Jeff Malec sits down with Jiro Fujisawa of Asset Management One USA to unpack one of the more unique quant approaches in the space: a cross-asset factor alpha (CAFA) strategy built on 80 synthetic markets. Jiro walks through his path from mechanical engineering to quant finance, the differences between engineering-style experimentation and market reality, and how Asset Management One USA thinks about risk premia, implementation details, and factor design.
The conversation dives into decomposing futures markets into orthogonal risk factors, running trend, carry, and skew models on top of synthetic price series, and why the real edge often lies in construction and risk management rather than “new” factors. They wrap with where quant fits in today’s equity-dominated world, how investors are using risk premia alongside multi-strats, and why systematic absolute return strategies still matter when the macro regime turns. SEND IT!
________________________________________________
________________________________________________
This podcast is provided for informational and educational purposed only and should not be considered investment advice or a recommendation of any specific security, strategy or investment product. The views expressed in this recording are the personal views of the participants as of the date of this podcast, are subject to change, and do not necessarily reflect the views of Asset Management One USA Inc. itself. Any discussion of investment strategies, market conditions, or portfolio construction is intended to illustrate general investment concepts and may not be appropriate or eligible for every investor. There is no guarantee that any investment strategy will achieve its objectives. All investments involve risks including the possibility losses as well as profits. Nothing discussed in the podcast constitutes an offer to sell or a solicitation of an offer to buy any security or investment advisory service. Listeners should consult their own financial, legal and tax advisers before making any investment decisions.
Asset Management One USA Inc. is a New York-based investment advisor and a part of Asset Management One Co., Ltd. group, a global asset management company headquartered in Japan. Asset Management One USA Inc. is jointly owned by Mizuho Americas LLC (MALLC) and Daiichi Life Group. Inc.. Asset Management Group consists of AMO USA as well as Asset Management One Co., Ltd and its subsidiaries including Asset Management One Alternative Investments, Asset Management One International Ltd., and Asset Management One Hong Kong Limited.
Check out the complete Transcript from this week’s podcast below:
Trading 80 Synthetic Markets with Trend, Carry & Skew: Jiro Fujisawa, Asset Management One USA
Jeff Malec 00:09
Welcome to the derivative by RCM Alternatives. Send it. Hello there. Welcome back. You found the derivative by RCM Alternatives, where I’m not going to tell you this time that there’s a new website with a lot of cool stuff on it. Won’t say a thing, not even going to say the URL. You’ll have to Google it. Anyway, onto this pod where I sit down with Jiro Fujisawa of Asset Management One, talking his new cross asset factor alpha strategy. It’s a mouthful-and we dig in. It’s one of the most unique approaches I’ve heard sitting in this seat, for sure. He’s got 80 synthetic markets built out of real markets, going long and short, run across three models in trend, carry, and skew. What? How do you do that? We’re digging in. Send it. All right, everyone. We’re here with Jiro Fujisawa. Did I get that close to right, Hiro?
Jiro Fujisawa 01:10
You did.
Jeff Malec 01:11
Yeah,
Jiro Fujisawa 01:11
you did, Jeff.
Jeff Malec 01:13
How are you? I’m
Jiro Fujisawa 01:14
good, thanks. How are you?
Jeff Malec 01:16
Good. Good, thanks. When when were we hanging out? In Austin, was it? Yeah, that’s
Jiro Fujisawa 01:21
right.
Jeff Malec 01:22
Talking heads,
Jiro Fujisawa 01:23
uh huh,
Jeff Malec 01:24
was a good time. And then it looks,
Jiro Fujisawa 01:26
yeah,
Jeff Malec 01:26
I’m in the middle of a. We might even lose power here. There’s a huge thunderstorm roaming through Chicago.
Jiro Fujisawa 01:32
Okay, so
Jeff Malec 01:32
I don’t know if we can hear that rain coming down, but we haven’t had rain in a while, so it’s good. Looks nice and sunny with darkness.
Jiro Fujisawa 01:38
Yeah, today’s nice. Apparently, it’s gonna start raining later this evening, and the whole week looks a little wet. But so far, so good. Holding up.
Jeff Malec 01:48
And you live and work in New York, or you live elsewhere?
Jiro Fujisawa 01:52
I do. So right now, I’m at the office, which is on Park Avenue, around 47th Street. I live about 10 minutes of a walk, so can’t complain about that. Live, breathe in the city. I sometimes wish I can live outside in the burbs, but I don’t think I can handle all the management of a house. I’ve got, I think I’ve got enough of management on on my portfolios.
Jeff Malec 02:17
And who would want to do that? You talk to some of these guys who do like hour and a half commute and whatnot. You are like, oh yeah, brutal.
Jiro Fujisawa 02:23
I think a good good number of our my colleagues in the office they live, you know, especially when they have family like kids, they live out, but very few in the city.
Jeff Malec 02:36
Love it. And then, are you one of these that tries to get out of the city, go to the Hamptons or whatnot for the when it gets steamy in the in the middle of the summer.
Jiro Fujisawa 02:46
Yeah, or just go abroad. You know, like I was in Japan just till last week, and I’m Japanese. Got my parents, my family in in Japan. So unfortunately, there’s no way around getting away from the heat. It’s pretty humid and really hot over there. So, but you know, it’s it’s nice to get out of the city and get refreshed, come back, and but yeah, it’s nice to be back as well.
