In this episode I talk with Aneet Chachra, fund manager at Janus Henderson.
In his role, Aneet runs flow-driven strategies. These are strategies that seek to find an edge in market events where trading volume creates a predictable pressure on price, such as index additions or deletions, corporate buybacks or issuance, or even the rebalancing of target date funds. Our conversation is wide ranging, from the basics of how Aneet categorizes these types of trades, to views on how changing market structure has affected the opportunity set, to the impact of social leverage on risk management.
While the approach may be highly niche, Aneet is bursting with broadly applicable wisdom.
I hope you enjoy this episode with Aneet Chachra.
Transcript
In this episode, I speak with Aneet Chachra, Fund Manager at Janus Henderson. In his role, Aneet runs flow-driven strategies. These are strategies that seek to find an edge in market events where trading volume creates a predictable pressure on price such as index additions or deletions, corporate buybacks or issuance, or even the rebalancing of target date funds. Our conversation is wide-ranging from the basics of how Aneet categorizes these types of trades to views on how changing market structure has affected the opportunity set, to the impact of social leverage on risk management. While the approach may be highly niche, Aneet is bursting with broadly applicable wisdom. I hope you enjoy this episode with Aneet Chachra.
Aneet Chachra, welcome to the show. Welcome to Flirting with Models. This has been a long time coming. I’m very excited to have you here because you don’t do a lot of media. You’re hard to get, you’re hard to book and you’ve got a lot of interesting things to say so thank you for joining me this season. It’s really a pleasure.
(2:45)
Well, thank you Corey, for having me on. I really appreciate it. I’m a big fan of the show. I think I told you this, that I once had to decide whether I should fly or drive to San Francisco and I chose to drive so that I could listen to all the episodes of Flirting with Models on the way.
Well, I hope you don’t regret that decision. I hope it lived up to whatever reputation it had beforehand. But it only gets better from here now that you’re on the show. So let’s dive right in. For folks, perhaps who don’t know who you are, maybe we can start with your background.
Sure. So Corey, I had a pretty unconventional path to becoming a fund manager. I grew up in Canada, I went to the University of Waterloo to study engineering. From there, Morgan Stanley hired me to be a programmer. So I think starting your career at a big bank and there’s a lot of good things about it, you go through great training and you very quickly get exposed to lots of different ideas and people. But the thing you realize that when you’re in technology at a big bank is that it’s not really their core focus. So if you build a system and I was working on derivative trading, as long as it’s working fine, literally nobody cares. But the minute that something is not quite right, everybody’s calling and yelling at you. So I realized that hey, I need to do something a little bit different and I became a quant at a hedge fund. And I was really fortunate with that, I ended up at a place with extremely smart people and I was, already had a pretty technical and programming and quant background, but I was able to really learn how hedge funds look at different trade ideas and how they put trades on and how they manage them. So from there, I became an equity analyst and that was useful because I got to really understand how to look at a company. But more importantly, I could find out how do long only managers in particular, how do they look at a particular stock? How do they look at their portfolio? And how do they manage their book with tracking error and other constraints?
So I had a pretty geeky background but along the way, I’d kind of picked up a bunch of market knowledge in different asset classes. So yes, I was looking to do something that combined the two and I was really lucky that a colleague of mine who I had worked with previously at the hedge fund, he had joined a British firm called Henderson. And there was a team there that was very focused on flow and price pressure effects. So I went there and I’ve been there ever since. Along the way, Janus merged with Henderson and the thing is, is now if you say to people that, “Hey, flows can be more important than fundamentals especially over the short run,” like almost everybody will agree with that. But when I started doing this, that was a pretty unusual idea. When I would tell people that, “Hey you know, we’re capturing flow effects, people would be like, “That’s crazy, markets are efficient. They reflect fundamentals, everything gets priced in, maybe there’s some very short term arbitrage to do, but really flows aren’t that impactful.”
And now there’s been a huge shift especially in the last couple years. Like I think your work on liquidity cascades has been a big part of that. That now it’s like almost when I talk to people like there’s this nihilism of like, “Hey, all prices are fake and any asset can trade at any price and nothing means anything anymore.” And really, I think that reality is somewhere in-between that flows can absolutely matter in the short run but fundamentals are important, especially when you look out to a longer horizon. So my background has been integrating technology and markets into flow and price pressure.
(6:37)
I want to start back at the beginning of your career really quickly, because you once told me a sort of a funny anecdote. One of your first bosses said to you, “When you fit data to a model, you don’t want it to fit too closely to reality.” What was the purported wisdom behind that statement?
That’s really funny because it happened on my very first day of work at the hedge fund. But really, it took me like many years to figure out what it actually means. And the thing is that when you start off as a quant, your temptation is to take like every piece of data you can find and shove it into your model. Because you don’t have like any judgment or any understanding of really how things work so you’re just trying to replace it. Like what you’re trying to do often is build an exact replica of the real world. So like you might want to say, “Hey, who are the holders of a particular position?” So you can look at something like Apple. Apple, you can go through all the 13D filings and ETF daily reports and mutual fund data and Swiss National Bank owns a bunch of Apple shares and there’s employee ownership. So you can build an imperfect but directionally pretty accurate map of who all the shareholders are.
