“Keep an open mind. But not so open your mind falls out.”
My guest in this episode needs little introduction: Cliff Asness, co-founder and managing partner at AQR.
Cliff has done dozens of interviews, podcasts, talks, and fireside chats over the years. He is also a prolific writer. So, my goal in this conversation was to try to find the questions he hadn’t been asked before or had not answered himself already.
How did his formative experiences in the dotcom bubble shape his perception of markets? Why should we stick to factors like grim death? Which of his dozens of papers have been woefully overlooked? Where has he changed his mind over the years and what is he most confident in going forward?
Cliff is fountain of knowledge of quant history, research, and practical experience and tells some fantastic stories along the way.
Please enjoy my conversation with Cliff Asness.
Corey Hoffstein 00:00
All right, let’s do this 321 Hello and welcome everyone. I’m Corey Hoffstein. And this is flirting with models the podcast that pulls back the curtain to discover the human factor behind the quantitative strategy.
Corey Hoffstein Is the co founder and chief investment officer of new found research due to industry regulations he will not discuss any of new found researches funds on this podcast all opinions expressed by podcast participants are solely their own opinion and do not reflect the opinion of new found research. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of newfound research may maintain positions in securities discussed in this podcast for more information is it think newfound.com.
Corey Hoffstein 00:51
My guest in this episode needs little introduction. Cliff Asness, co founder and managing partner at AQR Cliff has done dozens of interviews, podcasts, talks and Fireside Chats over the years. He’s also a prolific writer. So my goal in this conversation was to try to find the questions he hadn’t been asked before, or had not answered himself already. How did his formative experiences in the.com bubble shape his perception of markets? Why should we stick to factors like grim death? Which of his dozens of papers have been woefully overlooked? Where has he changed his mind over the years? And what is he most confident about going forward? Cliff is a fountain of knowledge of quant history, research and practical experience and tell some fantastic stories along the way. Please enjoy my conversation with Cliff Asness. Live Welcome to the show. Very excited to have you here today. I will tell you in preparation for this, I think I listened to four or five of your interviews, I think I might be able to tell some of your jokes, maybe as well as you now. I think I’ve memorized most of them.
Cliff Asness 02:05
Well, I do tend to repeat them. They’re good jokes, they
Corey Hoffstein 02:09
get a good laugh out of the audience every time. But I did do that in hopes that I could try to come up with questions here that you’ve never been asked before. So I’m trying to Yeah, I generally
Cliff Asness 02:19
don’t listen to myself and probably won’t listen to this. I don’t want you to take that the wrong way. But whenever I do listen to myself. It’s not the sound of my voice. No one likes the sound of their own voice, myself included. But I get very obsessive with what I should have said, which is highly correlated to what I said. But I could have said it better. And it just irks me too much to listen to.
Corey Hoffstein 02:41
I don’t listen to my own podcasts either. So neither of us will listen to this once it’s out.
Cliff Asness 02:45
That’s right. Hopefully many other people but not URI.
Corey Hoffstein 02:48
Hopefully. Well, let’s dive in. I want to start back at the beginning with you back when you launched AQR very famously right in the teeth of the.com. Bubble. And if my facts are correct, I think you guys were sort of a target 20% Vol fund very quickly having two standard deviation type sell off, which is a pretty meaningful drawdown in real drawdown terms, but ultimately proven right when you came out the other side, I am a big believer that people’s experiences are ultimately shaped, especially formative experiences ultimately shape their actions later on. As you sort of look back at that early history and the success you had sticking with your strategy. How much do you think that informs your mentality of just clinging to systematic strategies with quote you this clinging to them like grim death,
Cliff Asness 03:38
it was a little bit north of 20% Vol, which was an error. By the way, when we were at Goldman Sachs, and people always refer to us as hedge fund. And I’m always like, today, it’s something like half or more than half our assets, our traditional benchmark at Goldman Sachs, it was like six sevens of our assets. But it turns out that a whole bunch of 20 and 30 Somethings starting a traditional asset manager, you’re told, come back when you’re old, when you say we’re starting a hedge fund, and we’re closing people back then we’re like we’re in. So it was kind of the path of least resistance. We did basically a similar thing at Goldman, we ran one very aggressive fund. Here’s where I was a complete idiot in going around to raise money in the first six months. I must have had 20 clients say we don’t need 22% Vol uncorrelated to the market. How about you do a quarter of that? And I responded, in a flip. I know you find this hard to believe but with a flip tone, you want a quarter of all give us a quarter of the money. And I still refuse to admit I was wrong about the math because I was not. But I was staggeringly wrong. Because it turns out Cory, this will shock you that down to standard deviations at 20% upsets people more than down two standard deviations at 5%. And the con have odd parties, it really shouldn’t. The statistical event is a statistical event. If you gave someone a quarter of the money, because they were more aggressive, you lost the same amount of money. Math is on my side. But having served on many Investment Committee, since that we do not call in the largest standard deviation event, we call in the idiots who are down 35%. And it’s just how the world works. So it is an apropos question also now, because history doesn’t repeat, but it does rhyme. It does feel like we’re in a very similar time to maybe late 99, early 2000. We did start in August of 1998. I think the tech bubble is really taking off after LTCM collapsed, starting in September of 98. Ironically, we were thrilled our first month, because a lot of people didn’t distinguish between us and LTCM. And I’m big fans of some of those guys individually. And I don’t mean to be ex post about that. But we knew we were doing different stuff than them. So we were up in August, when markets were down 22% And LTCM began was kind of a three month end game. And so we were like, oh my god, proof of concept. Not that a month can be approved, but a nice baptism to begin with. And we smiled too quickly as anyone in markets always does. Because the next 18 of our first 19 months, we didn’t lose money every month, but we lost a lot of money, particularly cost of what’s still a big part of our process, the value factor are very similar period. You know this, we’re far from just value investors, though it does look like we’re value investors when value has giant standard deviation event. Just like in early 2009. When markets reversed and momentum got crushed. You look more like a momentum investor. I bring up all these bad things. It makes it sound like it never works. It has worked over the long period very well. But whenever extremes happen, you tend to look like that. We did stick