Jeff Malec 03:10
Here, were you born and raised in Japan? Were you here? This
Jiro Fujisawa 03:14
is going to be a long story on its own, but I was born in Japan. But after six months, I started moving around because of my my dad’s job. He just worked for Japanese securities firm and happened to move a lot. So first place was in Jakarta, Indonesia, for four years, and then Budapest, Hungary, for four years, then London, which was the longest I spent the better half of my childhood in London, but when I was in my sophomore year of high school, I moved. My parents moved back to Japan, so that was the first real time that I spent time in Japan. Yeah, so spent two years and then went back to London for for university, and then thereafter I started working in Japan. I joined Mizuho Bank or Mizuho Corporate Bank back then in 2013, and then I’ve been just going back and forth between Tokyo and New York across different functions. And here I am since in New York since 2021 March.
Jeff Malec 04:20
Nice. So when you went back to Japan, were you kind of you were trapped between two cultures? You were kind of considered an outsider, but you were like, “Hold on, I’m from here.
Jiro Fujisawa 04:29
Yeah. So I always, it’s I’m always an alien wherever I go, and I’m like, in a sense, homeless. But I also, you know, I sometimes consider myself homeful. Like I’m home wherever I am, as long as you know there’s some basic infrastructure and you know I can speak the common language. I’m I’m not really picky. I’m I’m a pretty simple guy in that respect. So, you know, once I go back to Japan, it’s like, hey, you’re that kind of outsider as Japanese-ish person. But I sometimes. Sort of, or some people get surprised by how how much I how much Japanese I speak, despite me not really living there. Yeah, and the opposite, or the same, sort of happens here, where I’m kind of, you know, speaking English, but I’m Japanese. Kind of, I guess a lot of people in New York are like that. So in that sense, this place also feels kind of like home.
Jeff Malec 05:25
Yeah, and that was all where you were just there. Mizuho’s Tokyo, or it’s all over. I’m sure in Japan, but when you go back, you’re in Tokyo.
Jiro Fujisawa 05:32
Ah, yeah, in Tokyo, that’s where the office is. My my family, but I do I did this time round travel a little, went to Osaka where they had the Universal Studio. My kids wanted to see the Super Nintendo World, which was really good. I I say my kids, but me included.
Jeff Malec 05:54
Yeah, I actually went to that the new one in Orlando. What was? Oh yeah, that was four months ago, and that was super cool. I remember I was a huge fan as a kid. I don’t know if you’re a New York Times crossword guy, but yes, the Sunday crossword was a Mario had like the tubes and the and the little question mark where you could jump up. It was a cool little theme. But that was cool. You feel you feel like you’re in the world for for a quick sake. It
Jiro Fujisawa 06:18
was surreal. Like it was really well built, so yeah. I didn’t know they
Jeff Malec 06:23
got built one in Osaka.
Jiro Fujisawa 06:25
Yeah, they did, and I think it’s like their fifth year. They had a fifth year anniversary.
Jeff Malec 06:30
Oh wow!
Jiro Fujisawa 06:30
When we went there, nice. All right, yeah.
Jeff Malec 06:34
And then where you went to school in London, or where did you go to school? And university. I
Jiro Fujisawa 06:39
went to a university called Imperial College. So i i took I got my master’s of engineering degree there. I was a mechanical engineer. I liked the maths and sciences, especially physics, as a you know high school student or even middle school. So I thought you know I I wanted to study more, but not be a scientist, but be a little bit more pragmatic. So the apply application of of science, which is engineering, and and so that’s where I ended up being. But then you know when I started off, or after I graduated, my my my my dad and also my brother being in the finance industry, I did my internship at a bank, and so I sort of I was interested in the application of sort of the numerical world in in or statistical slash numerical world in in finance, and so I wanted to join a firm that offered some sort of quant investment capacity, and I found my job at Mizuho in Boston. Actually, they have a huge careers fair once a year in Boston where they hire Japanese English bilinguals, and that’s where I got my job.
Jeff Malec 07:57
I want to see your like globe with all the pins on it. A lot of pins.
Jiro Fujisawa 08:01
Yeah, it’ll be a messy sort of a ball of yarns going back and forth. Yeah,
Jeff Malec 08:07
and always amazes me. I should one day go back of all the podcast guests, like way over 50% I think have some sort of engineering background, especially with the quant people. So it’s always amazing to me. Like what? And sounds like you didn’t actually know you were going to be mechanical engineer and then decided to be quant, which seems like what’s your theory on that? The people just see the world as a model that can be constructed and and deconstructed and put back together, and finance? Why not? It’s just another one of those pieces of the world.
Jiro Fujisawa 08:36
I mean, I I can probably spin a story all in hindsight, but you know, point in time speaking, I was I joined mechanical engineering or joined the program thinking I want to be a pilot. I want to build or even build a plane or something. But you know, as you do thermodynamics and also do some plastics sort of materials sort of studies as well, and you have some stats, you get all these elements that that are completely sort of different in its own world-they all come together to be mechanical engineering. But there’s many features within it, and so I think I wasn’t really thinking. Oh, you know, I want to be in in finance at the point that I started college. If I had, I’d probably go into some, or I may have gone into a different field,
Jeff Malec 09:23
maybe quant something or other.
