Quant Investor Cliff Asness Hasn’t Smashed His Screen This Year—Yet

quant
cliffasness
aqrcapital

#1


via bloomberg

The billionaire chief of AQR Capital Management is having a lousy 2018 with the $226 billion he’s running. He says factor-based strategies can still prevail.

Billionaire, blogger, charismatic quant, and fount of Twitter rage: That’s Clifford Asness. Over 20 years, the former Goldman Sachs managing director and graduate student of Nobel laureate Eugene Fama has built AQR Capital Management into a systematic investing giant by capitalizing on two trends: the growing power of computers and demand for lower fund fees. Today, AQR runs $226 billion in strategies built on so-called factors—behaviors that securities tend to exhibit over time. The problem is the “over time” qualifier. Asness, who turns 52 in October, is having such a miserable 2018 that he penned a 23-page essay defending AQR’s faith in factors. Here, the quick-tempered native New Yorker tells Erik Schatzker why he’s assembled a team of Ph.D.s on par with the finest finance faculties and how he keeps his sangfroid.

Cover artwork: Alan Coulson for Bloomberg MarketsES: Why have quants done so poorly of late?

CA: The big culprit on the year is systematic value investing. That one has been bad for quite a long time, probably since just after the financial crisis. And, to answer a question you didn’t ask, our faith in that doesn’t change a drop either.

So in a given year, value might have a tough time, but if quality, momentum, and carry all do well, we can do well. This year, the other ones are not making up for value.

ES: As you say, you’re not an unemotional person. You’ve been known to get upset and break things when your returns—I’ll use the word “suck.”

CA: Those were probably my words at the time.

ES: Any computer screen casualties this year?

CA: No. And the very fact that I have noted that to several people is a little embarrassing. I’m keeping track. It’s a thing. But I am not any kind of Vulcan about those things. Even this year. I find it very upsetting when we lose money.

ES: What exactly does AQR do?

CA: Rigorously following a model and doing so with great diversification are the hallmarks of quant. Now, when you get more specific about AQR, we do come from the world of academia. I was a Ph.D. student in the late ’80s. We were present for a lot of what’s become modern quantitative investing. It started out with the size effect and moved on to value. I was very, very early to discover the momentum effect. Over time the list of factors has grown to include a few others. They’re all steeped in the academic literature. We’ve helped create some of that literature, but we’re also huge consumers of it. Then we spend a ton of time trying to apply this with breadth and depth. We spend a fair amount of time trying to create what we would call alpha.

“I’ll still wake up sweating once a year, worrying that we’ve just gotten lucky forever”

ES: Can factors break down, become commoditized, or simply stop working?

CA: Absolutely. It’s something we actively monitor. The main tool that many people use these days to monitor the state of where we are and whether these are being arbitraged away is something called the value spread. Any factor can be thought of as a long/short portfolio. And on various value metrics—any of your favorites, price-to-earnings, sales-to-book—how expensive are the longs vs. the shorts? For the value factor itself, the longs always look cheap by definition. But for the other factors, it varies, and all of them have a range through history. None of that is at levels where the factors have not done well going forward.

ES: David Harding, the quant who founded Winton Capital Management, says this time is different. He’s concluded that at least one factor or risk premium—trend following—doesn’t work as well as it once did. He’s said he’s reducing the weighting of that factor in his multifactor funds, and he’s charging less for his trend-following fund.

CA: I believe he’s doing that in response to competitor pressure from people like us who do trend following and charge less than he does. So he can blame us for this, and I’ll happily take the blame.

I’m not really smart enough to figure out small nuances like, “It will work, but not as well as in the past.” We’ve always said a good estimate for the future is maybe half [the performance] of your backtest.

Trend following has been a very good strategy. It still belongs in most investors’ portfolios. And if I’m not mistaken, it’s still a substantial part of David’s portfolio, just less. We don’t tell anyone that their whole portfolio, God forbid, should be trend following.

ES: Harding blames the relative lack of success in that factor on crowding.