Now even if you have like perfect data, that’s actually a lot less useful than you would think because it doesn’t really tell you why somebody owns the shares and more importantly, what they might do if things change. And really like that’s the most important thing because it’s the marginal flow that sets the price. Think about you or I. We have $100 in our checking account and we decide, “Hey, we’re going to buy the S&P 500 Index Fund. So give our money to Vanguard or BlackRock or whoever and they take that $100 and out of that, $7 of it goes towards Apple shares. So they need to find somebody else who currently holds Apple and is willing to let go of it in exchange for cash. And then when you redeem out and that process happens in reverse where somebody ends up with the Apple shares and you take the cash.
So it’s much more important rather than having, mapping reality exactly to have even if you have a much more simpler model, but that incorporates how things might change in the future even if you’re not going to be exact at all, but directionally if different events happened, what are the flow impacts they’re going to cause? And the way I think about it is like imagine you take an airplane and you fly over Los Angeles or any other big city. So you can see all the cars on the ground and some of those cars are going to be in parking lots or driveways or on the road or on freeways. But really like that’s just a snapshot of where all these cars are. It doesn’t tell you like where people are going. And if you take the same picture a few hours later, it’s going to look completely different. So really, it’s much, much more about building a model of how things will evolve and change rather than knowing exactly where everything is right now. And particularly in markets like you don’t really want to know like where’s their traffic right now? It’s much more about like, where’s the traffic going to be?
(9:49)
So you’ve mentioned that in the strategies you run, you focus a lot on flow and price pressure effects. These might be ideas that aren’t necessarily familiar to all of the listeners. Can you describe what flow and price pressure effects are and maybe if you can, provide a couple of examples?
Sure. So I’m going to start by giving you my grand unified theory of markets. There’s really only four ways you can make money. So the first one is the most straightforward. You can buy and hold assets that have positive risk premia associated with them. So that’s what most of us do. You buy stocks, bonds, some other assets. You hold them for a long time. So typically with that, you need to have a long holding period so that you’re reasonably certain that you’re going to end up with a positive return. So these are things that you’ll do for one year or longer. And along the way, maybe you can add some extra return through rebalancing or using some leverage. Or if you’re really good, then some amount of rotation or tilting between different asset classes as something is maybe a little rich or cheap. But this is the source of returns that are available to almost everybody.
Now on the exact other end of the spectrum, if you have great technology and huge scale and amazing infrastructure and really smart people, you can be in high frequency trading and market making. And in that you’re just doing you know, huge, huge numbers of transactions and you’re capturing tiny, tiny spreads for providing liquidity. And all these trades are very, very quick. Almost everything is less than one day.
The third way you can make money is the sort of classic fundamental analysis. And that’s like, “Hey, I found this stock that I think is cheap and the market is wrong so I’m going to buy it and over time, the market will come to agree with my assessment and the price will rise up to my fair value.” And with those kinds of trades even if you’re directionally right, generally unless you’re very lucky, it’s going to take some time for that to happen. So maybe the company has an earnings report or some other catalyst and then the value that you saw that nobody else did, gets recognized.
(12:10)
So those trades typically are sort of between one month and one year to play out. So that brings me to the fourth category which is flow effects. So flow effects tend to be between one day and one month. So it’s things were, sub one day is high frequency trading, things that are longer than one month after that point, the fundamentals begin to generally matter more or risk premia associated with the asset. But in that zone, you see that there’s much, much more effect of flow. And so I define flow effects as trades done by market participants who are inflexible in some way. So that’s a market participant who needs to transact at a particular time or a particular security or particular quantity. Quite often, they actually have more than one of these constraints and usually they’re transacting with somebody that is more flexible. So what I’m looking for is, can I find flow effects where I know why the flow effect exists and I can reasonably at least directionally, figure out what the price impact will be.
So I’m looking to provide like a market service by guaranteeing somebody that their needs in terms of a fill or execution but on the other side of it I’m collecting a statistical edge on that. And I think you asked for examples of this. So here’s a couple. Almost all of us as I mentioned at the beginning, we’re in some sort of asset allocation product. So you might be in a target date fund or a balanced fund or really any asset model. So typically, all of these have weights and bands for different assets that it holds and then when you are beyond those bands, it triggers rebalancing. So either there might be rules that if your weight of your stocks is too high, you’re going to sell stocks, buy bonds instead or quite often, they’re calendar driven so these rebalancing might happen every month or every quarter. So that’s an example of a flow that not reflective of somebody’s opinion of stocks versus bonds but really, they’re doing this rebalancing because their model is telling them to do it.
Another thing is taxes. So we all have to pay taxes. The IRS only takes fiat money, they’re not taking stock, they’re not taking crypto and because of that, tax payment dates create interesting effects. There’s lots of stock strategies around year-end because people do tax loss harvesting and then lots of people work for an employer where they get shares or options as part of their compensation. And so as those vest, employees are very likely to at least sell some of the stock that they get to meet tax obligations and that’s another example of where there’s a flow that’s not reflective necessarily of anything about the company itself – people have to make those payments.
(15:17)
So after initially talking to you, I sat back and thought about the different ways in which these flow trades could emerge within the market and it seemed to me like there was a wide variety of potential trades. And I’m curious whether in your experience, there’s a way to sort of categorize or apply a taxonomy to the different types of flow effects that you would typically see?
