In Episode 3 of the The Acquirers Podcast, Tobias chats with Corey Hoffstein of Newfound Research. The thing that really stands out about Corey is his risk forward approach to investing. Everyone distrusts back tests but Corey has elevated his distrust into an art form, or at least an investment philosophy, he’s really taken to heart Richard Feynman’s famous proclamation that the first principles is that you must not fool yourself and you are the easiest person to fool, or in other words as Tony Montana said it in Scarface, “Don’t get high on your own supply.” Corey provides some great insights into:
- No Pain, No Premium In Investing
- Why Investors Should Beware Of Backtest Results
- How Often Should You Rebalance Your Portfolio
- In Investing The Counter To Fooled By Randomness Is Fooled By Narrative
The Acquirers Podcast
You can find out more about Tobias’ podcast here – The Acquirers Podcast. You can also listen to the podcast on your favorite podcast platforms here:
Tobias: Hi, I’m Tobias Carlisle, this it the Acquirers podcast, my guest today is Corey Hoffstein of Newfound Research, I’ve known Corey for a few years and I discovered him years ago reading some of the super smart research that he was putting out. The thing that really stands out about Corey is his risk forward approach to investing. Everyone distrusts back tests but Corey has elevated his distrust in it to an art form, or at least an investment philosophy, he’s really taken to heat Richard Feynman famous proclamation that the first principles is that you must not fool yourself and you are the easiest person to fool, or in other words as Tony Montana said it in Scarface, “Don’t get high on your own supply.”
Speaker 2: Tobias Carlisle’s the founder and principle of Acquirers Funds, for regulatory reasons, he will not discuss any of the Acquirers Funds on this podcast, all opinions expressed by podcast participants are solely their own and do not reflect the opinions of Acquirers Funds or affiliates. For more information, visit Acquirersfunds.com
Tobias: So it takes an enormous number of backtests to get that kind of distrust of back tests, and so I want to know how did you develop that distrust and how do people fool themselves with backtests?
Corey: Yeah, this is a rabbit hole, well first and foremost thank you for having me here, really excited to be chatting with you. This is just a rabbit hole, I mean you could just stop now and if want I can go on for an hour. But this is one of those areas where I think people are very fooled by back tests because they seem very scientific, and I think unless you have a deep history with backtests they seem very deterministic, like, “This was the process, this is the outcome and it is trustworthy.” And I think I see that most often with the community I work with of financial advisors who might be looking at the newest smart beta product and trying to evaluate how it does and saying, “Oh it did better than this other product that came out in the backtest, therefore it must be superior.” And the way I often try to draw the connection for them to get a better understanding of a backtest is, I say, “Look, you run Monte Carlo simulations for your clients all the time right when you do financial planning.
Corey: Imagine if you went to a client, ran a Monte Carlo simulation and picked out the very best possible simulation for their financial plan, showed it to them and said, “Here’s a backtest of your financial plan.” It seems sort of ludicrous but in a way, that’s a lot of what comes to market with backtests is this very, unfortunately, data mined unsophisticated approach to bringing something to market, and it’s taken as if it’s some wonderfully deterministic scientific process when in reality, there’s all sorts of uncertainty around it, and I don’t think that uncertainty is really appreciated by those who it’s being presented to.
Tobias: We do a lot of back tests as well, and the purpose of a backtest, aside from presenting it to somebody else, you’re trying to learn something yourself, which of these metrics is the better metric, and that’s something that you’ve been critical of in the sense that you’d rather not choose one single metric, you say, “Let’s look at several different ones.” And you’ve also said that rebalancing is something delicate and you have to be very careful when you’re determining which rebalancing data, can you just expand on that a little bit?
Corey: Yeah, so the very sort of naive analogy I use when it comes to building a portfolio, is there’s ingredients and then there’s a recipe, the ingredients are some of those sort of signals you talked about, is it better to use price-to-book, is it better to use enterprise value to EBITDA when trying to come up with value signals, and then there’s the recipe which is, well how often do we look at these signals, how do we rank stocks, how do we weight stocks? And ultimately it takes both of these ingredients in the recipe to bake the final cake that we’re ultimately going to serve. And both of them can have a tremendous impact, I think very often in the industry we’re very focused on what’s the secret sauce, which tends to be the signals. But very often, it is in the recipe that we determine how consistently we meet our objective, it’s in the recipe where we add extra constraints into what we’re doing to try to hedge our uncertainty. And a lot of that area of research goes very overlooked, I do want to go back to something you said though, which is we backtest to learn.
