedges. Vol players, people who were sensitive to volatility, were having to force de-risk. All of that stuff creates autocorrelation in the market that can potentially be exploited. It's just a question of how far and how fast. So March, we saw it happen very, very quickly. I think August 2008 was more

of a rolling situation. And I still think those rolling situations are certainly possible. The question you brought up about the Fed put at the end of the day, I think brings us back to more fundamental questions of if the Fed has ultimately eliminated left tail risk, why would we ever expect equities to give us a return above the risk free rate? Yeah. You know, if there's no risk, why is there any reward? So then the question becomes, well, if there is a reward, is it because the risk is ultimately that the Fed could one day fail? That is the ultimate tail risk that we need to think about. And I don't have the answer to any of that, but I think it's worth going back to first principles and sort of trying to think about.

Yeah. And I think a global macro approach would say like, well, the Fed put doesn't exist on all currencies, on all agriculture markets, on all energy sectors. So you can have these moves that will, by definition, affect those other things. So even if the Fed put is always there, it can only go so far as well. Absolutely. Some other questions here. We're like taking a request from Twitter here. I love it. That's not good. It's a dangerous road.

We won't do them all day. And we're not live. So how's that possible? But I wanted to say, do you guys do any research on what the other players in trend are doing? In terms of very specific players? Well, maybe not just in general.

My view is that to that this convexity mismatch and people were sick of the bleed, and so a lot of the firms went to, I'm going to add some long bias, I'm going to stretch out the look back because then there's less carry and there's less noise. So those things, when I looked over the short term, from 2010 to 2015, long bias and longer term outperformed,

uh, neutral stance and, and shorter term. So I think a lot of these guys shifted and it almost by definition that's made the short term crashes less apt to be captured. So there's a little, uh, style drift kind of stuff baked into there as well. And questionably a little bit of reflexivity too i i do wonder um i do wonder with the mass adoption of trend following post 2008 how much of those firms were really forced to go longer term in nature just to put that capital to work if everyone had stayed shorter term um the market impact may have been much larger.

And so it just might have been a more natural bleed. So the answer is, okay, how do you deal with more capital? Well, you have to slow down your system and trade less. And so you're not creating as much commission issue and you're not creating as much market impact, which inherently pushes you towards convexity at a different time horizon. So I've seen that. Like when you look at sort of the SG trend index and you try to correlate it to different trend speeds, it does seem like everyone has slowed down quite a bit.

And so the question is, does that mean that trend is less effective as that speed? Does that mean there's more crowding? I still don't have answers there. Yeah, or to your point, like who cares? That's the one model those guys chose or the tilt those guys choose, use an ensemble and choose all the tools. Right. Yeah. And an individual can still go out and find a short term manager, right? They're, they're out there.

So you can say, okay, I, I, I know most of the market might have shifted towards intermediate and longterm. Let me buy some broad market exposure. And then I'm going to go make sure I partner it with a short-term trend follower. All right. I'll ask one more from at economic pick. I don't know if you know him. Yeah. I know Jake. Jake. All right. Best strategies to park your capital and when equities are in a downtrend besides cash or bonds. He knew what I was going to answer. So this is a really interesting one because my personal view is you go to cash. When you start, and this isn't, a lot of people might

say something else, but when you see negative trends in equities, I prefer to go to a de-correlated asset rather than take on any basis risk of trying to profit or taking on that negative correlation, trying to profit from an actual short selling in equities. I think short selling in equities has proven to be incredibly, incredibly difficult. I think the nature of those markets are far more volatile than you get with positive trends. I think in those types of markets, you are just trying to preserve your capital. And so I don't particularly like going short equities when you see negative trends, or certainly I don't like just doing this sort of opposite trade that when there's a negative trend, you blindly go short. In those mandates where we do have shorting, we tend to take very specific profit targets. So once we go short, we say, if the short gets to this level, we're just going to take it off

and go to cash and take our profit and sit there until a positive trend comes back rather than try to ride a short all the way down because we think it's just far more difficult to manage trend following in those type of market environments. Jake might be referring to, you know, what about going to some other types of positions like long ball or gold? I think what you tend to see is in a true... Bitcoin. Bitcoin. I think what you see is, and you saw this in March, is even something like treasuries,

which are typically a flight to safety asset or gold sort of right before March 23rd, you just had margin calls everywhere and you had forced liquidation. And even those assets aren't safe. So anything that can be sold will be sold. So my view is if you don't have to hold anything, don't hold anything. As the crisis gets deeper, there may be some positive convexity plays if you wanted to start nibbling. So in 2008, you might look towards things like high yield bonds. I know Jake is a huge fan of some levered closed end funds that might represent a great opportunity for positive convexity. But if your goal is just pure preservation of capital, I think it's hard to beat short term US treasuries for that need.