with it. But very similar to today, we spent a ton of time, I hoped with an open mind saying might we be wrong. We were the first back then to do this thing that people all casually referred to as the value spread. Like a lot of terminology, I don’t think we use that term we may have I don’t think we used it in the, in the paper. Very similar to stuff I’ve written now the papers data ended in November of 99. And I remember three months later presenting because our conclusion was cheap versus expensive. We were at great extremes. Things were very cheap and very expensive. Higher than any time in history. The three year going forward prospects for value were very good. And three months later, I had to say no paper has ever been this wrong that quickly. The paper thankfully, we did forecast three year horizons. And it was dramatically right when it all played out from the time we wrote the paper. But we all know being right and surviving are not always the same thing. We did learn even shudder to say we’d learned a lesson. I think if you asked us before, do you have a five Sharpe ratio we’d say no, we have a good Sharpe ratio call it one for want of a nicer round number. We often say if it’s above point five in an uncorrelated process, everyone would want that in the portfolio the markets not a point five Sharpe rates about point four Sharpe ratio if you could find another stock market, uncorrelated, which you can find on the stock markets, but they’re pretty correlated to the US find another one uncorrelated with the same risk premium, you’d be excited. Having said that, you take a point five to one Sharpe ratio itself, in a normally distributed world has tremendously bad periods. In the actual world, which is certainly fat tailed. Everything has I remember, fourth quarter of 96. And I’m dating myself a bit here literally, it was better. It was a positive four standard deviation quarter. We’ve seen negatives. We’ve never seen 10s or 20s. I don’t think we will, I don’t think it is I don’t want to beat on LTCM. But I don’t think it’s that kind of an issue. But anyone who is in finance or doesn’t think the world is somewhat fat tailed hasn’t been paying any attention. You may have noticed I’ve had a little bit of a Twitter fight with someone who seems to think that we don’t know the world is fat tailed. That particular fight is about whether it’s priced into Options enough, but my dissertation adviser gene fama co with Ken French, I believe his dissertation on the market being fat tail that was a shorter horizon. So we’ve known that forever. So you sit on top of a modest but we think real Sharpe ratio, you got to think it’s real. And then your job really becomes to defend it, stick with it, and realize that Sharpe ratio over the long term, and that is just far harder, so I don’t think before 19 On, we would have said something dramatically different in terms of the facts. But a lesson on how hard it is to stick with something that you absolutely believe is true and real. Not just for clients, but even for you. That lesson, and I guess, having seen it be successful, and we had it was a different set of circumstances. But a similar experience with convertible arbitrage, and value for part of the GFC. A lot of things were fine in the GFC. But those two were not and we stuck with them, they look super cheap, and again, came back. So experiencing that a few times, make sticking with it like grim death harder. But you also have to acknowledge it can bias you the other way to maybe this time is different. I think 99 out of 100 times the world has not changed as much as people think. But that doesn’t mean one out of 100. It does. And if you’ve been successful fighting that, maybe you’re too stubborn. So we try to fight that, too. We try to examine all this stuff with as open mind as we can. I later found out this wasn’t her quote, which was disappointing. But the first place I heard this was my Nebraskan mother in law who said an open mind is a great thing, but not so open that your brains fall out. And I don’t even know who said it, but I think she got it from somewhere, but it was perfectly applied. And that’s kind of what you got to do with an investment process. You can’t have your mind so open to every new idea that you’re willing to trash. What has worked for 200 years and for you for 20 years. Because it’s not done well for a while. That’s way too open. But if you swear that you’re legally and morally entitled to the value premium for eternity? No, I’m not don’t know if I actually answered the question. But I talked for 11 minutes. So I’m, I’m filibustering your podcast,
Corey Hoffstein 11:55
that’s perfect. They actually kind of nicely leads into my next question. Because you’re talking about this concept of having an open mind evolving your thinking over time through these experiences. I’m always curious with people who have been in this industry for a while, if you had the opportunity to go back to when you started AQR getting a time machine, go back, pull yourself aside, don’t create any time travel effects and issues like we would expect. But you could just say something to yourself, whether it’s a sentence, or to give yourself a piece of advice, maybe other than, hey, invest in growth for the next 12 months. What would it be?
Cliff Asness 12:30
It’s funny, I asked myself those questions too. And I can see you’re struggling with it too. Because you have to answer in a realistic way. Go back and give me the Wall Street Journal every day for the next 10 years is not a fair answer, I probably would still have a lower Sharpe ratio of the Jim Simons, but it’ll be pretty good. You know, I’ll get philosophical here. I would like to tell myself, it’s all going to be okay. Because if you do unreasonable things in investing, and you are sticking with them, it will be okay. This is why I’m somewhat biased not to change our minds easily. The only way to lose if you’re right long term is to get too cute with what you’re doing. Let’s say there really is value premium, and a momentum premium and low vol premium. And they’re absolutely true. If you jump around between them, if you try to time them. And you know, we’re we’re kind of famously reticent, but not absolutely unwilling to do this. But we don’t like. And one of my main reasons is you can take something that is absolutely in the world, I just said it’s absolutely true. So I’ve done so you can’t do an investing I’ve guaranteed long term work, but only for the hypothesis, you’ve taken something guaranteed to work and you’ve introduced the possibility that will not work for you, because you will time at exactly wrong. So I would tell myself, I might even shorten it to calm down. Because that could apply to my non investing life also and be even more useful.
Corey Hoffstein 14:04
So I want to talk about again, with this idea of keeping an open mind you very famously and your collaborators at AQR have very famously, than prolific publishers of your research over time. I think SSRN credits you to something like 29 different research papers as the author which my guess might actually be just the tip of the iceberg if I’m but you’ve won a number of awards from the FAA, J, the best paper from the jpm for I think five years, really just an unbelievable amount of well appreciate writing. So first, I’ll say thank you, because I’ve been a big reader of the research,
Cliff Asness 14:37
and yet we can still lose gobs of money for two year periods occasionally. The question
Corey Hoffstein 14:41
I really want to ask you though, is which of your papers do you think was the most underrated or the most underappreciated?