Corey: And I think one of the things that often goes overlooked with backtests is, people look at a backtest as evidence of something, and I think to me, if you’re looking at a backtest as evidence, you’ve already sort of lost the battle. Ultimately what you should be looking for with a backtest is, first the evidence should come from the process, we should believe, for example, that buying cheap stocks should outperform expensive stocks and we should have in some sort of economic intuition as to why that is. And then what we should look for in a backtest when we evaluate it, is not something that works all the time, but we want to see the warts, we want to see where it fails and we want to make sure those periods of failure line up with our intuition. A value strategy should underperform in the dotcom bubble if someone presents to you a value strategy that does well during the dot-com era, you really have to scratch your head and say, “I this value?” And I think a lot of that goes overlooked.
Corey: People sort of design in the review and they have these, I see it all the time, tactical strategies that are supposedly trend following, and somehow magically get back in at the bottom of 2009, I’m sorry you’re not trend following if you got back in March 2009, all of the trends were negative. And so I think a lot of what we should be doing with the backtest, it’s not confirming evidence, it is simply trying to look at different scenarios and see how a lot of these things play out. But to your point, you said there’s a lot of these signals, a lot of the recipes, so much of that can affect a backtest and again I think it goes often overlooked and to your point, rebalancing is one that I’m more than happy to talk about, can have a really dramatic impact and that’s one of those little, tiny details that really goes overlooked.
Tobias: So I find that particularly interesting, and it’s something that you may not familiar with it if you haven’t looked at a very large number of back tests, and there are lots of different ways to deal with rebalancing, this is something that I’ve learned from you, that you don’t have to rebalance your entire portfolio on a quarterly or annual or half yearly basis. A smart approach might be to break the portfolio into a number of sort of sub-portfolios and then rebalance each on a different date, but then each is still a quarterly or monthly or annual, whatever the rebalancing date is, but you’re only rebalancing one fraction of the portfolio on that date, so do you want to talk about that a little bit?
Corey: Yeah, and this is something that came up for me because my background is being a quant, so all of our strategies are very systematic in nature where we have very defined rebalance processes. I think sort of nuance often gets hidden when you have a more discretionary manager who might be more opportunistic in their buying or selling. But when you have a very systematic process, what you can see is that the opportunities presented to the strategy is going to be very much driven by when does the strategy rebalance. So let’s use just a very simple example, let’s pretend we have two value strategies, deep value strategies, they’re trying to buy the junkiest stocks out there and one of them rebalances at the end of the year, every December and holds the stocks through the rest of the year, and the other one rebalances at the end of June. And I think when you think of that, even though they’re identical strategies applying the same process, it’s very easy to understand that, well the opportunities that are available in the market in December might be considerably different than what you see in June, particularly the more concentrated your portfolio is. The more you’re willing to hold just a few super deep stocks, deep value stocks, those few super deep value stocks can change dramatically in six months.
Corey: And so you can end up with very materially different performance even though both strategies are ultimately trying to achieve the same thing. So I actually just wrote a paper about this, and I tried to distill it to maybe the most naïve example which is just a static 60/40 portfolio. And what we tend to see with a lot of advisors is they rebalance their client portfolios once a year, and so what I wanted to show was, “Look even something as simple and naïve as a static 60/40 rebalanced once a year can be subject to this tremendous amount of timing luck.” Which is what I call it, and the example that I showed is if you have a 60/40 portfolio that rebalanced at the end of every February versus one that rebalanced every August, somewhat arbitrary dates but it’ll make sense in a second, the performance disparity in any given year is normally about 50 basis points, and it’s pretty random. But if you looked at the one-year performance disparity from March 2009 to March 2010, it was 700 basis points.
Corey: Which makes perfect sense when you think about it, it was a contrived example but it was because the portfolio that rebalanced at the end of February almost perfectly market time to bought them, rolling back into equities versus the one that rebalanced in August was the furthest away from that 60/40 mix you could get. And so, what you saw was that by rebalancing right into equities near the bottom then having them rip back versus the delayed rebalance that took place six months later, you ultimately saw a really big difference in the performance that was materialized and it’s not something you expect to mean revert, it’s not necessarily something you expect that that August portfolio is going to recapture from the February portfolio at a later date. And so, it’s one of those tiny details that often goes overlooked, but when we talk about managers getting hired or fired after a three or five year evaluation period, that can be the difference.