Jiro Fujisawa 09:25
Exactly, but as I sort of, you know, went through college and then the internship that I mentioned earlier, I thought, okay, maybe this is a little bit more exciting than you know building a wind turbine or or doing something else, yeah. Especially back then, you know, a lot of my classmates they went to Darby or somewhere a bit more rural. In that’s where all the factories are, and you know, even even if some of my mates or friends also they went to the f1 industry, they’re an engineer. Well, that sounds silverstones. That does now only only now I start to like or I I watch Formula One, so it’s kind of nice to see some or I don’t see them on TV, obviously, but you know I or maybe one day I might. But it’s nice to sort of see that connection that I that I have,
Jeff Malec 10:23
yeah. And then, did you come to a point, or have you? Do you feel like finance in the quant world is a closed system that you can solve, or do you come to the point where you’re like it’s unsolvable, and we just want to get as close an approximation and work on the percentages and things like that? That seems to be where some of the engineer quant connection breaks. Of is it is it solvable or not?
Jiro Fujisawa 10:46
I think it’s not solvable for very many reasons. But I think in the world of sort of tangible, real physical engineering, you know, you can run experiments in in a real controlled environment, whereas in finance you can sort of, or in quant quant finance you can sort of simulate and backtest, but that’s not the real world that we want to apply into. The real world that we want to apply into is in the future, but you know, in in engineering or in science, you can sort of replicate what you expect to happen in the future in a closed environment. So I think that’s a huge difference. And also, I guess data availability is quite different, or or some sort of homogeneous set of or well-behaving data set is is something that we lack in in finance, whereas we do in many cases, in in the sort of physical world,
Jeff Malec 11:46
and that’s like any physics classroom around the world. If you put the right ingredients and the right amounts in this experiment, it’s going to come out the same. And any point in time, the same students doing something in the markets is going to get a completely different.
Jiro Fujisawa 11:58
Yeah, and I I think that way. I’m obviously probably really underestimating the sort of complexity, and I’m sure people working in you know in in the engineering sort of sector would probably also argue that you know not everything is simulatable or replicable in in in real life. But you know, as I see it from where I stand right now, I think it’s a little bit more. There’s much more unknowns in in finance.
Jeff Malec 12:28
Yeah, we used to work with a guy who did a estimated the nuclear contamination in the groundwater of Pigeon Forge, Tennessee, or somewhere like that, where they had tested and built nuclear warheads. So that’s right. You can’t actually measure it; it’s underground. So it was sampling and little pieces, and so he would probably argue to that latter point. Yeah, like
Jiro Fujisawa 12:50
yeah, even
Jeff Malec 12:50
in the physical world, you have to use a lot of estimation and and processes. While I mentioned, but I was just arguing this with your kids. I don’t know if they’ve done it, but like you have the classroom. Hey, we’re gonna do a stock picking contest or whatnot. They drive me crazy. I’m like, this isn’t teaching them anything. It’s just like random luck. Who picks? Like, first of all, if you want to win it, you should just pick the highest beta stock probably and hope that it goes up, right? So anyway, I went. I went off tangent there. But if you have any thoughts on how to win the school class stock picking contest. Let us have it. Design a model.
Jiro Fujisawa 13:25
That’s a that’s a difficult one. If I knew, I would probably do it on my on my PA.
Jeff Malec 13:30
Exactly, because it’s like what’s going to happen over the next, and it’s such a finite time. It it’s teaching them bad lessons. So you mentioned Mizuho. You technically work for Asset Management One, correct? So tie that whole picture together for us. How that all came to be, and and what’s happening there.
Jiro Fujisawa 13:56
Sure. So it’s a little bit of a complicated or a different animal, or the the whole char system in a Japanese large institution is quite different from the standard practice over here. But there’s a lot of transfers in Japan, so even even if you join a firm, you get you tend to get rotated, or at least that was the sort of traditional way of incubating or building in-house capabilities in a Japanese firm. So, but by your
Jeff Malec 14:30
own choosing, or they would just take you and say, “Now you’re in this department.
Jiro Fujisawa 14:33
Kind of both. Yeah, it’s a mixed sort of system. But I, when I joined, I specifically joined a under the terms that a stay within the sort of asset management division, and I had my specific goals and things that I wanted to do, and I thought I and I think that I’m gonna, which is kind of in the in in the space that I’m at. So when I. Joined Mizo Bank. I was first sent to a sort of funds of funds slash gatekeeping division. So I was looking at global macro players like the Brevin Howards of the world, and yeah, you know Millennium, and usually the the blue blue chip names of of hedge funds. And I was advising Japanese institutional investors, mainly in the pension space, as to how much or what what they should invest in and how much, et cetera, et cetera.
Jeff Malec 15:32
Were any imposter syndrome of like, why am I in between these billion dollar transactions as a gatekeeper? I just I’m just getting used to this stuff.