And what are your views on the ability of U.S. treasuries to provide that flight to safety in the future at this zero bound here? So this is another really interesting one. The question becomes, if you know that treasuries are going to be a hedge for U.S. equities, should treasuries have a positive premium any longer? Right. I mean, that's another just interesting fundamental question. If we expect... You could almost say the past 30 years have proven that out, right? As it's become more and more of a flight to safety, the yields have come further and further down. Right. So the question is, can it provide the same protection it has in the past, given where rates are?

I will, in financial markets, never say no, it can't. Certainly, there could be a market environment where from where rates are today, they could plunge negative. And given duration on 10-year, you could get plenty of participation. But when you compare prior crises where the market had to fall hundreds of basis points, I think you're talking about going to very, very negative territory. So is our longer dated treasuries, is the juice worth the squeeze? Again, I think you just sort of, if your goal is pure preservation of capital, I think very short-term U.S. treasuries are probably the safest place you can be.

I would say from a classic managed futures program that would just go long bonds in a crisis, right? They're going to... I don't think that basis point move matters because they're going to size up because the volatility is much slower. So they should make the same on a 15 basis point move that they used to make 15 years ago on a 200 basis point move. That can create some real market structure issues if every single managed futures player is massively levered long treasuries at the same time and margin rates go up and everyone has to liquidate.

Yeah, and I think the other problem there is it was great as not managed futures, but overall like, yeah, I want to have some treasuries as a crisis alpha hedge kind of thing. That's great when they're paying you 2, three, four, five percent or something. When they're paying you zero or you're paying them, do you really want it as that hedge or do you look for something else that might have a little positive carry or something else like that? It's an interesting piece. Great. The last one jokingly here was, hold on, my phone just turned off. From Mike Lambris, at Mike Lambris, why are you so good at what he does and what's his favorite snack to eat while doing so good? Well, I don't know if I'm that good at what I do,

but what's my favorite snack to eat while I do it? Definitely, if it's a cheat day, peanut M&Ms for sure. Really? Oh, yeah. You were a health nut and'm for the cheat day peanut M&Ms for sure really oh yeah you were a health nut and then you you hammer the peanut M&Ms they're well I I pretend that they're healthy you know that dark chocolate plus the peanuts I assume there's some health in there somewhere you're a coder like every programmer coder quant that I've ever known is like pretty much some of the most unhealthy people I've ever seen. Just like bags of Doritos and like, what'd you get for lunch? They're like fried chicken or barbecue. It's hard to be healthy when you're sitting in front of a computer typing

all day. Yeah. I like the stories of some of these guys during lockdown have dual jobs at like Google and Microsoft or something something some of these super high-end programmers are like this is i'm so good i can do this they're moonlighting i just envision them with like two hand doing their work with one i love it well it's it's funny to me all the all the guys who are used to their sales and tradings job that would have six monitors or something trying to get that whole thing set up at home and you're seeing these home setups and it's like you know you need to have it like you need to be able to see the markets but it's so funny seeing these huge monitor setups on these like really tiny desks that weren't built for it

anymore yeah the uh my neck has started to go because i never i didn't bring my monitors set up home i just have my laptop so you're constantly in the yeah prone neck thing but whatever i'm getting old. So just wanted to ask quickly, all this stuff and 30,000 models would be no problem, you said, for the computers and whatnot. What, are you guys using any machine learning or AI to kind of process all these different data points? So I would say both yes and no. There are machine learning or AI to kind of process all these different data points? So I would say both yes and no.