Tobias: That’s amazing, and it’s something that we looked at, not rebalancing, we wrote quantitative value well before I had heard of that idea, but what we were conscious of that January effect is a very well known sort of equity, there’s a little bump in January because I think the idea is, that there’s for value stock particularly there’s some tax loss selling in December to try to capture those short term losses and then rebuy in January. And so they see a little bump, and if you assume that you’ve bought ahead of that, then you capture that then you do fairly well, you also got the reporting, the annual reports come out a little later than that but then you’re sort of assuming that you’re trading on that. So we tried to rebalance in June using the year-end data, so it was six months old, so it’s a slightly different effect that it’s capturing, but it is material when we looked at different rebalancing dates that we could have included and one of them, if you’re rebalancing on a quarterly basis then you’re close to the bottom in that March rebalancing, that’s a good one, you also capture the January rebalance date. So you do see some better performance.
Tobias: This is reasonably heady stuff, and I see it in your research papers, very well thought out very thoughtful research papers. I’m sort of interested to know how you got to that point, how did you start out as an investor? What’s your academic background?
Corey: I think with a lot of these things, there’s a lot of nature and a lot of nurture. I sort of tripped and fell into finance, but my approach to finance has always been one that’s very risk-based. So instead of the pursuit of returns, I’ve always had this sort of view of how can I try to control my unintended bets? How can I try to control the things that are going to be the most damaging to my portfolio? Again, I think part of that’s nurture, as embarrassing as this is to admit, my parents their nickname for me, when I was younger, was Safety Boy, I was never the kid running around the pool, I always had my swimmies on you know? So I think just very naturally I’m a risk-averse person, and I look for those areas where, “Well what is the unintended bet I’m taking, how could this go wrong?” From a nurture perspective when I was in graduate school, I attended Carnegie Mellon’s Masters of Science and Computational Finance, and its sort of a cross-disciplinary program all about pricing complex derivatives.
Corey: And as I was going through that program, I noticed this sort of repeat formula like approach that we were taking which was trying to identify a risk, trying to isolate it, extract it, package it into some derivative, and then price it and trade it away. And that the price you were paying was ultimately a transfer of that risk, and what it ultimately did was it sort of informed my view around finance, that everything we do in finance is not really about the pursuit of return, it’s actually about the transfer of risk. Who’s willing to bear it, and therefore if they should earn a premium, or at least expect to earn a premium for bearing that risk, and who’s trying to get rid of it and is therefore willing to earn a little less. And so I think for me those two experiences, both sort of the nature side of it and the nurture side of it led me down this path where, as you mentioned, I write a lot of very sort of maybe heady quantitative material, a little wonkish by nature but a lot of it is sort of circles the drain on these topics of well how do we think about managing risk? Can you do it with correlation, can you do it with different types of payoffs with these different style premia, or trend following, different ways of designing a portfolio.
Corey: What is sort of the opportunity timing aspect of it that we need to be aware of, and all those things have some interaction effects, they all sort of culminate to how much diversification you’re trying to take advantage of in the portfolio. And then it’s all about the tradeoff of, what do we have confidence in, where do we think we have an edge and we want to bear risk versus what risks do we really think aren’t worth bearing and how should we think about diversifying those away?
Tobias: When you came to launch your firm, was that something that you thought would be a … it is rare to hear somebody … risk first is an unusual approach, most firms come out thinking that they’ve found some trick to the market that there’s something that others are missing. You’re risk first, that’s the first thing that you see on your webpage that risk is your focus.
Corey: Yeah and it was from day one, and again embarrassing to admit, I sort of tripped and fell into this business. We just passed a decade of being in the business, but it wasn’t something I had actually intended to start my own business in. What had really happened was, when I was younger, when I was an undergrad, I was actually sort of doing an internship for my father’s financial advisor a couple days a week. It was one of a couple internships I held at the time, and one of my jobs was to interview different managers, so this was around 2007, and this small-cap values manager came in the door and ran a mutual fund and before the meeting really started, I just sort of asked him what he thought of the market, and again summer 2007 had the most bearish outlook of anyone I’d ever heard, just said we were headed for this horrible recession, “batten down the hatches, it’s going to get ugly.” And I was a little taken aback, most people at that time, in retrospect the cracks were showing but most people I spoke to weren’t that emphatic about it. And so I said, “Oh wow, what are you going to do?”