Cliff Asness 14:49
I think they’re all grossly under appreciated. There’s a wealth effect. Now. I’m glad you didn’t ask me that the other way because a lot of my stuff is co authored and I wouldn’t want to it would be like pick up a call out there, we wouldn’t be there for It’d be my fault also. But I got to start with a couple that were never published that were just working papers. Because almost by definition, those are going to be under predict not going to get the same circulation. One is actually perfect dovetail with talking about the tech bubble. I wrote a thing called bubble logic. I started writing it in mid 99. I think the final draft was mid 2000, I had the stupid idea to make it into a book. And basically, it’s the first time I tried to be funny in print costs, the tech bubble kind of began the long term effort of markets, impinging on my sanity. So I allowed my personality, or lack thereof to come through. And I think it was well done. And next post, it was pretty massively, right. It was also writing a book I was chasing, once it started to come down. I had new versions saying it’s still cheap. And finally I’m like, Alright, we made a ton of money, and it’s halfway back. And there’s nothing to publish here. So it always makes me sad that one of my best ex ante predictions, which was real, didn’t get published. Similarly, another paper that was unpublished was a very early work. first draft of this was 1994. And I even forget the title, we changed the title around a few times. But it was really one of the first papers, probably the first unpublished but public paper to look at how the various factors in investing, and back then it was really value momentum and size. We weren’t even looking at low risk at profitability and other other things, how they do for two separate questions like the fama French work, most of the academic work, just sort stocks. And that means you’re taking both an industry bet, and a relative value within the industry bet. And we separated it too, with a fairly neat way to do and I thought, and we found that most of the juice and particularly for value investing, came from within industry, like value did not do well at all, for choosing amongst industry. So that doesn’t mean you couldn’t come up with a better indicator for it. That doesn’t mean an active value manager. I’m often asked to contrast systematic with active doesn’t mean they could do it, perhaps it doesn’t work, because the accounting comparisons are very, very hard. And perhaps they could come up with a better way to do it. But this is a little bit of a story to my co author on this paper, quit at Goldman Sachs to go start a competing firmness. I then did the same thing. About a year later. He and I we weren’t I don’t think there was anger. We weren’t not speaking but we were not speaking. When I say that. I mean it passively not speaking, it wasn’t like I wouldn’t talk to that guy. But we just didn’t talk. And it was an awkward thing. So the paper languished. And by the time we got back to it, it was old news. So I certainly have a competitive streak when you think you’re the first to something and not to get it published. And something that has become a big part of our process. And I think a lot of other quants do to overtime, that one, and if I could pick a third, that’s an odd choice. And this one was published, I think in the JFe. Though, I really should remember this after all these years. I don’t always remember the JF versus the JFe. Don’t tell the publishers that they’ll never accept another paper from us. But a really paper I love called betting against correlation. You know this well, and I know listeners to your podcast have probably heard it a million times. But low vol low beta strategies, they tend to outperform in a risk adjusted sense, not necessarily always in total return for return for the risk, sometimes a total return. But for the risk taken famously, two of my colleagues wrote a paper called betting against beta, which they fully credit Fischer black, they resuscitated Fischer blacks findings that low beta outperforms, it’s become a real kind of staple of the industry. But one thing people are still fighting about print fights, academic fights, is why. And there are various stories for it. And it occurred to us at some point, that if the story is one that we tend to favor, that is people don’t like to lever. It’s about beta. If the story is like lottery effects, it’s more about vol. And it’s pretty hard to distinguish these typically, because they’re correlated, we realize that beta is simply the sum of all in a correlation. So we literally separated them and built separate factors. And we discovered, and I’m sure because you’ve read everything, you know, the result that it’s really coming from both very strongly, they’re both kind of almost equal contributors. In fact, correlation maybe was somewhat more after adjusting for factor exposure. And we think there are no proofs in our world, there’s simply the probability start swinging more your way. But I kind of liked this because it’s very pure paper. It has very little practical implication, it doesn’t really change what we do. Coming from both means beta, still a pretty good summary of the whole so it’s kind of pure research, but We are kind of obsessed with not just doing something, because it has worked in the past was trying to understand it. So I think that is one of these cute, almost little papers that doesn’t have some earth shaking result. But I really love that one just because it’s so simple. I really like simple. And when you can do something simple and find something cool. I like it. And few people compliment me on it. So I know it’s under appreciated.
Corey Hoffstein 20:25
Well, I thought it was a great paper. So there’s another compliment for you. But it does teed me up really nicely into my next couple of questions where I will do want to dive into some of these research concepts with you less the minutiae of the research, but more again, towards this idea of evolving research over time, you must have read my mind about that second paper because I was going to use that one as an example. So from my perspective, it seems as if there’s been a growing trend over time to continue to evolve factors to avoid those sort of unintended bets, right to go from that very simple, naive, cross market sort. Okay, we’re doing industry sorts, we might then start to customize the type of characteristics we look at in a given industry, depending on how those industries work, and you start controlling further and further and further. At what point, in your opinion, does this just devolve into sort of outright discretionary investing?