So I classify flow effects into like three big categories. One is mandates, the other is incentives and the final one is behavioral. So the first one mandates like that’s the strongest so it’s when a market participant has to do something because of a legal or a similar reason. So a good example of that is like index fund providers as I was mentioning before. If a company like when Tesla got added to the S&P 500 Index, every manager had to go out and buy Tesla. It didn’t matter what your opinion of the company was or your opinion of Elon Musk was, if you’re tracking the index, you have to be following the rules of that.
The second category is incentives. So incentives are not quite as strong but I think they’re still very, very important. So you can think of something like option hedging. Like if a bank or a market maker sells an option to a retail investor, generally the bank is not legally obligated to delta hedge their position, but they will almost certainly do so for their own risk and portfolio management reasons. Because if you’re an option market maker, if you can get rid of the risks in the market that you easily can, then that will let you run a much bigger option book.
The third category is behavioral. An example that would be share supply. So like imagine you know, you’re a founder or you’re a venture capitalist and you invest it into a tiny startup, maybe it’s Newfound Research and eventually it ended up being a huge firm, goes public, there’s an IPO. So now you’re sitting on this very concentrated stock position where you have these big unrealized paper gains. So your incentive or your behavioral reasons, you’re going to want to diversify your portfolio. Now you might have quite a bit of leeway in terms of the exact timing or the exact scale of your sales, but the direction of this overhang is pretty predictable. Like if you’re sitting on a billion dollars’ worth of stock as a founder, you’re much, much more likely to be selling shares than going out and buying more of them.
(17:57)
So how do these flow effects ultimately change over time? Do you constantly have to be looking for new flow effects in the market or just sort of the classifications and strategies you can apply remain constant?
So these flow effects do change over time and I see kind of two big reasons for that. So the first one is that what things people are interested in evolves over time. So an example of that is option trading. We all know that there was a big pickup in individual investors getting involved in options post-Covid. First, people were out buying lots and lots of call options. More recently, people have been buying put options. And retail tends to trade options differently than institutions. They’re much, much more focused on shorter dated, they tend to pile into a small number of names at a time. And then also generally, they’re taking a directional view. So institutions might buy options to protect their portfolio on an index level versus you know, retail is out making a very clear bet.
So because of that, the impact on markets of delta hedging of options changes over time depending on what particular securities retail is trading the most, or just the overall amount of trading that’s happening. Like there’s always been option trading or at least for a very, very long time, but it does go through periods where activity picks up or it drops. So the impact of option trading on markets is going to move through time depending on how much delta hedging there is to do in the market.
Now the second reason is much less obvious and that is that even the same amount of flow can have a different impact depending on who the holders are. An example of that is share buybacks. So companies generally try to buy back shares at a relatively consistent pace, they might decide that, “Hey, they’re going to buy a certain number of shares or spend this certain amount of money every month.” So when they first start their share buyback program, there might be a lot of flexible holders who are willing to let go of their shares. And they can start affecting a buyback without creating much impact on price.
But as time goes on especially if they’re significantly shrinking the float, then they’ll find that a lot of their main holders left are much more inflexible. Like there can be index funds to hold it and index funds are really like the ultimate diamond hands because they’re not going to sell no matter how high the price gets. Also think about like what happens when a company stock price is going up, it tends to attract a lot of momentum funds and strategies. And a momentum style by definition, will not sell until the direction of momentum turns. So even the same amount of flow can start having a much bigger and bigger impact as you try to incentivize the remaining flexible holders to sell. So this is the kind of non-linearity that you can see in flows.
You had asked me about the classification, whether that changes. And big picture, it doesn’t. The big three categories that I gave, mandates and incentives and behavioral, like these I think are actually fundamental characteristics of financial markets and even human psychology. But definitely new flow effects within these categories keep coming.
An example that actually I think you brought up recently was buffer funds. So buffer funds are an example of something that generally they are trying to protect by selling a call option and then using that to fund a put spread. And buffer funds are interesting because they’ve just grown so much in the last couple of years. So that’s an example of a flow source where probably if you ran a back test you know, buffer funds probably didn’t have a huge impact on the market. But it’s pretty reasonable to think that they’re going to be more important going forward, especially since buffer funds tend to have fairly fixed rules on like how they execute their hedging programs, what strikes they buy, what maturities they buy. So this is an example of where you can apply some amount of quant knowledge you have from other derivative, systematic derivative trading programs, but also layer in some human judgment that, “Hey, even though buffer funds were not meaningful before, they are probably going to start creating much bigger flow impacts in the future.”
(22:31)
At the risk of being a bit redundant to your answer there, I want to dive in a little deeper because you brought up a bunch of ideas around market structure and how the market itself has sort of changed over time. You talked a little about the growth of option strategies, the growth of indexing. I personally think about for example, the growth of ETFs, index ETFs and basket trading and how a lot of active money has moved to sort of this semi-passive structure. And all of that has occurred slowly over the last 15 to 20 years. I’m curious from your perspective, having done this for a while how that’s impacted flow effect strategies.
Well you know, 15 or 20 years is a long time. A lot has happened in markets, but I definitely think that there’s been two big structural megatrends. So the first one is as you just mentioned, just the sheer rise in the amount of inflexible strategies whether that’s ETFs or passive mutual funds or vol targeting, risk parity, momentum, factors, target date funds, balanced funds, like I can keep going on. There’s just more and more money is in styles where there’s rules that dictate when they buy, when they sell and what assets they hold.