Corey: And he said, “Well, I’m small-cap value manager, I manage a mutual fund that needs to be 95% invested by mandate, and you want to know, I don’t even know who’s invested in my mutual fund. So what I’m going to do is just give the best small cap value exposure I can. And you wanna know what, if it’s not the right exposure, it really should be up to the financial advisor or the individual to determine whether they should be in the portfolio or not.” And I said, “Okay, that kind of makes sense right.” He doesn’t know who’s invested, who is he to say, “Hey now’s not the time for small cap.” If someone wants small-cap if that’s what they’re buying. So after the meeting, I go to my father’s financial advisor, and I sort of relay the conversation and I say, “What are your thoughts?” And he said, “Well, I think that’s crazy, how am I supposed to know whether it’s a good or bad time to be in small-cap value, that’s why I hire an expert.” And I’m sitting there thinking to myself, “You’ve got two people pointing the finger at each other as to who is ultimately responsible for managing risk.”
Corey: And risk in this sense really being managing that how much downside participation are you going to have, who is ultimately responsible for saying, “I lost the money and it’s on my shoulders.” And so, I looked at my very paltry sum of life savings that I had invested in the market, and said, “All right, I’ve got to figure out how I’m going to manage risk.” And it sort of ultimately evolved into this research endeavor, I didn’t know it at the time but a lot of what I was researching were different trend following techniques and designing trend following portfolios. And ultimately started Newfound research because I was asked to license some of that research I had done to another firm, and so it really started as a research firm. But again always about, at its very core, that aspect of, “How do I manage risk?”
Tobias: Do you remember what the small cap manager was looking at while he thought that the market was so toppy?
Corey: I don’t, I will tell you a funny story though, I was once at a conference, and I had told that story to someone else, the other person was supposed to introduce me at the conference because I was supposed to speak, and I was planning as using that as my sort of introduction as to who I was, and this guy gets up in sort of the intermission to prepare and talk, let me come up and talk, and he tells the story as if it happened to him. And I’m like, “Wait, hold on, hold on this is my story.” And this was my whole introduction that I was going to give about me, and he just totally stole my story. And I don’t even think he realized he had stolen it, it was just like one of those, he just sort of passed it off as his own, but he didn’t know I had been the one who had told it to him like a week before, and it was one of those like, “Oh, now I’ve got to come up with an on the fly story.” But I don’t remember who the value manager was, I really wish I did, I would love to reach out and just say like, “You don’t know how you changed my life in that moment.”
Tobias: It’s one of those things that, I’m a value guy so I think about value a lot, and one thing that I’ve learnt over the sort of 20 or so years that I’ve been looking at the market is that, values a terrible timing tool. And so, the examples that I always give, when I started working it was the early 2000s, April 2000, it was the top of the market for dotcom stocks and they all fell over after that but value did very well through that period. So if you’d been a value guy, looking at the market and pulled out, that would have been catastrophic because you would have missed one of the best periods for value. And similarly in 2007, value had had a spectacular run, it had been sort of underperforming and it did some catching up, and sort of once again if you looked at value, it just wasn’t at all predictive. And any of those measures that Cape, Shiller PE, Tobin Q, and of those sort of market level measures, they look at different things, they’ll tell you that the markets very expensive, but that’s kind of meaningless, you can be five years, ten years beyond the point where the market looks really expensive and here we are sort of 20 years past the point, more than 20 years past the point where the market first started becoming unusually expensive.
Tobias: There have been some crashes, but we’re up a lot over that period, and you could have done quite well. Do those other things give a better insight, does trend or momentum, do you feel like there’s any more in that?
Corey: I think you bring up a couple really good points, the first is that I think it’s very hard to look at a particular factor and say it’s going to do something in a given market environment, every market environment is different by definition. I think the null hypothesis we should generally hold is that the market is ripe, there’s a lot of hyper-competitive, very intelligent individuals that are involved in the markets, and if it were easy to make money everyone would be rich, but the reality is we’re all competing, making the market more efficient, which ultimately benefits everyone to have more efficient transparent prices. But at the end of the day, it does make it harder for any of those individual signals to have meaning. Does trend or momentum necessarily have more insight? I think one of the interesting things that you see with something like trend is that it has historically been a great predictor of short term future returns. So, if you have positive returns over the last, let’s call it six to 12 months, you tend to get positive returns statistically over the next couple months, similarly, if you see negative returns, you tend to get future negative returns.