There are machine learning techniques that we utilize and ideas that we've stolen from machine learning. So we might use something like subset resampling, where we will take certain features and kick them out and try to rebuild a portfolio. Or if we're building a portfolio of a number of names to try to create stability, let's say there's 20 names in the portfolio, we might kick out five, build a portfolio, and then kick out another random five, build a portfolio. So ideas like that, that we've ultimately stolen from machine learning, yes. More explicitly, things like are we using deep neural networks

or anything like that? At this time, no. We've messed around with random forests, which I sort of have a preference for, gradient-boosted random forests, that sort of stuff. Haven't found a tremendous amount of benefit statistically to justify the added complexity. But it's certainly an area that we have a lot of ongoing research on. How about just for your research and like a brute force approach to like crunching through data and whatnot? I mean, most of that I wouldn't say is machine learning. We're just, after a decade of doing this type of stuff, you just build up a whole lot of different models and algorithms that make the research process that much more efficient. I can't wait for the near future with just for writing blog posts when the machine just spits

out some interesting charts for me and says like, Hey, this is pretty cool. You should write on this. That really would be wonderful. I mean, I think what would surprise most people, at least for us, a lot of people comment about the pace at which we're able to write research. I think a lot of that comes back to the systems that my team has built over time for us to be able to rapidly do research and say, okay, this is an interesting idea to be able to quickly get the data, test out an idea and a hypothesis, be able to quickly generate graphs and charts and all that sort of stuff is something that's been accumulated over a decade. And if I was starting fresh today, it would just be near impossible for me to get back to the working speed I have now.

Did you ever struggle over the years with like, do don't teach like that, you know, selling it more than giving away the research or anything of that nature you know i so the research for me actually sort of serves a dual role which is you do a lot of research over time and a lot of it goes nowhere but a lot of it it doesn't mean it's not interesting you can learn lessons from research that goes nowhere and the risk of doing research that goes nowhere. And the risk of doing research that goes nowhere and not memorializing it is that five years later you go, I think I did something related to this. What did I do? And then you go back and try to find the code or the data. It's just not there for us. The commentaries actually served as a way for us to memorialize everything we were working on. Um, the ancillary benefit, and I certainly will be very upfront

about this, is it serves as marketing. People find us, they see what we're working on, they see the thought process behind it. They either really like it or they don't. And if they like it, they'll very often engage with us. And so it sort of allows us to say, okay, let us memorialize everything. It forces us to create a code base. It forces us to isolate the data that we're using so I can go back five years from now and replicate exactly the project I was doing. And then the added benefit of being able to show people the sort of things that we're working on. We've been doing similar approach for way too long now, like 15 plus years. But I can barely have a conversation with people without,

oh, we wrote a blog post on that. Oh, you should go see this post we wrote. Well, and that's a great point, which is when you're in the asset management game, you spend a lot of time educating on your process. And you answer so many of the same questions over and over that you can often say, I'm just going to write a piece on this. And it really helps. You can put it out there and people either come to you pre-educated, or for those people who want to take a deeper dive, you can say, I already wrote a 10-page white paper on this topic. Have at it. And then when you're done with that, if you want more, I've got another 10-page white paper you can read. So it really allows those people who want to go deep, if you've been

writing consistently over time about your process and your research and your discoveries, it allows them to really go as deep as they need to, to get comfortable. When it's the modern world too, right? Like all the successful companies these days, they prove they're worth to you. They give you value for free, Google Maps, yada, yada, yada. They give you the value and then they extract something on the backend. So I think like an old school hedge fund or asset management firm would value for free google maps yada yada yeah they give you the value and then they extract something on the back end so i think like an old school hedge fund or asset management firm would have been like what was the old smith barney we earn it or whatever right they tell you how great they are and trust me i'm great versus i think new school is just hey here i'm sharing this value

with you i'm giving you value both because it helps me and I believe it helps you. And if you like it, let's do business. If you don't like it, you know, go down the street. Right. Exactly. Exactly. It's in a world of asset management. We are, I think many people would say we're close substitutes, right? Why would you ever do business with one firm versus another?