And then also as you mentioned, generally the amount of option trading has gone up. Also, there’s much more structured note issuance, there’s convertible arbitrage. So all those things also create a lot of systematic flow related to delta hedging. When you add those things up, the amount of inflexible flow out there keeps going up not just in terms of the absolute number of dollars, but also the share of the overall market. And on the flip side, there’s just a lot less active funds or other discretionary type capital out there to be more flexible.
But the other big mega trend that’s happened is you know, who is intermediating these kinds of flow effects? Like when I started my career, banks dominated this space. Like banks had huge balance sheets and they were very, very active making deep and liquid markets in all sorts of products. And banks were also pretty interesting that they not only had you know, very big flow trading businesses, but they often tended to have very big proprietary trading businesses. So for a whole bunch of reasons after the great financial crisis, banks are still important but their share of intermediating flows has gone down significantly. And so what banks used to do, a lot of it is being kind of captured by two other categories. So on the very short time horizons, there’s a handful of high frequency trading and market making firms and they’re really dominating on the intraday level.
But in terms of these kinds of flow effect type trades, hedge funds and particularly multi-strategy hedge funds have gotten much, much bigger in these areas. And that’s a big shift because the thing with high frequency trading firms and hedge funds is that they have a lot of flexibility in terms of what asset classes or markets they get bigger in or get out of and they can make that decision really quickly to get in and out, depending on how market conditions are changing, or volatility. So the intersection of these like two big mega trends of more and more inflexible flow with this kind of very much more fragmented intermediation can cause pretty big price moves. In my opinion, it’s actually one of the reasons why we’re seeing these big intraday swings, that when you see this coalition of more and more trades that have to get done with market participants that are fragmented and they can choose to provide liquidity or not or widen out spreads, depending on it. So you’re just seeing that there’s just a much, much bigger impact and market structure in general has gotten to be a bit more unstable. Now on balance, I actually think that this is, it’s good for flow strategies because frankly, there’s more opportunities to capture. But it also really changes how you have to look at risk and portfolio management.
(26:56)
When I think about some of the flow strategies you’ve mentioned, two categories sort of come to mind for me. There’s one where a counterparty might be a party that with nonprofit motives like an index fund just rebalancing you know, introducing Tesla for example, they just have to do it versus a counterparty that maybe does have profit motives in mind like a corporation doing a buyback or issuing shares where they might have information you don’t have and it almost seems like an adverse selection problem for those style of trades. So I’m curious, how do you handle a situation where the other side of the trade does introduce that adverse selection problem?
Sure so Corey actually, I wish I had like a great answer to this question because with most flow strategies, that if you can avoid adverse selection, they’re a heck of a lot more profitable. One thing you can do is try to look at incentives. Like you mentioned, a company selling stock. So can you get some sense of why they’re doing it? So if a company, let’s say a company is in the middle of doing an acquisition and they need to raise cash to close the deal. That’s probably a pretty benign reason for selling stock.
But then you can have the situation where there’s a private equity firm who sits on the board of the company and they’re selling a big block. Now that firm may not know anything specific about that company, but they probably know a heck of a lot more about the industry and the sector than I do. In general, I think it makes sense to think that you’re the less informed party in any transaction. You should assume that I’m the idiot around here. And really, like that’s going to be my main answer. It’s kind of like what people say on the Internet, “That’s what the money is for.” And what I mean by that is when a company is selling shares or a large holder is selling a block, typically they’re doing it at a price discount to market price. And part of that discount reflects this risk of adverse selection that you’re taking.
Because on any of these individual trades, you’re taking irreducible idiosyncratic risk. It’s like playing poker. Like you can execute your strategy absolutely perfectly, but you’re still going to get hit by bad beats. And frankly, like these strategies are not generally good as one-off trades. You really need to be executing a lot of them to kind of smooth out your returns. It’s really one of the reasons why these trades exist. Like I’ll ask lots of people actually, I’ll ask you, right? Would you take a million dollar bet, no tears on a 60/40 coin flip?
Probably not.
And most of you will give the same answer that as a one-off trade even though you have a pretty big expected positive edge and expected return, it doesn’t really make sense for anybody to take it. But you know, most people give me this answer that, “Well, if I could do the trade 10 times or 100 times or something, I would do it, but not on a single coin flip where anything can happen.”
(30:00)
Well that also introduces this… Something you and I have talked about. I know in the past sort of the ensemble risk versus time sequence risk, right? If you have to take those bets in a row and you go bust before you get to the wins, that doesn’t help you. And it makes me think of the event-driven nature of these trades, that you are having these trades come in and out and I would imagine that you go through periods of glut when there’s lots of events and periods of drought when there are very few and it might introduce sort of a sequence risk element to the strategy. I’m curious how you sort of navigate those swings in the number of events that you’re seeing or the types of events that you’re seeing at a given time?
You’re absolutely correct. You go through swings. Part of the reason for that is seasonal. There’s just less activity or issuance around the holidays or maybe in the middle of August. Also, a lot of these effects tend to be calendar driven. So you’ll see a lot more around month-end or quarter-end or around earnings dates or dividends. But a lot of these activities depend on volatility and that can be very, very interesting. So when volatility picks up, normally there’s a drop in the amount of discretionary activity. So like if a company was planning to go public, you’ll start seeing these notices that well, this firm has postponed its IPO due to market conditions. But at the same time, systematic strategies, they continue. So things like index rebalancing will continue to happen and risk parity and CTAs will adjust as prices move up and down.