CA: We don’t think there’s a tremendous case that it’s too crowded; we just think trends have reversed more in the last few years. That’s a tautology, trend following not working, but it happens.

ES: You believe there will be a reversion to the historical mean?

CA: I do believe going forward, these strategies will work at a relatively similar level to how they’ve worked in the past. Reversion to the mean doesn’t mean a huge comeback. It just means the factor is about as good a deal as it used to be. One thing I try to avoid, and there’s a lot of pressure to do this, even internally, is to tell people the comeback is coming. Every once in a while there’s a tactical opportunity. But I’m not sure you want that, because it’s usually preceded by body bags.

ES: Is it reasonable to wonder if rising yields and quantitative tightening, and the impact they may have on everything from volatility to liquidity, will change the game for quant strategies?

CA: It is reasonable, it’s just absolutely not borne out by the data. A lot of these tests go back longer than the bull market in bonds. Look at the core, old-fashioned but still wonderful value and momentum factors for picking U.S. stocks. The data in that starts in the ’20s, and the initial tests on it were done in the mid-’80s. There was no bond bull market to drive that. The last 30 years have been a pretty massive, on net, bond bull market. But there have been some pretty horrible bear ones, and we don’t see a tremendous pattern there either.

ES: Quants have become the whipping boys for sudden sell-offs, the so-called flash crashes, and periods of extreme volatility. According to one estimate, there was almost a trillion dollars in quant hedge funds alone at the end of 2017. Has quant become too much of the market?

CA: I don’t know how people calculate these things. It sounds like way too much. Some funds use some leverage, so they could be getting to that number based on assets instead of equity. I’m trying to be polite about this.

ES: Don’t be.

CA: It’s a load of crap. One of the problems I think quants have is we are reasonably transparent. No one frickin’ knows what the average judgmental active manager is doing. If they caused the [May 2010] Flash Crash by all panicking at the same time, we would just say the market went down. Quants are easy to identify and lump together. There has been one time quants, I think, were responsible for a short-term radical market movement: August of ’07, the Quant Quake. It was a whole bunch of quants selling that led to a very sharp decline. As is often the case with price pressure-driven declines, the market came back very quickly. You know who was really upset? A bunch of quants were very upset, but we didn’t upset the world.

“People say active management is on the wane, and therefore everything must be crazy. That’s certainly plausible, but they don’t actually ever point to any data”

ES: Is it possible that many systematic strategies are so similar that they respond to the same signals by liquidating the same ­positions?

CA: It is. If I live to be a hundred years old, I expect to see something like the Quant Quake again. There’s no strategy in the world that isn’t somewhat susceptible to what other people do. You have to size your strategy at the survivable level.

ES: If it’s not the machines, what was behind the volatility in February, in June?

CA: Our system of trading is entirely different than 20 years ago, and that is somewhat scary. It has nothing particularly to do with quant, but if everyone’s trading using new systems, I think you can get some weird short-term results. If anything, I would say the world has been calmer than we expected since the global financial crisis, or compared to long-term history. If anyone wants to say, “Quants have grown a lot in the last 10 years and things are calmer than normal. Yay, quants,” it would be at least as valid as the negative comments you hear very often. But, unfortunately for me, also not true.

ERIK SCHATZKER: The performance of quants in 2018 has turned some faithful into skeptics and raised doubts as to whether some of these strategies are still viable. Is there a crisis in quant land?

CLIFF ASNESS: There might be a minor crisis in confidence for some. I think it’s misplaced. We’ve seen periods like this quite a few times before. It’s been a while since the last one, and people kind of forget. Plus there are new investors in quant who weren’t in it before.

There may be investment strategies that never, ever, ever have bad periods. A fair amount of those turn out to be scams; and even the real ones, the few people who have invented stuff like that, can’t run a ton of dollars, and they kick all the clients out. Real-world investment strategies that are relatively uncorrelated to markets and have positive long-term average returns and are scalable are hard to create. But they don’t generate win-every-day, win-every-month, win-every-year risk-adjusted returns. We absolutely expect to win long term, or we wouldn’t do it. I think the evidence bears out that we do.