Jiro Fujisawa 15:43
Since I was, I was kind of like first role. I didn’t really, you know, have really hard expectations as to what is good of the the the the standard. So it was, it was a good experience. Love it. And then just to back up, what’s the what’s the company line is a the the financial group has various arms from banking. It covers retail and also wholesale. It also has securities, kind of the investment bank division. It also has a trust bank and also and before they used to have many small subsidiaries that that have been doing asset management, i.e. running external capital and doing money management. Now, because we had multiple firms, we merged all that into one big entity in 2017, and that’s where asset management one. That was where when when it was born, so asset management one, where I work, for whom I work, is a company owned by Mizuho Bank and Daichi Life. So we’ve got two parents, Mizuho and Daichi Life.
Jeff Malec 16:53
When you were allowed to do that as a bank, yeah,
Jiro Fujisawa 16:55
yeah, exactly. And then at that point, they we or they we sort of decided that we want to run external capital, and then you know, 2010 ish risk premia one point came about, and then 2.0 later on, and QIS recently, whatever you know. There’s a lot of yeah. What are we calling what we do? Yeah, I don’t know, just quant. Let’s keep it simple. Risk premium, and then so you’re actually in there, or over your career, have been in there coding, creating various risk premium models, various QIS. So I started this particular role 2017. I initially started as a trainee here. I did program back in uni days. I did MATLAB and R, but not in Python. So I had to sort of, you know, not relearn, but sort of transfer my sort of language from from those languages to Python. Develop my own strategy. Get grilled at the investment committee, and you know, come up with some prototype. You let it run on on no capital. So we typically once once we have a strategy developed, we we tend to have a sort of dry run period of a few months, and then we would put some capital allocation, and then gradually increase as if if it makes sense.
Jeff Malec 18:21
And then are these are all supposed to be or designed to be betas, for lack of a better word, right? They’re just known risk premia, or maybe they’re a little bit only known to you risk premia, or something of that mix
Jiro Fujisawa 18:35
bag of both. I think arguably it’s it’s very rare to find any sort of groundbreaking, revolutionary idea that that prints money and is different from everyone else. I think that’s there’s various reasons as to why that is the case, but you know, firstly, by definition of a quant, it has to be statistically significant. So, whatever phenomena that you’re trying to capture, you need to you need it to be. It needs to be significant. You can’t just you know try to capture a really short term thing in in size and scale. So, I think a lot of the times we do source our ideas, or at least more on the sort of traditional factors from from academia world. So it’s much more well covered, discovered. Usually, that comes with some intuition, which is something we always like to have. Right. There’s a
Jeff Malec 19:36
paper of buy on Tuesday morning, sell on Wednesday evening, and that just was the whole concept. You’d be like, “Well, what’s why? Yeah, so you’re against that. You want to know why?
Jiro Fujisawa 19:48
Yeah. Sometimes it’s a little easier to explain than some other times. It it may be a little bit more heavy on the empirical evidence as opposed to more fundamental. Oh. You know, there’s these kind of flows are driven by these kind of retail flows or whatever. It’s it’s not always fully explainable, but it at least needs to be intuitive. We can’t just be looking at data and trying to data mine it and come up with a beautiful back test of a sharp three or whatever,
Jeff Malec 20:23
and then give us like five examples: carry,
Jiro Fujisawa 20:27
carry value, trend. Maybe not really for screaming, but like we do have some defensive strategies. Shortfall, I guess that’s kind of an extension of carry, but different from like a curve carry, but yeah, those would be the typical sort of traditional factors. But
Jeff Malec 20:48
then that we do also. Sorry, go ahead.
Jiro Fujisawa 20:50
Sorry, I was going to say we also do have a little bit more of a niche, or if you like, bucket which we call market imbalance, which is a sort of omnibus category for various things that we the premium that we try to harvest by providing liquidity to the markets and getting that premium out of it. It can be like a like a really well discovered covered one is you know congestion. There’s certain flow patterns or dynamics in markets that we try to provide liquidity and and gain money out of it,
Jeff Malec 21:22
and that as the bank can provide that liquidity, or any investor could access that premium, provide that liquidity.
Jiro Fujisawa 21:28
I think any investor can access it. It’s just whomever can provide that liquidity, at least theoretically, gets that premium. So it is a it is risk premium in that sense. If if that flow or if that thesis doesn’t hold, then you you that is the risk that you hold.
Jeff Malec 21:58
And then talk for a minute. There’s been some papers. I mention this all the time. I should go actually get the actual name of the paper, but right that once a risk premium is published, its efficacy declines quite a bit, right? And once it’s like out there in the public and all the investors are using it, have you guys seen that in practice? Do you have a number on it? Like, what is it still good enough? Even though we know it’s not going to be like what was in the paper. Like, how do you think about all that?