Cliff Asness 21:19
There are a few different dimensions you can answer this on is discretionary versus systematic. But there are other ways to think about this data mining versus whatever the opposite of data mining is being more driven by logic and saying simple. At first, I’m going to disappoint you. I’m sure you know this, but there’s no bright line test for this. This is inherently subjective. Having said that, I have for years said there is no process that doesn’t have a ton of discretion in it. A systematic one uses the discretion in the building of the process. To us non discretionary means we listen to the models, and a lot of people will have cases where they might override maybe risk cases if the world is going mad. But the vast preponderance of the time, we simply follow the models. But that doesn’t rule out rather large discretion in how you design your models. It’s funny, then this is why I say dovetails with a data mining concerns. That is very justifiably a major concern in our I don’t know we in industry, in our field. Having said that, if you want to take old discretion out of a process, you totally can because you have a prime mover problem in setting up the process, but you can push it down pretty hard. You can have a process choose factors and rules and whatnot. But then you’re the ultimate Data Miner. And I think most of us not necessarily all but most of us think that is a pretty scary way to approach things you find patterns that don’t exist. So the simple answer and discretionary is no matter how bespoke the model and how many bells and whistles you’ve built in. If you follow it with a decent amount of rigor you are being a non discretionary manager. A separate question is how to decide when too many bells and whistles are too many. And there you’re coming down to decision and all investing is almost always the same. It’s never do I believe this that’s binary. It’s how much do I believe this? And how important is it. So if there is something within industry value being better than across industry value, that result is held up for the 26 year out of sample period. Since we built it we did it first only in the USA because we’re all jingoistic and start with actually we probably do that because we have much longer term data in the USA, but it is held up outside of the USA. So we thought it made sense. Also, we thought we could understand momentum was the least affected by industry effects, because it’s really easy to measure across. So in deciding whether that bell and whistle makes sense is doesn’t make economic sense. And what out of sample evidence do you have, and you just repeat this process again and again. And they are discretionary choices. And people do differ on how, for instance, what value measure to use, we do very little tinkering of that with that across industries more than a little bit for REITs. REITs are a case where we think on first principles, some of them just don’t make sense. Some in Financials, they famously a lot of the research, either excludes financials or comes up with another way to handle financials. They’re far more levered companies often than, than others. But by and large, we don’t do a ton of that. You could literally run a process across. I think last time I looked, we are 25 in our broadest, most Alpha oriented value composites. We had 25 measures. Now actually, I know it’s 25 because last year I was presenting to a group in Europe actually, and I mistakenly said 15 Because I’m old and don’t remember everything exactly right anymore. And one of my partners have fallen in Lukash was in the audience, and I love Lukash. because he raises his hand while I’m talking, it’s not the q&a period. And he works with me and says, Actually, Cliff, it’s 25. And I use it as an occasion to tell the crowd that I will be rewarding not punishing this behavior, just so you know, but it is a little odd. I don’t think 15 versus 25 was much of the point, I don’t think I was misleading anyone. So discretion, you are a discretionary manager if you come in, and fairly often, or even anything, but rarely, say, I’m going to tweak this, I’m going to tweak it. Invariably, you’re a bad discretionary manager, if you’re always doing that, for what would have worked better lately. As an aside, I do think the only five Sharpe ratio strategy I know how to build Well, something that traded way too much, you could have a T cos negative five. But true gross, negative five, is trying to figure out what has worked over the last what has hurt you in particular, over the last few years, and then doing the opposite. If you try to test that and build a systematic virtue, you will not be able to build there is some mean reversion. But it’s not a negative five. I think human nature, my personal subjective experience, is every time I’ve succumbed to that tendency, which I’ve been very good about for many years, but wasn’t as good early on. It’s kind of a disaster. So I do think discretion is about not whether or not you use judgment, another almost a synonym for discretion in building your process. It’s about how often if ever, you override your process, and there, you can get as complicated as you want. Not that you should we still try to keep it fairly simple because of data mining concerns. But you can get as complicated as you want still be non discretionary. If you’re following the process.
Corey Hoffstein 26:50
I’ve always sort of said, If you don’t override the process, and you call yourself a discretionary manager, you’re really just a closet systematic manager, because then you can codify it. So I would argue if you’re a discretionary manager, the only thing you should really get paid for is overriding the process. You’re sort of idiosyncratic bets.
Cliff Asness 27:07
Yes. And in that Calvinistic sense, we’re all non discretionary managers.
Corey Hoffstein 27:11
Absolutely. So I want to stay with a thread that you started to go down a little bit with your last two answers, talking about these ideas of having the first principles of not just necessarily leaning into the data and data mining, but having some economic and behavioral theory behind it. And that’s something that you and AQR seem to have a very strong preference for factors and styles that are not only well founded in the data, but have a strong corresponding economic or behavioral rationale behind them. But I’ve also heard you say things such as well, if value stopped working, it wouldn’t be a negative performer, it would just be random grocery costs. So I guess my question to you is, if there is that asymmetry in picking a factor that doesn’t really work, or a trading strategy, that doesn’t really work. And your downside is random noise plus trading costs, but your upside is alpha, why not risk more type two errors, why not lean into factors and things you find in the data that are maybe a little less clear cut,
Cliff Asness 28:12
you want to know something really embarrassing doctoral program, 2530 years ago, one of my prelim exams, I decide to take in statistics, I still have to think through and remember, which is type one and type two, every single friggin time for 30 years, it’s quite exhausting,
Corey Hoffstein 28:29
you would think they could come up with a better phrasing than type one and type two,
Cliff Asness 28:32
yeah, something that indicated what was going on. But it’s another really good and really hard question. First, you have to distinguish what we yap about in public and write about constantly from what we might do internally. I don’t want to get too much into CQRS business. But very broadly speaking, we implement factors in two different ways. And one way we call styles, which were honestly telling people were doing things that are mostly publicly known not that we’re, we’re not above a little tweaking, if we think we have something that’s really easy and high capacity, and, and can make it a little better. But it’s really exposure to some what we think are great long term styles. But it is not alpha, and we charge a lot less for it. It’s a different concept. And then we have kind of full model versions, which are still very exposed to the factors, one dream every five years. We say what if we hedge that all the factors can only had our own skill, and I have to reprove to myself that we’d have to lever that another nine times to one because one of the ironies, which is an interesting thing is, the more risk you take out, the more leverage you need. Not to segue back to what we were talking about before, but we wrote another paper, extending the industry concept to low risk investing. Whenever people talk about low risk investing, they talk about it in terms of industries. Turns out it works better if you don’t take an industry that But you got to leverage the portfolio more, because you’ve taken out a fair amount of risk. And if the whole thing works because of risk aversion, leverage aversion, that could be why you get a higher Sharpe ratio. So, in our styles portfolios, it really is, again, not above trying to make it a little better at the edges. But it’s largely about doing what we think is a really great version of stuff that is publicly known. And as we’ve seen publicly known stuff doesn’t make it easy to stick with. I actually think that’s a short term hire and a long term positive, because I think it’s why it doesn’t get arbitrage away. Believe it or not, two and a half years ago, the most common question I got was, why doesn’t everyone do this in arbitrage right away? And now the most common question is, how can anyone do this? I wish those people two and a half years ago could ask the question of those people. Now, and you know, I’m not above asking that myself. So in the places where it’s we consider kind of where we’re really, it’s still in pursuit of alpha, not a style, we have things like you’re describing, we do in proportional to our confidence, there’s still not the lion’s share of what’s going on. We don’t stick with the more known economically strong rationale, things because of religion. We stick with it because it makes us believe in it more, when you test something like value, low risk momentum, in not just for stocks around the world, but for picking stock markets for picking bond markets for where to be in a yield curve for credit for commodities, all in their own fashion. And they hold up. And when I say hold up positive Sharpe ratios, not Jim Simons, but they hold up, your confidence goes way up, and almost by their nature, more one off type two error risk things, almost by their nature, it’s going to be hard to get that ubiquitous 200 years back, test, front test, do it in every market, it’s usually a more bespoke kind of thing. So even if you have a story and strong evidence, you don’t have the same ubiquitous evidence. So we absolutely do those things. We don’t think of ourselves as assuming those. We think of ourselves as doing them in proportion to our confidence. But the thing we have most confidence in are some of the things we’ve written about, where do