I’m trying very hard to sound all calm and sangfroid with you right now, and I am, intellectually. But I’m as emotional as anyone else. People at AQR would laugh at me for trying to sound calm, because they get my emails: “Another frickin’ down day!” When we have a bad period, I want to figure out why, and I want an answer.

We looked at some of the standard possibilities, like maybe quant’s gotten too big. We looked at the actual numbers in quant. We looked at the transaction costs, which we think would be directly affected by size. It’s plausible. I don’t reject that, but we don’t see any evidence.

Are the strategies too expensive? If the long positions were much more expensive than the shorts compared to the past, you could say there’s some evidence that maybe quants have changed the world. We absolutely don’t see that.

ES: Why have quants done so poorly of late?

CA: The big culprit on the year is systematic value investing. That one has been bad for quite a long time, probably since just after the financial crisis. And, to answer a question you didn’t ask, our faith in that doesn’t change a drop either.

So in a given year, value might have a tough time, but if quality, momentum, and carry all do well, we can do well. This year, the other ones are not making up for value.

ES: As you say, you’re not an unemotional person. You’ve been known to get upset and break things when your returns—I’ll use the word “suck.”

CA: Those were probably my words at the time.

ES: Any computer screen casualties this year?

CA: No. And the very fact that I have noted that to several people is a little embarrassing. I’m keeping track. It’s a thing. But I am not any kind of Vulcan about those things. Even this year. I find it very upsetting when we lose money.

ES: What exactly does AQR do?

CA: Rigorously following a model and doing so with great diversification are the hallmarks of quant. Now, when you get more specific about AQR, we do come from the world of academia. I was a Ph.D. student in the late ’80s. We were present for a lot of what’s become modern quantitative investing. It started out with the size effect and moved on to value. I was very, very early to discover the momentum effect. Over time the list of factors has grown to include a few others. They’re all steeped in the academic literature. We’ve helped create some of that literature, but we’re also huge consumers of it. Then we spend a ton of time trying to apply this with breadth and depth. We spend a fair amount of time trying to create what we would call alpha.

“I’ll still wake up sweating once a year, worrying that we’ve just gotten lucky forever”

ES: Can factors break down, become commoditized, or simply stop working?

CA: Absolutely. It’s something we actively monitor. The main tool that many people use these days to monitor the state of where we are and whether these are being arbitraged away is something called the value spread. Any factor can be thought of as a long/short portfolio. And on various value metrics—any of your favorites, price-to-earnings, sales-to-book—how expensive are the longs vs. the shorts? For the value factor itself, the longs always look cheap by definition. But for the other factors, it varies, and all of them have a range through history. None of that is at levels where the factors have not done well going forward.

ES: David Harding, the quant who founded Winton Capital Management, says this time is different. He’s concluded that at least one factor or risk premium—trend following—doesn’t work as well as it once did. He’s said he’s reducing the weighting of that factor in his multifactor funds, and he’s charging less for his trend-following fund.

CA: I believe he’s doing that in response to competitor pressure from people like us who do trend following and charge less than he does. So he can blame us for this, and I’ll happily take the blame.

I’m not really smart enough to figure out small nuances like, “It will work, but not as well as in the past.” We’ve always said a good estimate for the future is maybe half [the performance] of your backtest.

Trend following has been a very good strategy. It still belongs in most investors’ portfolios. And if I’m not mistaken, it’s still a substantial part of David’s portfolio, just less. We don’t tell anyone that their whole portfolio, God forbid, should be trend following.

ES: Harding blames the relative lack of success in that factor on crowding.

CA: We don’t think there’s a tremendous case that it’s too crowded; we just think trends have reversed more in the last few years. That’s a tautology, trend following not working, but it happens.

ES: You believe there will be a reversion to the historical mean?

CA: I do believe going forward, these strategies will work at a relatively similar level to how they’ve worked in the past. Reversion to the mean doesn’t mean a huge comeback. It just means the factor is about as good a deal as it used to be. One thing I try to avoid, and there’s a lot of pressure to do this, even internally, is to tell people the comeback is coming. Every once in a while there’s a tactical opportunity. But I’m not sure you want that, because it’s usually preceded by body bags.