Jiro Fujisawa 22:25
I think it differs from factor to factor. Some factors that rely on, especially on certain flows, where you know maybe if if everyone piles in on that flow, then you might have certain market concentrations that sort of essentially killed that that premium that was that used to be available, but in in other cases where you know maybe in certain commodities markets where you know that the impact of that isn’t as large as the underlying sort of flow mechanism, then the impact is marginal and shouldn’t impact the the efficacy of of that factor. And I think it’s difficult to say that okay, because a lot of players are piling on on the same bed. Risk premia has a sort of life expectancy of X number of years. I think the the really fundamental risk premia strategies should work in the long run, but I think it’s also about parameterization. You don’t want to be sort of always tweaking it, but I think the design of a strategy should be able to take in account of these adaptations in the market. So I think that’s what we really value in our research process. So sort of tying that into what what we talked about earlier, there’s very little groundbreaking ideas, but I think the the importance or the a lot of value comes into the detail of the construction. We we just call them implementation in our shop, but I think that’s the key in delivering a good strategy within the risk premium space.
Jeff Malec 24:14
And something as simple as like execution algos or speed of execution, things like that can be part of the part of the game. Agreed,
Jiro Fujisawa 24:22
agreed.
Jeff Malec 24:24
And then, so it’s always funny to me. We talked about well, there’s 510, risk premium, but then how many are actually on the platform? Like hundreds? Is that just tweaks or different parameter sets of of known factors?
Jiro Fujisawa 24:38
If I had to put like a number on it, I’d say maybe 4050, purely different ideas, and then there’ll be hundreds of different variations for different purposes. I think that’s the sort of ballpark, right? If I run like
Jeff Malec 24:54
a PCA analysis on that, I’m going to come up with like 50 unique bets, maybe,
Jiro Fujisawa 24:59
maybe so. The I think the sort of ex ante and ex post worlds are two separate things. Conceptually, one risk premia strategy may be completely different from another, but at the end of the day, they kind of look similar. If you I don’t know run a correlation on on those strategies versus the South Gen CTA index, you might get like point 6.7, sort of correlation. So in that sort of long run, maybe the correlation structure may be quite similar, but that’s kind of looking at the longer term trend or the longer term behavior and the similarities. But I think when you’re looking at your PNL day to day and the positions in certain markets, the slight difference could make a huge difference in the sort of end return of if you compare different strategies over over the same period.
Jeff Malec 25:58
And then, do you feel like investors? To me, it’s like what happened to cable, right? Cable, we had all the channels bundled together. I’m getting one nice product, and then everyone’s like, “No, I don’t want the Hallmark channel. I don’t want this. And unbundled, now I can get all these different pieces. So it’s basically what QIS or risk premium is, right? The investors can get just the pieces they want. They don’t need the Hallmark channel, or I’ll flip that. The people who love the Hallmark channel, you don’t have to have ESPN, but I’m assuming most of our listeners want ESPN and don’t want the Hallmark Channel, which might be put put a note in the comments. Let me know if you want that Hallmark. Anyway, in the cable space now, we’ve seen it kind of rebundling. Like people got too much choice, and they didn’t want to have to manage all these different streams, and now they’re like, “Can’t you just put this back together to me? So, any risk of, or maybe opportunity there, of like, are people trying to manage this, or how do they manage all those choices? Are they starting to say, like, “Can you just bundle these ones I want for me?
Jiro Fujisawa 26:55
Yeah, I think the key is in having the the flexibility to provide both, so they can choose to choose or not to choose. Typically, when we have conversations with certain investors, we would sort of start off by, you know, having a sort of default basket of strategies, one sort of portfolio, if you like, and then certain investors, we would we would sort of go into the details of what goes into that portfolio, and then maybe some some investors may already have trend, so they don’t want any more trend. So if I weed that out, then okay, that directional component with the within the portfolio is gone. So, how do I sort of make that basket a little bit more well-rounded between the different kinds of strategies? So, I think if certain investors may want to just have just end the conversation there, but others who want to have that sort of piece-wise offering, we can you know sort of have a conversation that goes more into detail of of the specific factors or strategies that go into it.
Jeff Malec 28:11
And then, do you see investors more? Are they doing risk premium instead of a multi-strat hedge fund or whatnot, or in concert with of like okay right the multi strats are hiring a bunch of different PMs they’re doing a bunch of different risk premia there’s a center book they’re weighing it all or a pod shop if you want to call it a pod shop but they’re basically using all those risk premia and balancing it internally giving you one return what what are these investors after that same thing they want to build their own multi strat essentially.
Jiro Fujisawa 28:43
I see both. I think some some people might just want the sort of you know some the multi strat strategies have already some stat arb or some sort of quant element into it, and then they actually want some additional exposure that that’s complement complementary to what they already have, which is where sort of the QIS slash risk premia stuff comes in. We’re sort of we’re not a hedge fund that sort of is set up in such a way that we have multi multiple PMs separated by you know separate different capital allocations and sort of competing PM by PM to to get the the best single portfolio. Where the approach we take is a little more collegial and more sort of institutional approach, where you know we manage one book, one book, different ideas, different factors or strategies that go into one single portfolio, but it’s more homogeneous, if you like. Yeah, more kind of same building, exactly. Same building block, but different mix and match for different clients.
Jeff Malec 29:54
And then my last question, maybe, but I’ll I’ll say it’s my last question on the risk premium. Say I wanted to build these all myself using Cloud Code or whatnot. Where where does that lead me? You think it is achievable, or I end up with just an implementation problem, right? Like I feel like that’s the people are going to start to say, “Well, I don’t even need these banks now. I can just build this risk premium myself.