Corey Hoffstein 32:16
you draw the line between craftsmanship and alpha? At some
Cliff Asness 32:20
point, you’re going to ask me a question that I can actually answer right, as opposed to filibustering and giving you examples, because you know, there is no literal answer. We call craftsmanship. This term comes from a paper my colleagues wrote, they know this, so this won’t surprise them. I hate the term. I told him, I think it’s very pretentious title. Of course, we not immediately, but not long thereafter, we went into a bad drawdown. So you never want to refer to yourself as a craftsman. And then. And then it’s like, oh, what do you build there? So I thought it was a pretentious way to phrase it. And my colleagues said to me, all right, so just the title. And I said, In the immortal words of Ralph kramden, homina, homina, homina, I had nothing, I couldn’t come up with a better one. And it does sum up the idea even if I don’t like the, the tone, but we think of craftsmanship, as alright, you believe in value? Do you diversify across many factors? Or do you stick with with the one you think is best? How do you implement equal weight among liquid stocks that you can do that you can’t do? equal weight in any size for very small stocks? Signal weighting, which is even more extreme, which goes proportional to how strong the value signal is cap weighting, which has the ultimate liquidity, the blend knees trading, how efficiently Can you trade, that’s a craft. That’s not economics, that’s something you work at, and you get better at sort of a lot of ways we think of craftsmanship, as relatively small things, they may add up to something big when you put them all together, that aren’t a new factor, a new factor, I would not call craftsmanship, I call that research. It will probably be alpha for a while. And then we wrote a paper a long time on this, it was just a white paper, things take a journey from Alpha, they can journey all the way to zero if they are very arbitrage. Arbitrage will not actually show that’s a word, but I’m going to use it I think the thing is most susceptible to that are strategies where it’s about moving fast. Those tend to have a half life to them and may be great for a while and you should avoid them. But but they can go away. But others, when they become known are still painful to stick with and move to becoming a style exposure. And the funny thing about craftsmanship is it’s just the same thing. It’s an improvement from a small thing that has a positive risk adjusted return. In this case, each one because they’re minor has a relatively tiny, right if you do a slightly different weighting scheme. You think your Sharpe ratio got better by point oh two. And on first principles you like it and it’s held up in a lot of places. You might do it, but a one Sharpe ratio shocks people with the fact that it can easily have a down decade. Even if things were normally distributed, we’re talking about squared a 10. It’s a three kind, it’s it’s rare, but a point five Sharpe ratio like the stock market or less, it’s even to point point four, that can have a down decade in a heartbeat. So when you get into point O two, if you’re waiting for proof, you’re right. It’s going to take a long time. So we rely a lot on common sense on on as much out of sample evidence as we can get. But again, I am filibustering again, because there is no bright line between craftsmanship and alpha. If the world knows about it, it certainly craftsmanship not alpha. If it’s something that is more implementation oriented, we tend to call it even though it’s not, it’s still alpha, in a sense to the investor, we tend to call it craftsmanship, if something is a new factor, no one ever heard of, we would clearly call that alpha. And then it would take a path of its own through time. But there are no bright line tests. Ultimately, by the way, we all you, me, everyone spends a ton of time trying to put nomenclature around this and trying to put semantics ultimately, we don’t really care. We want it to work over the long term, and cause us as little pain as possible in getting to the long term. But I guess these discussions are important because having a coherent intellectual framework probably gives you some faith. But sometimes I think we may be and AQR is probably half responsible for this in the world. But over beat to death, these distinctions
Corey Hoffstein 36:28
on that line of faith, we’ve been spending a lot of time discussing stick to itiveness. That’s probably not a word either, but I’m going to use it. I’m gonna give it to you. Thanks. And this idea of whether it’s data driven economic rationale, behavioral rationale, having the ability to stay with a factor long enough that you do have a certain amount of faith with it. I’m curious, what is something that you’ve actually changed your mind about recently,
Cliff Asness 36:54
I thought there’s a very low probability we’d have a pandemic. Sorry. I guess one thing we’ve been public about this again, this was a colleague’s paper. But I have moved in the last maybe five years, from thinking the size effect was underwhelming to thinking zero. Which for a TA of jeans, who a dissertation advisor being gene Fama, and I still have great affection and respect for gene to move there. That was a hard one for me, I’ve made the move there slower than I should, because of a prior. But if anything, the original results were overstated. And this is widely known. It’s not the author’s fault. But the Chris, I believe, understated, how bad the listing returns, were, and of course, small stocks the list more than large stocks. So the effect is smaller than originally was. Beta accounts for most of it. And then if you adjust for the fact that small stocks are less liquid, you can do this with simple lag betas. It’s literally nothing for 95 years, it explains a lot of risk. Like if you want to explain the portfolio’s variance, which, as you know, is different than whether this positive mean, yeah, you certainly should include small, but this might sound minor, but for someone with my kind of intellectual roots, it felt big. I have moved from thinking the size effect was small to thinking it’s non existent. Now for fans of size, I’d point out a few things. One, we do think most anomalies, at least gross, how much you can capture net, it’s a different question, but at least grow stronger, in small and I think people mix these two, do you believe in small value, that’s not the same as believing in the small firm, effect and value might work better and small, and that’s completely legitimate. Also, adjusting for beta, small stocks have considerably higher betas than large stocks. And again, particularly when you do this liquidity adjusted with some legs are one of my favorite. And I haven’t written this up yet, I’ll give you a preview. When you do this, at the monthly level, small stocks have a larger beta than large stocks. But when you add in at least a one month or even longer lags, you find it’s even larger. And what happens is they don’t always trade or don’t trade nearly as frequently as large stocks. So when the whole market moves, some of them haven’t traded. So they move in response to prior moves when they finally trade. They really have moved, you just haven’t seen it. But what you go is I’ll make up the numbers that go from a 1.2 to a 1.33. Beta, completely. I’m getting it directionally right, but do not go with those numbers. If you do this at the daily level, which I had never done before, but I got this just from Ken’s website and did did the daily. The beta of small is less than one at the daily level. And it’s the same thing going on. The illiquidity effects are just way stronger at the daily level, the amount of things not trading, and then the legs come in like maniacs. So it’s the same exact thing and you get the same result when you put enough legs in. You don’t have a premium, but I think going to A beta on a lonely small portfolio of less than one or a long short portfolio negative beta at the daily level, and seeing that extend to healthily positive and wiping out the premium. I haven’t written that up yet. If you write it up immediately, I’m going to be very upset. It’s getting published next Monday. I don’t think it’s standalone, a publishable thing. It’s just a neat way to illustrate this. But again, I tend to spend 11 minutes not answering your question to finally bring it back. I’ve gone from somewhat skeptical about the size effect to a complete heretic from, from my upbringing.