As a small boutique, I think you need to earn trust. And for me, brand and trust are sort of equated and earning that trust over time as a quant firm, I think really means showcasing your work and your thinking. And obviously it has to come out in the returns of the portfolio as well. But every day, it's just one extra day of return. So the question is, what can you do to help really enhance that trust and show people the evolution of your thinking and how your team continues to build upon the work. That's where I think publishing is really important. I used to joke with people are like Winton, you know,

it's the biggest trend follower out there. They've shifted gears a little bit to equities, but back in the day I joke, Oh, they have, people would say they have 62 PhDs on staff or something i'd be like yeah 53 of them are in marketing and asset raising um so let's jump over finish off the pot here with your favorites uh just go through here quick fire so going to school in cornell favorite finger lake favorite finger like i don't even think I could tell you the Finger Lake names anymore. Oh, no. Kuka?

Cayuga? I think Cayuga. I'm going to say Cayuga because I think Cornell was on the Cayuga Lake. Yeah. All right. You never spent time on those? No. You were coding the video games?

Exactly. You know, the funny thing about Cornell is the week it was most beautiful at Cornell was always finals week. And then you left. So it was like, by the time you wanted to go visit the lakes, you were, you were gone anyway. You were out. Um, how about, was there any beer drinking? Favorite Genesee beer? Was there any, I mean, yes, there was beer drinking. I was too poor to afford anything though. So I think it was all Bud Lights at the time. No, that was the Jenny Cream Ale is like the New York. Yeah, I think my buddies at Geneseo would drink the Jenny Cream. But at Cornell, I think it was, I think Cornell was like a Coors Light school. Nice. I was before that time, there was no such thing as Coors Light or Bud Light anywhere in our

area unfortunately uh favorite podcast besides your own oh favorite podcast there are a bunch of really good ones out there um I mean I think it's it might be cliche to just say Patrick O'Shaughnessy's he does a phenomenal one uh our our mutual friends at, I think, have a great one that they showcase. Yeah. How about Bill Simmons, a Boston buddy? You know, I don't listen to a lot of non-trader podcasts. I'm trying to think of who else. Top Traders Live is another one that I think is really high quality, worth listening to almost every week. So favorite Boston restaurant? It just shut down actually. It was called Coda. They had the best burger in the South End, in my opinion. I would say another one that's near and dear to my heart is Vahigantes,

which I have no idea whether it's still there or not, but that was a date night favorite for me and my wife. All right. Favorite Boston actor between Affleck and what's his name? Damon. Damon. We're going Damon. Yeah, Damon. You're going Damon and remembering his name for me.

Yeah. And favorite Boston sports team. Are you a Patriots fan? You know what? This is embarrassing. I just, I've never cared about sports. And living in Boston for close to the decade I did where it just, every team came home with a trophy.

I just was literally annoyed at the parades, which is the worst thing you can say. Because I was so ungrateful about all the victories. So you don't listen to Bill Simmons. He's, he's constantly, I will say hands down, best one to go watch was always the Bruins. If you were going live, had to go Bruins. Yeah, that's a great spot. And, um, so favorite Santa Monica activity. Oh, beach can't beat the beach in LA in LA. Just going out and spending an afternoon on the beach is tough to beat. And lastly,

well, I got my last one. Second to last is favorite snowboarding spot. It's either Vail, Colorado or I might go Beaver Creek, Colorado. Okay. You don't go out to the Tahoe or anything closer by? No, I haven't actually yet gone there because I just moved to LA about a year ago. My wife is not

a big winter activity participant. So this is more when I get to either go with my family or friends. And Vail is one we've been, as a family, my family has been going out there for a little while now. So it's got some good memories to it too. I'm going to organize an annual like volatility trader. You can come along as well. I will gladly join. Yeah, we'll get some of these guys out there

uh and finally we ask all our guests favorite star wars character favorite star wars character i'm who well who's not gonna say it's either gotta be han solo or obi-wan kenobi right does anyone say anything different uh this morning i was doing a pod with a guy from shanghai and he said Jar Jar Binks and I almost fell out of my chair so yeah there's there's different for sure well you know someone who probably knows Star Wars way better than I do could probably name some like super obscure character that I had I just don't even know exists well I think Meb went with uh Ahsoka Tan who's like from the cartoon she hasn't even been in the movie.

Yeah, there you go. I didn't even know there was a cartoon. That's good. All right, Corey, this has been fun. It's been a pleasure. Thank you for having me on. Yeah, sorry we ran a bit late, but that was fun. 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 alt and visit our website to read our blog or subscribe

to our newsletter at rcmalt.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 at rcmalt.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.