ES: Is it reasonable to wonder if rising yields and quantitative tightening, and the impact they may have on everything from volatility to liquidity, will change the game for quant strategies?

CA: It is reasonable, it’s just absolutely not borne out by the data. A lot of these tests go back longer than the bull market in bonds. Look at the core, old-fashioned but still wonderful value and momentum factors for picking U.S. stocks. The data in that starts in the ’20s, and the initial tests on it were done in the mid-’80s. There was no bond bull market to drive that. The last 30 years have been a pretty massive, on net, bond bull market. But there have been some pretty horrible bear ones, and we don’t see a tremendous pattern there either.

ES: Quants have become the whipping boys for sudden sell-offs, the so-called flash crashes, and periods of extreme volatility. According to one estimate, there was almost a trillion dollars in quant hedge funds alone at the end of 2017. Has quant become too much of the market?

CA: I don’t know how people calculate these things. It sounds like way too much. Some funds use some leverage, so they could be getting to that number based on assets instead of equity. I’m trying to be polite about this.

ES: Don’t be.

CA: It’s a load of crap. One of the problems I think quants have is we are reasonably transparent. No one frickin’ knows what the average judgmental active manager is doing. If they caused the [May 2010] Flash Crash by all panicking at the same time, we would just say the market went down. Quants are easy to identify and lump together. There has been one time quants, I think, were responsible for a short-term radical market movement: August of ’07, the Quant Quake. It was a whole bunch of quants selling that led to a very sharp decline. As is often the case with price pressure-driven declines, the market came back very quickly. You know who was really upset? A bunch of quants were very upset, but we didn’t upset the world.

“People say active management is on the wane, and therefore everything must be crazy. That’s certainly plausible, but they don’t actually ever point to any data”

ES: Is it possible that many systematic strategies are so similar that they respond to the same signals by liquidating the same ­positions?

CA: It is. If I live to be a hundred years old, I expect to see something like the Quant Quake again. There’s no strategy in the world that isn’t somewhat susceptible to what other people do. You have to size your strategy at the survivable level.

ES: If it’s not the machines, what was behind the volatility in February, in June?

CA: Our system of trading is entirely different than 20 years ago, and that is somewhat scary. It has nothing particularly to do with quant, but if everyone’s trading using new systems, I think you can get some weird short-term results. If anything, I would say the world has been calmer than we expected since the global financial crisis, or compared to long-term history. If anyone wants to say, “Quants have grown a lot in the last 10 years and things are calmer than normal. Yay, quants,” it would be at least as valid as the negative comments you hear very often. But, unfortunately for me, also not true.


#2

Photographer: Ike Edeani for Bloomberg Markets

ES: Once, all prices in the stock market were set by people making decisions about whether to buy or sell companies. Now as little as 10 percent of all trading in stocks is fundamental discretionary. The rest is machines. What effects has that produced, and on the whole are they positive or negative?

CA: I think discretionary [investors] still have more assets under management. Last I saw, depending on your definition, quant or indexing—and you know that I’ll rant and rave that we’re not a passive index fund—are still less than half of the world. We probably do trade more.

Everything is within historical norms. People say active management is on the wane, and therefore everything must be crazy. That’s certainly plausible, but they don’t actually ever point to any data. Another second-order effect we worry about, of course, is trading costs. If it costs triple what it used to to trade because a whole bunch of people are trying to do the same thing at the same time, and they’re moving the price before you get to it, you could have a strategy that is more costly to implement and therefore its edge gets smaller. Our trading costs look very normal to us. We don’t see a big deterioration, and it’s not for want of looking.

ES: How much active trading is needed for accurate price ­discovery?

CA: It’s probably less than many think.

ES: The new market structure, with its high-frequency trading [HFT] companies, has reduced trading costs. But is it possible those companies won’t supply liquidity when it’s most needed?