Corey Hoffstein 40:34
This next question, I’m a little afraid to ask because I think it might be a little redundant to our conversation. So maybe it’ll be your 10 second answer. But I am curious, as you
Cliff Asness 40:42
have I struck you as someone who gives many 10 Second answers,
Corey Hoffstein 40:45
you might have an idea. You never know, the question, I guess, as you look forward, which strategy or factor do you have the least conviction in? And which one do you have the most conviction in?
Cliff Asness 40:58
Can I answer conditionally on current valuations? Absolutely, take 10 minutes, whatever you want to do? Well, if you’re answering conditional on current valuations, it’s too obvious. I’ve written about tilting more towards value I’ve said, we think market timing and timing factors are a sin, and investing sin, not a moral sin. But we think one should sin a little, rarely, and not a huge amount. Meaning when you do have an opinion and opinion should be coming from something more formal than just an opinion. The way we describe it is we won’t do this unless things are really epic extremes. And then we will not tilt so much that the long term results would be way off of optimal. Not optimal is not just an optimizer, there’s some again, judgment in it. But you have in geekspeak, a fairly what’s called a flat surface and optimization. Imagine you believe in value momentum, low vol and profitability. I’m oversimplifying, but that’s it, the exact mix of those. If you’re 3030 2020, and I’m 2020 3030, we’re gonna have pretty similar means, we’re not gonna be able to tell from long term back test who’s really better, or whose choice is really better. And we’ll have some considerable deviation over short term. So this surprises people, these are correlated portfolios, but we’ll have deviation. So when we will tilt, we won’t tilt so much that that longterm would be off of that flat surface. So if we’re wrong, we’re still within hailing distance of optimal. But now, we’ve written a ton about this. Another version of the same thing is 99. Never been this wrong about this something this quickly. I think I’ve exceeded 99, for how wrong we are in late last year, we talked about tilting towards value because it did look kind of near tech bubble level, cheap, meaning the difference between cheap and expensive, was stretched. And it’s moved to considerably more in our in our measures than then tech bubble level. Cheap. Here’s something that it’s not exactly your question. But if you had asked me in 90,000, will ever see this again, in my career, I would have made the same error many making in investing, I would have said that. So that’s a Black Swan. That’s whatever’s blacker than a black swan. That’s not going to happen. Again, it’s the craziest thing we’ve seen in 100 years of data. And we think if anything, it’s hard to judge, which is more extreme, because back then, the companies that were super expensive, were actually worse companies. Nowadays, they’re not worse companies. They’re just ridiculously priced. So maybe they’re similar levels. But I went to thought we’d see that, again, in my career. We certainly have. But with all that preamble told you, not a 10 second answer. I can’t run away from things I’ve said publicly value, not on a one month horizon, we’ve proven quite clearly that we into my mind, no one, but we certainly don’t know how to forecast that. But uncovered and that same three year horizon, we use the 99. The hell we’ve lived through makes me more confident going forward and should. And if you’re more confident about value, I think you got to be less confident about momentum, not so much that you leave that flat surface, because we can be wrong about this for a long time. But the last part is a giveaway, or is it a gimme? Right? Because if you have two highly negatively correlated factors, and you’re most confident one, it’s gonna be pretty odd, if you’re not least confident in the other. I know
Corey Hoffstein 44:25
you’ve been spending a lot of time discussing and researching and doing deep dives into the Value Factor lately. So maybe I’ll set that one aside. But what other areas of research are you finding to be the most interesting today that are really peeking your personal curiosity? Curiosity is
Cliff Asness 44:41
really the right word. I’m too old to be the guy doing big data and machine learning. We joke very often these are of course, two separate things that people always say as if they’re one phrase, though, often they do work very well together big, unstructured datasets we didn’t have before. We do pursue that Remember, we talked about things that are alpha that aren’t styles we absolutely do pursue, it’s hard not to have that be the most fascinating thing, just because it’s so different. I have a general view that I’m not even sure everyone on my firm shares, that anything you find there will be arbitrage away faster. It’s not the value effect, it’s being the first to a new Parson new data set are gonna have to be literally the first but among the first and, and then that part of the world will get more efficient, I think it’ll be more of a continuing arms race going on. But again, you don’t want me building this. I still program in Fortran. I don’t know anyone working in big data, who really uses a lot of Fortran code. But we’re doing it intuitively. I love it. I gotta be careful. And I want to overdo how much I love it. I’m actually still agnostic on how much it’s going to matter going forward. But I love the study of it. It’s fun, it’s new, it has potential to be really different. But the jury’s still out on whether we or anyone will actually revolutionize the world with us.