Jiro Fujisawa 30:18
Right. It does a good job in trying to get the the overall gist of things, but I think kind of repeating what I mentioned earlier, the the devil’s in the detail. Those details, Claude may not, or you know, any sort of AI bot may not be able to capture, and that’s where exactly things blow up in your face when when when things don’t go right. So it becomes an issue of accountability. Do you want to risk your capital in doing in doing it in taking that approach? That’s probably where where the cost of you know relying relying on on a AI agent, yeah.
Jeff Malec 31:02
What’s a bet on is the public knowledge of everything that could go wrong public, or the secrets of what can go wrong internal to a group like you, right? That hasn’t been published out there, but has been fixed and corrected, and that and right, you know where the pitfalls are, yeah, and maybe didn’t publicize all that,
Jiro Fujisawa 31:21
right?
Jeff Malec 31:23
Interesting to think about. All right, we’re going to rename the pod “Bury the Lead” because you started a new strategy, and now we’re just getting to it. 30 minutes later, inside asset management one, cross factor asset strategy, the new strategy. Give us the 30,000 foot view of what what you’re doing with that, and then we’ll dig in some more.
Jiro Fujisawa 31:52
Sure. So it’s actually called a cross asset factor alpha. You missed the a at the end.
Jeff Malec 31:57
Yes,
Jiro Fujisawa 31:59
we’ll call it Kaffa, Kaffa acronym. So, sort of rewinding, setting the background as a management one, we’ve been running a suite of products that that invest in risk factors. What I mean by that is instead of investing in ES or TY or G C or specific markets, we sort of try to decompose the investment universe into the different components of risks, and those are called risk factors. So each risk factor is a a basket of assets. It’s a long short combination, and it changes over time. Not every day we try to stabilize it, but it’s essentially a a basket of assets. So we try to trade those baskets, and we’ve been doing this for a long time. And but there were certain constraints in some of the existing products where you know they these funds had to be long biased, so that we’re not overly short at any point in time, just by the sort of nature of the client demand at that point in time. Now, while that was, while we have been doing that on the other side of the pond in New York, we’ve been sort of doing the you know quant CTA type strategies where you know we’re long and short. We’re we’re market. We try to be market neutral, and so we we sort of combine the two worlds of this risk factor investment and the sort of long short CTA type approach, and came up with the new concept, which is where what Kaffa is. So we look at an investment universe of about 80 different futures market. We try to decompose that investment universe into a finite set of risk factors, and then we invest in these risk factors in a long short manner, how do we do that? We look at various risk premiums, risk premia ranging from trend, looking at skewness, and also carry. So, what we try to achieve in doing this is, firstly, we get orthogonal bets by the fact that we’re investing in these synthetic assets, which are by construction orthogonal to each other. When we do the risk factor extraction, we design it in such a way that the we’re extracting orthogonal factors, and then the second layer of diversification benefit that we get is that we’re taking multi-factor approach. We’re not just doing a trend play or just doing a carry play. We’re taking multiple approaches.
Jeff Malec 34:54
All right, a lot to unpack. Let I’ll start with the risk factor. Do you have examples of? Of the risk factor is like liquidity, or give me a couple examples of different risk factors. We
Jiro Fujisawa 35:06
try not to put names to it. We we try to stay a little bit more statistical at that point. We don’t want to sort of inject our own view as to what is driving markets, especially when it comes to a sort of wide range of assets that you know the 80 different markets we look into ranges from equities to bond futures to FX and commodities and especially commodities you know there’s a whole wide range of things so we want to keep we want to remove some any sort of bias so typically you know PC one would be like sort of the growth sort of factor, if you look at general investment,
Jeff Malec 35:45
and how how many are there again? Risk factors.
Jiro Fujisawa 35:49
So we would go up to about 80, same as the breadth of the markets. Obviously, that
Jeff Malec 35:56
you got to change that because that really confused me in Austin, and we’ll we’ll see.
Jiro Fujisawa 36:00
I see,
Jeff Malec 36:00
right? It’d be better if we had 100 risk factors in 60 markets. Then we could keep those. But it’s so it’s 80 risk factors, 80 markets.
Jiro Fujisawa 36:08
Yeah, which I think so. That’s just a more one to one, right? It’s yeah,
Jeff Malec 36:13
yeah,
Jiro Fujisawa 36:13
yeah.
Jeff Malec 36:15
And so that risk factor, call it growth, and I’m just for simplicity’s sake going to call one liquidity, but you’re saying it’s just a statistical, like we were talking about before. I’m going to run my analysis, and there’s 80 distinct factors here, and that’s before putting them into the synthetic market. So that’s after the synthetic market. So these are the synthetic assets. So let’s say the first factor, risk factor, is growth. Growth is, you know, maybe it’s long equities and short bonds, and many. It’ll have 80 different positions across the different markets. That is the synthetic asset. Synthetic asset number one. So each risk factor is a synthetic asset. Ah, that’s why it’s 80 to 80. It’s one. They’re the same thing, just kind of viewed viewed separately, viewed at a different angle. Okay, so then you have all all 80 are in each each bucket, all 80 markets. Yep, that’s crazy. And so, got it. So in bucket one for growth, maybe I’m long stocks, short bonds, short gold, yada yada yada, long crude oil, something like that. Then factor 71, I might be a different mix of those, or I will be a different mix of those, right? Short equities. So at the end of the day, is this all 80 are working together? Is it weighted amongst the 80, or they just all offset in some way, and whatever you’re left with, the net is the position of the portfolio.