CA: I don’t actually believe this is true. Of course it could be true. There’s an assumption that prior to HFT, old-school market makers used to buy something they knew would lose money because they had to make markets. No old-school market maker ever said, “Well, it’s currently trading 20 percent below, but I’ll pay you only 10 percent below because I’ve got to make this market more stable.” It didn’t happen. Liquidity has always run away in a crisis.

ES: How much of your assets under management today would you call hedge fund money?

CA: It’s probably still in the ballpark of half of what we do by assets. What we do in the hedge fund world tends to be fully hedged. A lot of hedge funds are still very net long. I hate even to call our hedge products hedge funds because it conveys that we’re charging 2 and 20 [percent] for $100 billion worth of assets, which is not even close.

When we left [Goldman Sachs] to start AQR, we raised a billion dollars out of the gate. If you told people you were closing the fund, they wanted to double their investment. There have been various hedge fund frenzies. I think we had the good fortune to start AQR in the midst of one. The month we launched our first fund, Long-Term Capital [Management] blew up. That hedge fund frenzy stopped really fast for a while.

ES: Of all the investment managers out there, which do you consider your toughest competitor?

CA: It depends on the area. If Renaissance [Technologies] were to open the Medallion Fund, they would beat us in every competition ever because they beat everyone else also. Every once in a while an investor says to me, “How do you compare to Renaissance?” I’m like, “Oh, we’re way worse than Medallion, but they won’t take your money, and we will, and we’re pretty damn good.”

ES: More broadly?

CA: Well, let me go to the traditional side because a few spring to mind. There’s a firm called Arrowstreet [Capital] up in Boston. I think they’re a very impressive quantitative firm. I’ve had quite a few public spats with Rob Arnott of Research Affiliates. But he’s a brilliant guy, and they run some really solid quantitative products. The things that BlackRock is building—Andrew Ang leaving Columbia to go there—they’re tough competitors.

ES: Dimensional Fund Advisors?

CA: In many ways they’re one of the best quant firms, and they’ve managed to get the world to think of them as kind of a different thing. Quite a few people refer to Dimensional as passive. They use value. They even use momentum, which is the least favored factor. They’re some of the people I admire most in the industry.

ES: What makes AQR different?

CA: I’m fond of saying our secret sauce is not even close to secret. We’re quite open about it. We’ve made a very conscious decision to share our research over time. I think it’s been a wonderful business decision. We were really a bunch of academics who wanted to see if this stuff worked in the real world and make money for real clients. It’s in our bones and in our DNA. We get to hire some of the best academics because they want to come to a place with peers. They know they can continue to do what they love. But they also know that if they come up with something proprietary, we won’t let them publish. If you picked 20 of AQR’s top researchers, I think it could be a top finance faculty at any school in the world.

ES: What’s the most interesting advance taking place in quantitative investing today?

CA: We’ve just made a big hire in machine learning, Marcos López de Prado. We’ve been doing versions of machine learning here in isolated groups for a while, but now it’s such a big effort. We think it has applications across a ton of what we do. One of the reasons we were comfortable hiring Marcos is that he’s very comfortable with trying to fuse machine learning with a need for some economics behind it. He’s not a “just let the machines run wild” guy. We continue to work on how to measure the old factors better, but, of course, that will be very incremental. This has the potential to move the dial more.

Photographer: Ike Edeani for Bloomberg Markets

ES: Are there functions people now perform at AQR that machines will perform in the future?

CA: Almost by definition, yes, because I believe that across the economy. I’m a little less pessimistic on it being such a terrible thing for the world than some. In trading, we’ve absolutely seen the ratio of machines to people go way up, particularly in the last 10 years.

ES: Do you see the pursuit of big data and the application of machine learning as an arms race?

CA: Yeah, and I think we just hired a howitzer.

Processing the big data first—which is what Marcos is about—if you’re marginally better at it, can be a marginal advantage forever. Part of big data is acquiring the data, negotiating the fees for the data. Like I’ve said about investment products, there is no product so good that there’s not a fee that can make it bad. Well, there’s no data source that’s so good that there’s not a fee that can make it bad. So in that sense it’s not just a theoretical geek effort, it’s a business effort.