Corey Hoffstein 46:05
Are there any areas of investing that you think purely systematic approach would struggle more or perhaps even not work at all?
Cliff Asness 46:14
Mostly no. But you got to pass a few tests, you got to have good data over a long period, and reliable data. And it’s got to be tradable. Most quantitative approaches are not Warren Buffett, my preferred holding period is forever. So anything illiquid, I can’t imagine how you’d use a systematic approach to do venture capital, figuring out why Google is going to kick Yahoo’s butt, which I think happened post the venture stage, but you get what I’m saying it’s pretty hard for me to imagine a systematic process. We’ve dabbled in the private equity side, because there’s, there are at least real companies that have been around. And I’ll keep an open mind and will if I come out two years later and say, Cory, I want to come back on the podcast talking about the AQR private equity systematic fund, I’m saying right now, there’s a relatively small chance of that occurred. But I think it’s pretty darn small. And I’d love to do that. Because while private equity, I live in Greenwich, Connecticut, so I have a lot of private equity friends, I think it’s something the world absolutely needs. Some things it doesn’t it don’t need long short, I think long short actually helps the world I think it makes markets more efficient, more, more liquid. But you can imagine a world without long short products, sadly, enough private equity, there are firms, companies that fitness, and they have to be invested in. So this asset class should exist. I have written quite a few cynical things about people over valuing the opacity. I’ve written about how we may have the illiquidity premium backwards, where the fact that it’s illiquid, and you can’t mark it to market is actually a positive. And I love to say, if we were private equity, we’re not down. We’re not having a value drawdown. Not we haven’t realized those returns yet. And by the way, I don’t think we will be down on a long term horizon. So the counter argument is yes, we know this is just makes us better investors. I’m iffy on that counter argument. It is true. But it’s also somewhat frustrating to those of us and you’re in this world to who are marked to market being actually worse for what we were all told was a feature. Over time. With all that ranting and raving, I’m saying I have every incentive in the world to do a better private equity product that systematic that charges much lower fees, as we’ve done on a lot of the style stuff over time. But I don’t see it as likely ever. And certainly not in the near future with the reasons being the situation’s are bespoke the data is not great, and it certainly can’t trait. So again, if two years from now, I say we figured out a passive way to be the Jack Bogle of private equity, and to do it efficiently and not to have a lemons problem, which you worry a lot about approach like that if you’re the people not doing the due diligence. Alpha is hard, but negative alpha seems to be easier. So I have every incentive to say you can do that I’d love to. I don’t think it’ll happen. So those things, pretty much anything else that’s liquid, I should also mention liquid good data, and enough to diversify across. I literally, it will take a tiny amount of risk and on most unconstrained portfolios and just old school ta direction of the market, but trivial compared to everything else very small. It’s hard to get a lot of confidence in that. So it’d be a really good standalone systematic process, you need a fair amount of diversification. You don’t use a systematic process to pick a single stock all the time. By the way, people ask me what my favorite stock is. You probably get this to the media in particular. First time I ever went on TV. I’ve only been on TV about 10 times they rerun a few of the things so I think people think I’ve done it more often. But probably the first time certainly they did it maybe six or 10 times I’ve been on TV some But he asked me what my favorite stock is. And I’ve started actually discovering what’s in our portfolio. Just so I can answer that when I first was answered, I think I flubbed it a little bit. Over time I use it as, as a way to teach that. We don’t have strong opinions about that active discretionary managers have an advantage, where they get to know everything about a situation and then they take a contracted position, their disadvantages, they got to take a concentrated position. So they really have to be right and they don’t get the benefit of the factor as much because it’s not as reliable. First time I encountered. This was 1996. at Goldman Sachs, I had just started the quant research group, there was a separate quantitative equity group, run by a great guy talking to another great guy who was a discretionary stock picker, the discretionary stock picker was super excited. They had just added a stock for years. I’ve said it’s Philip Morris. I have no idea if it was actually Philip Morris. That’s I’ve been telling the story for years. The story is true. I don’t know if it was Philip Morris. But we just added Philip Morris to the portfolio at maximum weight. We almost never do this. We’re excited about it. He asked the quad. What do you guys think of Philip Morris, and the quant says, In the immortal words, I don’t know. And the discretionary manager looks like him from you know, Mars and Venus kind of thing. But if you have make up your number, hundreds, even 1000s of stocks, longer overweight and shorter underweight, you are betting on average phenomenon, not individual names, it is kind of weird. If you know a lot about the specific names, you can study him, but you’d have to memorize a ton. So oh, so that story to try to give the idea of the difference between the two. But without enough diversification to bring it back. I also don’t think it’s systematic approach. But almost anything else I listed a lot of them earlier, I think is amenable, if your results are real, if they’re based on actual human biases, or actual risks that you get a premium for. They should work for many, many decisions. In fact, that’s one of the criteria we use for weather we really believe I’m fond of saying it’s all started with US equities. And if for some insane reason, we had to only trade US equities, we still feel much better about the core factors, because they hold up everywhere else.
Corey Hoffstein 52:15
So stick with me on this question. It’s a little bit morbid, but I’m really curious how you’re going to answer it. So let’s pretend for a moment you get to the end of your life. And you meet Saint Peter at the pearly gates. And he says, Thank you for your contributions to finance. Because you’ve done so much you can ask any question you want about the markets and I will tell you the eternal truth. What question do you ask?
Cliff Asness 52:37
First, I do like your odd assumption that a quantitative investment manager can go up not down. In this scenario, that’s very kind of you.
Corey Hoffstein 52:48
I have to hope. I mean, I’m in the same boat here.