So the impact of flows generally goes up when volatility is higher and that’s generally positive for flow-type strategies but at the same time, it kind of pushes up the risk because most flow strategy, what they’re doing is they’re providing a market service. And when volatility is higher, the average compensation for providing that service tends to also go up because somebody’s getting a certain outcome and over time I’m collecting this statistical edge. The main part of the answer to your question of how to deal with these swings is, it’s acceptance. It’s like being willing to know that you will go through periods of time where there’s just lower volatility or there’s just not that many attractive opportunities out there and you shouldn’t go out chasing marginal trades.
On the flip side when the opportunity set is greater, you need to be kind of willing to flex your portfolio up and take more trades. And part of the reason why these trades exist is somebody needs to transact and the other side, you need to be a sort of a flexible and adaptive capital provider. So these swings are just, it’s part of inherent to the strategy and it makes it difficult that they don’t really fit into an ETF or a mutual fund structure and that’s one of the reasons why it makes it a bit difficult for them to be completely competed away.
(33:01)
Well that sort of leads nicely into a conversation about position management itself, going on maybe a little down to the microscopic level here with this idea of the market environment is going to change the opportunity set. You could have a great opportunity set where risk is really high or a thin opportunity set perhaps where risk is really low. How do you think about position sizing and entry criteria and exit criteria for these trades?
So for most of these flow strategies, you have a small edge and you’re trying to capture that repeatedly. So your biggest risk tends to be idiosyncratic. There’s known unknowns or unknown unknowns and the best way that you can kind of defend against these tail outcomes is by doing lots of trades, both simultaneously and repeatedly. Like when I go back to that coin flip that I offered you, ideally you want to be flipping 10 coins at a time and then repeating that 10 times. So that way, you’re diversified both at any given point in time, but also across time. And generally, I don’t want to be reliant on correlation or volatility in terms of position sizing or management because both correlation and volatility can be pretty unstable especially when things change.
Correlation in particular is interesting. Like there’s the big market people saying that, “Hey, all correlations go to one in a crisis.” But like really if that was true, that would not necessarily be such a bad thing. Think about you are running an equity long short book. If all your longs and all your shorts go down 10%, you’re still roughly hedged. Like your real risk is that the correlation suddenly shifts and it’s completely different than what you thought it would be. And that’s true for volatility as well. It’s easy to look at different assets and say, “Well historically, this asset has been very low volatility,” but that’s not a good representation of what could happen. I think we’re seeing that with something like stablecoins where if you looked at the VAR of it historically, it might look great, but we’re seeing that one of them is getting wiped out.
Now even if it recovers, that was a much more volatile path than anybody would have done on, looking on a past basis. And leaving crypto aside, even if you’re just looking at traditional stocks, like there’s lots of large cap companies like Boeing or Disney or Facebook or Alibaba that at one point, were considered to be low volatility, even boring companies. But then some news or some event happened and you just saw giant stock price moves and capitalization moves. So you can’t rely really on volatility weights and this is going to sound like really simple and dumb, but if I was going to summarize position sizing, I would just say be small. Because you really don’t know what are all the future outcomes of any trade. And so in like both markets in life, you just need to survive long enough to get lucky.
And then you’d ask me about entry and exit. So entry and exit, this is one thing where I think like flow trades are different. So with most trading strategies, price is your primary guide to how you get in and out. With flow strategies quite often, it’s about a time horizon. So like if you’re modeling a particular flow and you’re expecting to have an impact, so let’s say rebalancing a portfolio or index add delete or a bond auction, then that flow effect should happen sometime around that event and maybe you’re right or maybe you’re wrong. Like sometimes, maybe the flow was already priced in by the market or it just wasn’t that large or some other news came up and there wasn’t a flow impact. So regardless of whether the trade worked or didn’t work, you need to exit. So again, it’s like playing poker where you see your cards, you play your hand and then you move on.
(37:00)
When we discuss these trades in isolation, they all sound like convergent trades where you’re waiting for a spread to close and the big risk is that the spread ultimately blows out against you. So you I guess in theory to have these sort of positive expectancy, negative skew-type trades, how do you think about building a portfolio? Or I guess how do you prevent a portfolio comprised of these types of positive expectancy, negative skew-type trades from ultimately blowing up in your face?
I don’t actually think that’s generally true across the entire portfolio. I definitely think that there can be negative skew. You know, we’ve talked about a company selling stock. A really public example, that is what happened around Archegos where a whole bunch of banks found themselves with basically the same position and when they all had to dump stock, it created a price impact and a flow impact that was probably much, much larger than anyone had anticipated.
But even a company selling stock can sometimes have positive skew. And I’ll give an example of that, that around Covid, a whole bunch of companies share price had collapsed because people were pricing in that there was a real risk of these firms would go bankrupt so like you can think of things like airlines or cruises or so on. So if these firms were able to place shares, even if they were doing at a big price discount or even if they were diluting out their shareholders, often once these companies had raised capital and sold shares, their prices rallied you know, not because necessarily Covid was over or their businesses are in much better shape, but this near-term insolvency risk got pushed out.
So broadly, I don’t think that flow strategies in aggregate are either positive skew or negative skew. But what I will say is that they tend to have very wide variance in both directions. So these things tend to be very low Sharpe ratio on any trade and it’s part of the reason why these trades continue to exist. Again, going back to you don’t want to be flipping a single coin and it also deters competition from entering.