ES: How much of AQR’s success, particularly in the past decade, would you attribute to superior returns, and how much would you attribute to superior marketing and thought leadership?

CA: Fifty-fifty, with me preferring the term “thought leadership” to “marketing,” shockingly. Thought leadership gets you credibility with people. They assume, which I selfishly believe is a good assumption, that it translates into a very competent running of the portfolio. We partially do thought leadership just because we can’t help ourselves. We consume it. But in part it’s because it’s very hard to demonstrate superior returns. The difference between a 1.0 Sharpe ratio and a 0.9 Sharpe ratio takes a ridiculously long time to actually prove.

ES: What has changed most in the company’s two decades?

CA: We were a billion dollars when we started. We’re somewhere above $200 billion now. That does change your world a little bit. We were 13 people when we started. We’re pushing a thousand now. I think a huge part of how we’ve dealt with scale is partnership. We stole a lot of our business model from what I would call pre-IPO Goldman Sachs. That’s the only Goldman Sachs I worked at.

Embracing that culture, sharing the wealth by making equity owners out of 36 of our colleagues, has allowed us to manage this size. So the biggest change to AQR would be that.

There was another huge change. Maybe it’s a tie. At some point, we looked at ourselves and the industry and said, “Some of what we’re doing is unique alpha. And some of the stuff we’re doing is not alpha, and we’re not going to charge alpha fees for it anymore.”

ES: What count as your biggest mistakes?

CA: Getting a little too levered and having too much convertible arbitrage before the financial crisis. We survived and actually did fine. I think we could’ve made more money if we’d taken less risk beforehand because we would’ve been in a better position to act on it. I think we were very, very early on low-risk investing. We used it early, but we didn’t launch dedicated low-risk products. I do kind of kick myself for that.

ES: Why not turn your strategies into exchange-traded funds?

CA: We don’t see them as a huge advantage over mutual funds. Some people claim they have very large tax advantages, but our tax people keep saying it’s less obvious. Therefore we don’t see a blindingly compelling reason to do it. If a whole bunch of our clients said, “We’re just not interested in you if you don’t do ETFs,” we’d probably do it.

ES: Tell me about the original decision to sell part of the company to Affiliated Managers Group Inc.

CA: Huge error. We had a terrible start at AQR, worse than today. That was the tech bubble of ’99, 2000. I don’t think we were one step from death, but if I squinted and looked far away, I could see the pearly gates. It became clear that AMG wanted to buy a modest piece of us at what we thought was a fair valuation.

I wrote a letter to our clients in 2004 announcing and explaining the AMG transaction. I said, “Whenever you hear about someone selling a stake, you’re told how great it is for the end clients, how it’s going to increase your capacity to do this or that.” I go, “Yeah, it’s none of that. It’s personal diversification.” Most of us had 99 percent of our net worth in AQR. Doesn’t 70 percent sound like a pretty huge bet? A very confident bet? Personal diversification was the entire motivation. And the second time was really a smaller version of exactly the same logic. And as part of it, I signed up, as did some of the other partners, to stay for seven years. If I was ever thinking of retirement, which I’m not, I took that option away.

ES: Will there be a liquidity event? Will you take AQR public?

CA: There are other ways to transition a firm. Maybe other, younger partners buy it. I certainly won’t take an IPO off the table, but I don’t see it even on the medium-term horizon. I think back to, say, the financial crisis and what that would’ve been like to go through as a public company. As a private company our valuation might have gone down 80 percent, but there’s also no nice man from Bloomberg talking about it on TV. I think it’s harder to manage a public company through extremes. At some point of maturity, when all the original people are in their dotage, maybe you don’t have a choice, maybe that’s the natural evolution. But I have no desire personally to run a public company.

ES: So if AQR has $226 billion today, what’s a reasonable expectation for where you’ll be five years from now?