Jiro Fujisawa 37:44
So when I look at the portfolio, there’s three different buckets, three major different, three different major buckets: trend, carry, and skewness. So if I just start with the trend, the trend bucket looks at the 80 synthetic assets and says which one should I buy and which one should I sell and how much. So trend may say okay, synthetic synthetic acid number one looks trendy in the positive way, so I’ll buy that and then you know synthetic acid number three is trending down so I’m going to short that, and do the same for all 80 synthetic assets, and then you have that trend bucket. The trend bucket, you know, technically has a little bit of a different sort of trend measure within it, but we’ll sort of leave that aside for some
Jeff Malec 38:38
posts. And so that different question. So each of the synthetic markets, you’re building a a record of its price movement.
Jiro Fujisawa 38:46
Correct.
Jeff Malec 38:47
Yeah. So then that model, all three of those models, trend, carry, skew, are going on top of the synthetic price action, and then I can see, okay, this synthetic one is trending up, synthetic three is trending down.
Jiro Fujisawa 39:00
That’s right,
Jeff Malec 39:01
and so sticking with the trend. Well, can it be 80 long trend, zero short, or just in practice? No, it’s always going to be somewhere around 50-50.
Jiro Fujisawa 39:11
So we do apply a constraint at the end to make sure that we’re not overly betting on any synthetic acid, or also right at the the end, we also apply a market level constraint where you know we sort of transpose from the risk factor or synthetic asset space into back in the real world specific contract base. We also don’t want to pile on to any sort of single any single market. So there’s multiple levels of caps that we apply, but even at the sort of synthetic asset level, we try to try not to take on too much directionality. But specifically for trend, because it’s a directional strategy by nature, we do we could be long biased or short bias. At any any point in time,
Jeff Malec 40:01
yeah. And depending on each of those models, and then is it is it hard? It’s hard to explain, right? It’s like how does a not even how it works, but I understand how it works. But now moving forward of like, okay, what am I rooting for? Like it’s hard. A trend is hard enough to explain what you’re rooting for to people, but now you’re like, well, it’s the synthetic is long short these different things, and we’re rooting on that. Yeah,
Jiro Fujisawa 40:26
I think that’s what sort of it’s difficult to explain, but because it’s difficult to explain, it’s it sort of captures a whole new dimension of yeah of price price actions, and we’re not trying to overcomplicate things to confuse people, obviously, or to sort of just for the sake of making something new, because statistically, or you know, at least each each process that we apply goes back to our philosophy. It has to be intuitive. So, as a whole, it may be complicated or a little bit difficult to explain, but every decision that we make ultimately leads to us, or we believe that it allows us to achieve much better orthogonality and stability in in the long run.
Jeff Malec 41:16
Right. That’s for my friend George. Orthogonal is just basically non-correlated, right?
Jiro Fujisawa 41:22
Exactly. Yep.
Jeff Malec 41:23
Yeah. Bunch of unique bets, but so it’s really diversification on steroids, right? Because each of these, right? You’ve completely eliminated single markets. So now these synthetic markets, each of that is a more diversified version of the underlying what’s inside of it, right? So each of those buckets have less volatility than, for sure. Right. Than the sum of the markets inside of it. The
Jiro Fujisawa 41:48
volatility of each component, it’s it’s uniform. I obviously within the portfolio you can allocate more vol or more more risk than others, but each unit, when we look at these, each
Jeff Malec 42:03
await them to be equal.
Jiro Fujisawa 42:08
Or when we look at each individual synthetic acid, it’s we sort of compare it on a sort of equal vol level. But when we how much we include it, really, it’s driven by the appeal appealingness from each you know trend carry or skew lens,
Jeff Malec 42:28
and then those trend carry skew are equal. Third exposure is equal.
Jiro Fujisawa 42:33
Yep,
Jeff Malec 42:34
and then each of the 80 synthetics is equal. See, right, so that
Jiro Fujisawa 42:40
would vary,
Jeff Malec 42:41
risk weighted equally, or if you’re like model 17 has been killing it for the last year, we’re going to overweight model 17 or synthetic 17.
Jiro Fujisawa 42:52
So if I if I take the trend model, it would look at synthetic asset one to say 80. the trend model will allocate more to asset number one, synthetic asset number one versus synthetic asset number two. So it won’t always have the same vol or same allocation. It’s more
Jeff Malec 43:15
trendy and
Jiro Fujisawa 43:17
exactly okay, and we don’t have any. So it’s like prior view,
Jeff Malec 43:21
and then so we mentioned these. So the the 80 synthetics change over time.
Jiro Fujisawa 43:29
Over time, it does.
Jeff Malec 43:30
How often is that?
Jiro Fujisawa 43:32
So we observe every day, and we adapt. We trade on it every day, but it doesn’t mean that we’re flipping around the the positions every single day. It’s gradual. We look at a long enough period, not too long, but not not too short set of time horizon to be adaptive, but also not be too sticky. So you know, if there’s a change in the market structure, we would be able to capture that, but we would incrementally change it day by day.