(39:05)
So given the different types of flow trades that you categorized and that you’ve got this varying number of flow trades going on at any given time, how do you think about portfolio construction and diversification?
So I’ve already talked a lot about finding edges and sizing trades small and doing lots of trades. So the only thing that I’m going to add here is diversification and I know that’s a kind of a boring topic so let me try to give you a visual. So imagine you’re running a casino. You want not just lots of customers coming in through the door, but you want to be offering like lots of different games whether it’s poker or roulette or craps or slot machines and that’s partially because of what I talked about before that different people are interested in different areas at different times. So flows in different strategies you know, may change depending on what people are interested in.
But another big part of why you want to have this diversity is maybe there’s one of your roulette wheels is tilted and some number keeps coming up more often than it should. Or maybe there’s like some really smart woman who’s sitting at a blackjack table and counting cards. So what she’s doing is that you think that you have a positive edge in offering blackjack but instead, now she has a positive edge that you didn’t anticipate. And the nature of statistics is that it can take a very long time to figure out whether a strategy has stopped working or you’re just kind of going through a period of bad luck.
I think Corey, like you wrote a paper that basically said that by the time we know for sure whether the value factor has stopped working or not, we’re all going to be dead. So instead, you want to be running like lots of different strategies so you can absorb the hit if one of them is not working because of bad luck or it’s been crowded or you’re facing a much more savvy competitor. If you only have one investing style, you can’t really diversify enough and this is why it’s really important to build an ensemble I know you love that word – of lots of lots of different strategies. Even within a strategy as much as you can, can you diversify across different assets and countries and sectors? So a big picture, I think I’m going to repeat myself, but it’s the same things. It’s like have an edge, be small in each trade, do lots and lots of trades and diversify as much as you possibly can.
(41:33)
One of the things that really dramatically changed in the last couple of years has been the role of social media in the influence of traders. So we can talk about the Reddit meme mania, but I would even argue that someone like Elon Musk was in many ways, able to meme his stock into the S&P 500 and meme his way into shoring up his balance sheet by creating people who were just avid fans and would buy the secondary issuance. Our friend Lily Francus has written about the idea of salience and social leverage in many ways. How do those risk factors change the profile of these trades?
That’s become much more important as you’ve given lots of good examples of that. The way I think about it is like lottery tickets we all know how a negative expected return. And I hate to admit this, but like sometimes I bought lottery tickets. I’ve been in an office and there’ll be like a big Powerball jackpot and somebody will put together a pool. And I’ll join in because I have this like tiny fear that what if everybody in the office wins the lottery except for me because I like I wouldn’t chip in $5? So meme stocks can be like that, that you have lots of people who either know each other or they meet online and they all decide to pile into a particular stock. But there’s a really big difference between lotteries and memes. So with lottery tickets, it doesn’t matter how many tickets I buy or my friends buy. We’re not going to change the outcome but that’s absolutely not true with stocks. If you have a whole bunch of people who are all piling into the same name especially if they’re using call options, it’s going to cause an impact on price. And this is what I tend to call social leverage. And what you see with that is that people who come early into a trend, they tend to benefit the most and then people tend to come in afterwards.
The other thing is, is that the flow effect is originally when people start buying these shares, the most flexible holders tend to exit out because they’re happy to see this pick up in price and then the people left are much, much more inflexible. So you end up with in the example of Tesla, index funds have a huge holding in it. Momentum strategies for sure have gone into it. But you also get these stocks tend to be “culty” or at least that there’s a lot of true blue believers that will not sell no matter how high the price goes. Then what happens is that even the same amount of flow can again start to cause this nonlinear impact where there’s just so much supply has been sucked out by all these inflexible holders that whatever flow has to jam up the price from a few people that are willing to transact.
Now like the old saying in markets has been “markets take the escalator up and the elevator down.” And I still think that’s like mostly true for at the index level, but that’s not true at all at the single stock or crypto level. Like you definitely see lots of examples of things taking the elevator up and taking the elevator down. And then once that happens to a particular stock, people know that, “Hey, it can happen again.” It’s like the famous for being famous, and that’s what I think Lily Francus, she coined the term “salience.” And once something happens that it becomes a salient or a meme, then the risk profile of that security totally changes, because it now has a completely different holder base and price impact of flows and option markets. Like nobody is out comparing AMC’s valuation to other movie theater chains. Nobody’s really looking at AMC’s valuation at all. Same thing is true with GameStop where you’re not going to try to build a long short model where you compare it to other mall retailers. And the hedge funds that didn’t take that into their risk management, there’s certainly examples of where they did quite poorly because of that.
But I don’t want to just pick on retail meme stocks. There’s lots and lots of examples of this. There’s thematic ETFs or you’ll see there’ll be growth hedge funds that all piled into a relatively small number of tech names. And the thing is that once that happens, then you will see that a whole bunch of businesses and securities that should not necessarily be correlated because their businesses are quite different. Like if you think of a thematic ETF that owns electric cars and biotech companies and internet companies, the underlying businesses are very, very different yet they will often all trade together because they have the same set of holders. Same thing we’re seeing across hedge funds where if you look at like the results for a lot of tech names, they range from being very, very good to not so much but yet, they’re all trading together. And so when you start seeing very high correlation across assets that shouldn’t necessarily be correlated, that’s a really good sign to me that it’s flow effects that are dominating.