CA: Well, nowhere near the growth we’ve had in the past. If five years from now we were in the upper 200 [billions], I’d be fine with that. I never want to shrink, but if we were in the mid-100 [billions] and we’re doing a good job and we’re just out of favor, I’d be fine with that, too. We’ve never had a goal in terms of assets.

ES: Where in finance would you like to see more innovation or more research even if you aren’t willing to spend the time, effort, and money?

CA: Private investing. I think there’s an opening for someone to be the Vanguard [Group] or AQR of private investing, putting in enough work to make sure we’re getting a fair price and then charging a third of what everyone else charges. I don’t know that world enough, but I think it’s an interesting place. One great way—in any part of the economy—to see where there’s room for innovation: Who’s making monopoly rents?

ES: Where is AQR most eager to expand?

CA: You don’t always have to grow, you don’t have to have a new asset class that you haven’t explored yet. But aside from illiquidity, which is possible but a much bigger leap for us, I think fixed income is the final frontier.

A bigger challenge we have going forward is investing capacity, rather than raising assets. I think we’ve been very disciplined in closing some products. There are places that still have capacity. Fixed income is a great example. Large-cap equities is a very big capacity place. We haven’t had a lot of trouble raising money.

Photographer: Ike Edeani for Bloomberg Markets

ES: You first developed the ideas that ultimately formed the foundations for this company while a student at the University of Chicago. Tell me about your relationship with Eugene Fama.

CA: I love Eugene Fama. I hope he loves me. I was his Ph.D. student. He was co-chair on my dissertation, along with Ken French. I disappointed him when I went to Goldman Sachs instead of becoming an academic. I said, “Gene, they’re offering me a job where I can do essentially the same research and get paid more. Doesn’t economic modeling say I should do that?”

ES: Is it true that you left Goldman in a dispute over pay?

CA: No. Not even close. It was probably about mid-’97 when we started seriously thinking we could do this on our own. At the time, we weren’t using Goldman data or research. I was a loyal Goldman guy—up to a point, obviously. I went to the partners I reported to and said, “We’re thinking about leaving.”

I wrote a whole memo on staying at Goldman. It had two aspects. One was money. As well as we did at Goldman, if we started our own asset management firm and it worked out well, we’d do better. We didn’t want Goldman to take us all the way there. That was unrealistic. But we said, “What if you can move a third of the way there, give us more of a performance fee on what we do?”

Second, and almost no one knows this, but we weren’t just managing money at Goldman. We also were the group in charge of everything quant. If some marketing person wanted an efficient frontier drawn, we drew it. This was, like, half our time. We were running billions of dollars and still having to produce efficient frontiers for other people’s marketing. So I asked to start a separate group to do that. The odd thing is, they met me pretty much the whole way on money, and I was airtight promised a partnership in the next cycle. But they greatly valued the other part of what we did, all those other things, and there was no give there.

ES: Tell me about your temper.

CA: I have a bad short-term fuse. I readily admit that. By the way, I’ve never hit a human being. This is me getting mad at my computer and hurting my own hand a couple of times. Even online. I left Twitter at one point because I was finding it too frustrating. I rejoined after a personal six-month penalty box.

ES: You tweet thousands of times a year.

CA: It’s 17 minutes a day. Go add it up.

ES: Why?

CA: I have three kinds of tweets. The bulk are quantitative finance, and there’s a great community on Twitter now of people really discussing these things. It’s just wonderful. I tweet on my view of free markets—I’m a believer in free markets. A little bit of politics. There are some things I wish the president would do differently, and sometimes I can’t resist mentioning that on Twitter, which is not wise for a CEO of a company to do. But I don’t always do the wise thing.

And then, anything I think is particularly funny. Humor is important to me.

“At some point we will see a non-normal event again. Our job as managers is to survive it”

ES: Tell me about your politics.

CA: I’m a classic English liberal, which is rule of law, limited government, don’t tell other people what to do. I’m not a Burning Man libertarian.

ES: What are the most important changes you’ve seen in the asset management industry?

CA: Easily the most important is the rise of passive. And the fees on passive—the idea that one can purchase the market for effectively zero basis points.