Jeff Malec 44:08
And something like I’m thinking back to like Swiss franc depegged or something. If you had that in the portfolio, that would be something that would like now that vol is 10x what it used to be. Right, time for a change.
Jiro Fujisawa 44:20
So that’s when you know those kind of scenarios. I think it’s it’s very difficult for a statistical model to adapt to, and that’s where I think our job as a portfolio manager step in to make that assessment. We we’re fully systematic. We don’t want to impose any of our judgments within our portfolio. Obviously, there are at the stage of designing a portfolio or even a strategy. There’s a human being is building it, so there’s definitely some human input. But once we built a model, we don’t override any decision unless. There’s some extreme market event, and we have to step in from purely from a risk management standpoint. And I think that sort of you know Swiss franc deep packing, or even a little bit more recent the the ruble sort of liquidy drying up. That those kind of events we would intervene, and we don’t want to be stuck with the positions that that we can’t get out of or can’t get out with a reasonable sort of spread. So only in those times we would intervene and step in and make changes. But we don’t take it lightly, and it’s not a simple exercise, especially when you have a lot of models that contribute to whatever positions that you have, if you just blindly just rip out that position, you may be exposed. If you’re doing some sort of relative value play on on that position versus another, you’re sort of susceptible or expose yourself to certain beta risk or essentially changes
Jeff Malec 46:01
all 80 synthetic markets. All right, if it’s in all 80, but to me, built the perfect thing to not tinker or have opinions on the because I don’t even know what, right? If if synthetic number 14 goes long, like I don’t know which markets are actually in there, so I don’t have to have the the normal systematic manager thing of like, why are we going long oil here? This is just going to reverse as soon as they make a fake piece deal or whatever. So you’ve solved that problem. What’s next? Will you go to 100 synthetic or or what’s in the research pipeline?
Jiro Fujisawa 46:34
No, I think just creating more new. I think one obvious avenue of research is: Are there more factors, not risk factor, but you know, alongside trend, skew, carry. Is there anything else that’s sound that we can sort of add into the into that mix if we were to specifically talk about Kaffa? But outside of the Kaffa world, there are other initiatives, other research topics. But I think for Kaffa, we don’t. We definitely don’t want to sort of increase more synthetic assets. It if if you start trying, if you if you start trying to expand on that more, you sort of incur this instability, and also you’re sort of adding on a lot more noise than than value, so I think we’re happy with where we are in the synthetics assets.
Jeff Malec 47:26
And then how did which did we cover before? Like how did you arrive on the 80 in the first place? Like that was you ran the component analysis and came out at to 80.
Jiro Fujisawa 47:36
So it’s more to do more a sort of mathematical exercise of how much breadth can you get with with n number of assets? Usually, that sort of if you go way beyond n, so if we have 80 different markets, if you try to extract 200, yeah, or you know something ridiculous, then you’re picking up a lot of noise, and you know, just trying not really extracting anything.
Jeff Malec 48:05
Law of diminishing returns, essentially.
Jiro Fujisawa 48:08
Yeah. So I think 80 was its limit, but you also don’t want to just pile on to like two or three different things.
Jeff Malec 48:17
Love it. I think I’ll see you here in Chicago in October, Rent. You’re coming to the conference.
Jiro Fujisawa 48:32
I believe so. I’ll have to check with definitely with Mark on that. But it’ll be nice to to do a trip there
Jeff Malec 48:39
for sure. Any last thoughts? Want to leave
Jiro Fujisawa 48:43
us? No, I think you know we’ve sort of. It was a great pleasure to be on this. Thanks again, Jeff, for having me. We’re, I think, Quant in general has had a a tough beating over the few years. Finally, it’s becoming a sort of field where people have started to revisit, and even though the stock market has been very bullish, despite what it’s all what’s going on, you know, I think it’s people sort of started to look at if and when markets turn around, how do we sort of diversify our portfolio out of sort of the PEs and the Mag Seven or AI stocks. Yeah, so I think it’s. We didn’t cover that.
Jeff Malec 49:28
Like the this is supposed to be an absolute return vehicle, right? Doesn’t care what the market’s doing, doesn’t care what the economy’s doing. It’s gonna do what it does regardless, right? Would you put it in that bug? It’s more of a absolute return versus a crisis period performer or something like that. Yeah, love it. Well, it’s cool. I think you’ve nailed it, right? It’s like a systematic multi-strat, systematic quant multi-strat. We’ll come up with some some cool words for it. All right, Gerald. Thanks so much. We’ll see you soon. Okay, that’s it for the pod. Thanks to Jiro for coming on. Thanks to Jeff Burger for producing. Thanks to RCM for sponsoring. Drop us a comment. Drop us a note. Invest at RCMAM.com. Check out that new website, and we’ll see you next week. I’m not sure who we’ll have. Maybe Jerry Parker, Trend Royalty. Maybe a newer commodity manager. One of those too. Peace.
This transcript was compiled automatically via Otter.AI and as such may include typos and errors the artificial intelligence did not pick up correctly.