(46:46)
In the call we had together in preparation for this episode, you made an offhanded comment. I don’t even know if you remember saying it, but I wrote it down, that “Signal turns to noise because of arbitrage, but noise turns to signal because of reflexivity.” And I thought it was a really interesting thought and I was hoping you could expand upon what you meant by it and why you think it’s an important concept.
So the first part of it, signal turns to noise, that’s just a reminder to me that markets are adaptive and flexible. Even if I have some great edge or signal or strategy, I’m never going to publish an academic paper about it. But there’s lots and lots of smart people in markets and it’s pretty likely that other people are going to discover the same thing and they’re going to start to do it. So even if general strategies can persist over time, how exactly you do something, it generally makes sense to adapt and evolve over time.
A good example of that is like what’s happened with option selling. So for a very long time, implied volatility was consistently averaging well above realized volatility. And because of that, you could instead of buying the S&P 500 Index, you could every month sell an at the money put option on it instead. And both those strategies ended up with roughly the same return, but the option selling strategy had a lot less volatility and smaller drawdowns. And that’s sort of the market equivalent of a free lunch. So you saw that lots of institutions piled into systematic options selling programs and because of that, if you look at put selling especially if you do the part of the compensation you’re getting is for taking equity risks. So if you isolate out the volatility component, the returns for selling volatility over the last five years is somewhere between zero and negative.
So this is an example of like there was absolutely a real signal there around in volatility risk premia but in the sheer amount of money that went into it, it took this signal and people made and firms made you know, huge fortunes from volatility selling, but it’s an edge that got completely ground out and now it’s basically noise, at least for now. This might change and it’s something to watch. But it’s something that’s really not a great signal anymore.
Now the flip side of that where noise turns to signal, that’s something that is a little bit more complicated, but I think is more interesting. And that is that a signal exists because people have belief that it does. And I’ll give you an example of that which is since I started my career, like the dominant trend average is the 200-day and you’ll see it everywhere. You’ll see it on TV, in research, it’s probably the single most cited technical indicator. Now, why 200-days? Like why not 225 days, why not 250 days which would be roughly a year? And the answer or at least what I think the answer is, it’s hidden in 1987. And look, I wasn’t trading then but I’ve looked at the data from that year and one thing you see is that if you were following a 200-day moving average strategy, the S&P closed just below the 200-day on Thursday, October 15th. And with most trend strategies you know, you take a signal on the close and then you transact the next day. So if you are following a 200-day moving average, the signal would have tripped on Thursday and you would have exited your long position on Friday, October 16th. So at that point, you would either be flat or you would have gone short.
Then what happened? Well, the next business day, Monday, October 19th, the market crashed. Stocks closed down more than 20%. But if you were following a 200-day moving average, you were not exposed to that crash at all and then as the market recovered eventually, you got back to being long again. Now imagine that instead of a 200-day average, you were following a 202-day moving average strategy. So when you look at that, the S&P didn’t close below it until Friday, October 16th. So you would not only have been long into the crash on Monday and lose 20% but because the signal had just tripped, you would be selling all the shares that you held on Black Monday which turned out to be in hindsight of course, the very bottom of the market. So there’s a huge difference in returns and outcome whether using a 200-day moving average or 202-day average.
Now to be clear, like I’m a big believer in trend strategies. I work with some really great colleagues who have an exceptional CTA model. But I don’t think that there’s any sort of magic parameter out there, that there’s some physical constants of the universe that, “Hey, 200 days is the right trend length.” Instead, the right approach is to use an ensemble of it, but there is some randomness that happened in 1987, that 200-day happen to work really well. And now you know 35 years later, that is still probably the most popular technical indicator and so it’s something that my opinion started out as almost noise or luck or randomness, that is now a very important signal.
(52:01)
We’ve spent a lot of this conversation talking about these different flow trades, but I know you also spend a decent amount of your time modeling the position dynamics for macro strategies like CTAs, risk parity, target vol funds, target date funds, all of which can have their own flow impacts. But I’m curious given the prevalence of index rebalances and additions and deletions and corporate buybacks, why is this higher level macro flow analysis important to your process?
So there’s a few reasons for that. One is that I’m not alone and most banks and research firms will build out some sort of position models of systematic strategies. But unless you build it yourself, you really don’t understand like what assumptions go into it and what are the limitations of your data gathering and your analysis? We all get these e-mails that say that, “Oh, well if the market goes up, then CTA firms will be forced to buy at this level and they’ll be long, but if the market falls then there’s going to be selling pressure at this other level,” and it’s really hard to like make sense of any of these things unless you’ve at least looked at it yourself.
The next reason is that look, these systematic players can be very important. I think you highlighted it in your work on liquidity cascades and one of the things that you showed was that one of the reasons why the market dropped so much and so quickly in March 2020 was the presence of volatility target date funds being forced to de-lever. And that was a really interesting feedback loop if you think about it because if somebody’s using volatility as a signal, it’s forcing them to let’s say, sell a huge amount of S&P futures at the close which pushes down the price, which then increases realized volatility and which makes it much more likely that the next day, they’re going to have to continue to keep selling more futures.