Cliff Asness 52:50
I’ll give you two once about markets once a little more competitive in nature. This is almost a throwaway. But we’re never going to know how efficient the market is. I think it’s certainly quite efficient in a long run horizon. I think it’s rarely wildly inefficient. But even when I wrote my dissertation, Chicago, a big part of it was on price momentum. You’ve heard the joke, Gene fama was great about it. He said if it’s in the data, write the paper up, but I was terrified and telling him I wanted to write it unsuccessful. Momentum failing would have been a great Chicago dissertation momentum succeeding was difficult, but momentum is one of the harder ones to reconcile with efficient markets. People have certainly tried over the years, there are some theories. If you believe this momentum to things, I think you’re most of the way towards believing markets, or at least not perfectly efficient. I should mention also gene fama won’t tell you they’re perfectly efficient. I sat through his class three times, once as a student and twice as a TA, just to make sure I kind of could answer the students questions. And every year, he probably still does that. I haven’t asked him in a while. He looks at the class at one point early and says markets are almost assuredly not efficient, or not perfectly efficient. And you get guests. Only in farmers class in Chicago. Could that statement elicit guests, he just says a point on a spectrum. Clearly, Jean thinks they’re more efficient than most people. So I would like to ask St. Peter Exactly. To explain this one. To me. I’m pretty convinced they’re not the obvious. It’s almost a tautology. They’re not perfectly efficient. I’m pretty convinced. I wouldn’t use the phrase irrational. I use the phrase less than perfectly rational. You and I both make errors on occasion, we both can do irrational things. I think we’d accept the verdict that we’re not perfectly rational. I think we’d be insulted if someone called us irrational. That’s a little bit different. So I’d like to ask St. Peter that and then in a purely competitive way, this is odd. I’m working this out of my head as I’m asking this. I’d like to ask them, Who are the best managers give me the top 10 But I have a very specific way I want to ask it. I want an ex ante. Everyone’s had good or bad net luck in their life and markets and this business is great. have dependents so luck can matter? A lot. But so St. Peter can answer ex ante. Maybe the person was a total failure because they took way too much risk early on, but who had ex ante the most been a quantitative or, or not? An answer in a risk adjusted dollars sets. Other fair amount of people, particularly in the hedge fund world that get into these blanking contests over Sharpe ratio. Well, that’s great. We want the highest Sharpe ratio we can get. But what’s relevant is how many risk adjusted dollars you can create for clients. risk adjusted means Jack Bogle did a tremendous thing. But us doing the same thing doesn’t create wealth for clients, but a style that’s not in their portfolio yet. So I’ve often thought about writing a piece it would be dangerous to write right now, because things can get turned on you when things are not going so great. But an ode to low Sharpe ratio strategies. And when I say low, I mean, we want to make them as high as we can make them the craftsmanship we talked about earlier. But value particularly in large cap, not a huge edge. It’s just an edge that’s existed everywhere for a long, long, long time. And you can add a lot of dollars to portfolios, because these things tend to be very, very, very high capacity. And not infinite. Nothing is that’s ridiculous, but certainly very high capacity. So I would like to know, in that sense, risk adjusted dollars. And clearly I’m altering the question to favor people like me a little bit by making it that sense not Sharpe ratio for the same amount of say risk of capacity higher Sharpe ratio definitely adds to it. But both count who would have been the best I don’t expect to be on this list. It’s on it, and ego fest. But I’d be fascinated to know the ex ante best managers ever in a risk adjusted dollar sense. It might be surprising.
Corey Hoffstein 56:54
You have done dozens of interviews, podcasts talks, fireside chats, been asked probably hundreds of questions. What is the one question you wish someone would ask you but never has?
Cliff Asness 57:08
I can’t go with that St. Peter one? No, that’d be different. Because that was the question I didn’t know I wish to be asked. But I did wish to be asked for the question I would like to be asked. Alright, I’m going to cheat the question. I would like to be asked in 2023 Oh, my God, you make tons of money by sticking with that? Why was this not obvious to the whole world? I am looking forward to being asked that question. And I do believe I will be asked that question. But what’s your answer gonna be? My answer is it’s way harder, and you sacrifice some of your internal organs to stick with this. And the end of your life comes 14 years earlier than it was supposed to. They don’t hand you the risk premium. A great man once said, No pain, no premium.
Corey Hoffstein 57:51
Alright, last question for you, Cliff. 2020 has been an upside down year for so many people. What are you looking forward to in the future? And this can be business or just personal?
Cliff Asness 58:01
To answer personally, you got to think about this to post COVID world. And the question really comes down to what do you miss the most from me and miss a ton of things. I’m sure we all do. But there are two things I miss tremendously and they’re related, but going to restaurants with friends. It’s not exactly I’m sorry, an earthshaking answer. I’m not unique on this. But we’ve done the Zoom cocktails with friends numerous times, even that’s getting a little old. It’s not the same thing as sitting around a restaurant for two and a half hours where people you love killing more wine than you should. And this is also very prosaic, it’s not different than anyone else, or many other people, but live sporting events. And I think it’s a long time restaurants I think we can do with table distancing and whatnot. Until a vaccine happens the notion that I’m going to sit in Yankee Stadium, they make the seats pretty damn small to maximize revenue, those damn capitalists, Broadway shows even worse, they cram in like, maniacs. So the notion that any of us will be comfortable doing that, in the near future is very sad to me, because I love sports. I love live sports. I’m not claiming to be good at any of these but I love them. So I am really looking forward to though I think that’s sadly a ways off. I’ll tell you what a lot of people say I’m not looking forward to is my kids having their liberty again. I have four teenagers in my house. It’s like a last hurrah to spend a lot of time loving that occasionally as you probably will edit out but occasionally they come and interrupt the podcast to ask me to fix the home AV system. Because again, I am the home it man. It’s a priority I get it. But I do keep reminding myself that in one very narrow sense. You know a lot of people have died. I’m not saying this is a good thing. But in a very narrow personal sense. The one gift from this whole thing is like a last amount of serious family time with teenagers who are going to go to college soon.
Corey Hoffstein 59:58
Well I clip I just want to say this has been a pleasure. I can’t thank you enough for your time. It has been a real honor. As I said, longtime fan, big reader your research getting to ask these questions has been a real joy. So thank you,
Cliff Asness 1:00:11
longtime listener first time caller. Now this was great. These are some of the best questions I’ve ever gotten and it was a pleasure to talk at you