There is part of investing that must be done with great skill. There is part of investing that was automatable. Part of what Warren Buffett does is to always own cheaper, more profitable, lower-risk stocks. I think it’s been a pretty good deal for Warren Buffett’s investors. But with most active managers, you paid through the nose for it. Nowadays, you pay tiny amounts for it. There are ETFs out there charging nine basis points on factors for U.S. large-cap stocks.

People were right to get massively excited about going from mutual fund fees of 2 percent, with a load upfront, to 30 basis points. But when the index funds start arguing about the difference between two basis points and zero, they kind of forget that they’re talking about two basis points.

ES: Fidelity Investments recently made headlines by cutting fees on a few funds to zero. Is the trend in fees inexorably toward zero?

CA: No. The trend in passive index fund fees is inexorably toward zero. There are people who will get mad at me for saying that, but you can’t really charge high fees when you’re not adding value. Any form of active management—be it judgmental, be it quant—is an assertion that we’re providing you something that the passive index doesn’t. If it’s true skill, the fee on that should still be fairly high.

ES: What might be the most important changes to the industry in the future?

CA: Well, not surprisingly, the acceptance of the famous factors and tilts. The fact that they’re now available for much lower fees is a very big change. I’m not taking any personal credit. Many people would have said that move’s under way. It’s not nearly as big as the move to passive, but it’s probably the huge one that’s out there.

ES: You must have a vast trove of failed experiments. Have all the risk premia been identified?

CA: People have up to six-, seven-factor models—somewhere in that range. I would be willing to bet 20 years from now we’re talking about at least one more. But not three more. And maybe one of them we decided is crap.

ES: You’re as confident as ever in your backtests, confident that the factors you’ve identified are replicable going forward, confident that they’re not the result of data mining or survivorship bias?

CA: I’m always in a panic that they’re the result of data mining or survivorship bias. We spend our lives trying to disprove that. I’ll still wake up sweating once a year, worrying that we’ve just gotten lucky forever.

ES: How do you know that the behavioral reasons that explain why risk premia exist in the first place are still present and persistent?

CA: This is obnoxiously phrased, but there’s a supply of investor error. A factor can go away from either the supply or the demand side. It can go away from too many people trying to exploit it, right? There’s a finite amount of error out there. Or it can go away from that error clearing up. I don’t think either has happened, from the numbers. I worry more about the demand side—too many people like us—going forward than I worry about the supply side.

ES: You reject the notion that the growth in quantitatively managed assets and the consequent emergence of so-called ­programmatic liquidity have added risk to financial markets.

CA: I reject the notion that it’s added a substantial amount of risk. I think it’s rather trivial. And the reason is these techniques are almost exclusively used to try to beat markets or in a market-neutral sense with far less leverage than pre-financial crisis. They are largely about reallocating returns among investors who pursue these techniques. It’s been a terrible year for quant, and I have not read apologies from all the critics who said we would bring down the world.

ES: Isn’t the whole point of programming machines to make them do what people are emotionally or intellectually incapable of doing?

CA: Yes.

ES: Doesn’t that mean that markets will behave differently as machines do more of the trading?

CA: Relative returns, yes, will change. To be frank, I’m not sure how. I’m not sure, for instance, that the market is less stable. I think a lot of quants are replacing active managers who are taking similar bets in a more concentrated way and charging much more. They were tilted, maybe, toward value. If you argue that models don’t get scared, they might be steadier hands. It will have an effect. I do strongly believe the assumption that it’s a negative effect is not borne out by the crises we’ve seen.

ES: Can a few bad quants ruin everything?

CA: A few bad quants can create short-term volatility. And that’s what we think happened in August of ’07. I think the quant world is a safer world than it was in 2006 or 2007, but not perfectly safe. Our chief thing is to tell clients at some point we will see a non-normal event again. Our job as managers is to survive it, and your job as investors is to allocate the right amount so you can stick with it. I would prefer, to be honest, it never happens again. And I don’t mean this year. This year is not anything weird.