So the other thing that’s really interesting there is it’s actually quite similar to what happened in October 1987 with portfolio insurance. People generally are trying to look for strategies where they can capture all the upside of equities while cutting off the downside. And those things, they don’t really work if everybody’s trying to do the same thing because there’s an inherent assumption in them that their selling is not going to have much price impact and that might be true in normal times when volatility is low and the selling is small. But when the numbers get big and volatility high, then these things have a pretty big impact. And even if you trade individual stocks, you’re still going to get affected by what’s happening at this index level. If a huge futures or basket or ETF trade is going through, it’s going to affect whatever stock you own, whether you like it or not.
Finally, the thing about all these flows is that even if they’re individually not that big, sometimes they can be additive, be quite important. I should be calling this like flow stacking, like you know, return stacking, that sometimes you can end up with multiple strategies that all need to buy or sell at the same time or similar instruments and because of that, you can again see these more nonlinear impacts because everybody models that their individual trades are relatively small. But if lots and lots of different market participants and actors are doing similar trades then effectively, they’re acting as a one much bigger, larger trade.
(55:30)
Sort of the last piece of the portfolio design puzzle that you and I have discussed in the past when it comes to these flow trades is the application of portfolio level tail hedges that this is a key part of your risk management strategy. I was hoping you could walk me through sort of the why behind that decision because obviously, tail hedging is a really hotly debated topic as to whether it’s additive or not. And then since you have elected to implement tail hedges, maybe how you think about implementing them in practice?
So hedging is something that we do on a team basis because it’s really important to look at the entire risk of the portfolio, not just any particular trade or strategy. So based on the entire portfolio, you want to be sizing and implementing your hedges. Now for the why we run tail hedging, there’s two big reasons. So the first one is like the obvious direct benefit that you get that hopefully during a period of market stress, you’re less correlated, you’re taking a smaller loss and you don’t find yourself in a big hole that you have to dig yourself out of.
But the second one which I think is much less appreciated, is that when you have some hedges on, it helps you take advantage of the market opportunities that come up. And what I mean by that is that look, if you’re down a lot, you’re not mentally going to be in a good position to take advantage of whatever opportunities are coming your way. I really liked the Walter Deemer quote that, “When the time comes to buy, you won’t want to.” But instead if you’re in a position where the hedges are helping you and you can look at what opportunities are out there and you can add positions at much more attractive levels. So we don’t see hedges as a profit center. Over time, I expect that our hedges will generally lose money, at least in isolation. We’ve actually done better than that but certainly that’s not something that you want to build in. But more importantly, hedging gives you this mental edge of being able to take on more risk when other market participants might not be able to because of financial constraints or psychological reasons.
Now for that how we implement hedging, we have sort of three main types. So the first market risk is gap risk. So that is like we’ve talked about a good examples of October 1987 and March 2020 where there’s like a huge move in markets accompanied with a jump in volatility. So for those kinds of events, you typically want to have a long volatility hedge that has a nice meaningful payoff in a crisis period, but that you can hold in non-crisis periods and not have a huge drag.
The second type of hedge is against more sort of gradual and slower moves, whether up or down. So an example that is stocks in 2008/2009 where they tended to grind lower until March of 2009. So trend/CTA tends to be ideal for those kinds of hedges. We’ve seen in the last year pretty consistently, commodity prices have gone up, bonds have gone lower. So that’s been a good environment for most trend/CTA type strategies and that’s been a good portfolio hedge, even during periods where vol has not necessarily been as reactive.
The third type of hedge is the hardest. It’s around events. So for that is where you’re looking at the world today and saying that there’s a market is not really pricing in that the world could look quite different in the future. Going back to what I was saying at the beginning around taking a look at the entire city from the sky that you could see that right now there’s no traffic, but you can also model in that you know, maybe there’s no traffic at noon but at 5pm, there is going to be. And a good example of that would be what’s happened with bonds. So a year ago, not only were bond yields low, but bond implied volatilities were also low. So you could structure something that you would have a nice payoff if rates were to rise, but with a fixed known downside cost. So that was a good example of an event hedge, but I will say that they tend to be the hardest because you need to get not only at least directionally, the timing and predict what’s going to happen in the future, but you also need to structure it in a way that you get a payoff that’s much greater – he odds of that happening is much greater than the market is anticipating.
(59:58)
Well, we’ve come to sort of the last question, I guess of the episode here, unfortunately. But I’m excited to ask you this because you’ve had a varied career. And the question I’m asking everyone at the end of the episode is to reflect back upon their career and think about what was sort of the luckiest thing that happened to you or the luckiest break that you got in your career?
So for my career, I think the luckiest thing was that even though I started as a programmer and then eventually in a quant, I was able to get the opportunity to move into a portfolio management role. So to me, the luckiest break was this colleague that I’d worked for at a prior hedge fund, remembering what I had done and when he moved to Henderson, contacting me. So that’s something where you know when I look back, it’s a relatively hard transition to make from being a much more quant to being able to run a portfolio so I think that was, I’m super appreciative of that. I think it’s much more common now to see that people with a much more technical background because of, ironically, the rise of all the quantitative strategies, you’ll see much more people with geeky backgrounds like me get into fund management. But when I got that role, it was definitely unusual. It was also relatively unusual for people to be doing strategies that blended both quantitative analysis with financial analysis. So I really lucked into having a role that really fit my skills and interests.
Well Aneet, this has been fantastic. I can’t thank you enough for joining me today. I know our listeners are going to love this episode so thank you.
Thanks, Corey. It’s been a pleasure to be on.
(End of Recording)