Corey: And so, some of it is, look you can’t have a bear market without continuous negative returns, it’s very hard to have that without it, especially a prolonged bear market. So, there’s a bit of intuitive sense there, that you’re going to see a negative trend, doesn’t necessarily mean trend falling will be successful right, you can get whipsawed along the way on the way down depending on the path-dependent nature of it. But I do think what the evidence has suggested is that something like trend following can be a great confirmation to other signals. So if you think the market’s overvalued, just timing on value alone really hasn’t been very successful, but if you look at times when the market overvalued and you start to see negative trends, well then all of a sudden the signals become a lot more effective in combination historically, because you’re not only getting a negative trend, but you might have a real catalyst for valuation contraction, that the market may lose its enthusiasm for risk. Everyone might say, “You want to know what? We don’t have as rosy an outlook.”
Corey: Or everyone just doesn’t want to hold securities and everyone decides they want to sell, the reality is there’s no money on the sideline, every stock that’s sold is bought to entice someone to buy, you have to see a deflation in price and decrease the valuation multiples to entice someone to ultimately buy-in and so you see that risk appetite decline and that’s when you get the valuation reset.
Tobias: I think trend is a little bit like value, over the last ten years it’s been whipsawed a lot and whichever version of it you prefer if you like the 200 day or you like a simple moving average, or an exponential moving average, whatever you like to apply. Anytime that it’s basically suggested that you hedge, that’s been a bad time to do it for the last decade roughly, so trend is a little bit, in my opinion, trends a little bit like value, you sort of have to believe that at some stage it’s going to start working again, and it does tend to, that’s not unusual behavior for it to underperform the unhedged market right up to the point that the market tips over, and then it’s that cascading behavior of the decline, of the drawdown that sort of somehow, the 200 they sort of pluck it out at that stage and then hopefully you miss the bulk of the decline, you might miss a little bit of the bounce too.
Corey: Yeah, you know, trend’s a really interesting one, so I think when people talk about trend, a lot of what we focus on at Newfound is what we would call trend equity, which is applying different trend following models very specifically to equity markets, that’s sort of our flagship offering. And so in that sense what we’re trying to capture is how can we participate as fully as possible with equity market upside, avoid those significant and prolonged drawdowns, but you do see it in sort of the multi-asset trend following stuff as well that’s really struggled over the last four or five years due to sort of even where it’s caught trends in some markets, it’s getting whipsawed in others. What I think is really interesting about trend and the reason we really emphasize it as a style at the firm is it’s one of the very, very few trading strategies that has a positive skew, a convex payoff. So what we mean by that is where most trading strategies, going back to the sort of risk-based thinking, you are earning a premium because you’re willing to bear risk.
Corey: Which means that you tend to harvest these sort of small alphas at the risk of a big negative left tail, and that’s true for a lot of the economic premias, so when you buy stocks you’re sensitive to growth and inflation shocks, you’re bearing the risk of uncertain future cash flows and hopefully you’re going to get a premium for bearing that risk. But when events like 2008 happen, you’re bearing the downside, so what you tend to see if yup, the whole return distribution is shifted a little bit to the right, your expected return is positive. But when you look at sort of the shape of the distribution, it’s got a big sort of asymmetric left tail, that is true for almost every investment strategy out there. Whether it’s value investing, more esoteric stuff like carry, most active approaches, there is sort of this negative skew that goes along with it, this concave payoff. You’re harvesting a small alpha and to use sort of an analogy, it’s that picking up pennies in front of the steamroller.
Corey: Trend following is the opposite, and it’s a little weird. Trend following tends to have a lot of small losses along the way and then one big payoff. And so from that perspective, I think it can be really, really useful as a diversifier in the portfolio, instead of your traditional correlation-based diversification, it’s sort of this payoff based diversification, that everything tends to have concave payoffs, this is a big convex payoff. And so you can get some diversification in that sense. But it doesn’t mean it’s not really frustrating to hold along the way, that when you say to someone, “Hey, 90% of the time 80% of the time, we’re going to be taking small loses, and you’re going to get whipsawed and you’re not going to participate fully, and nine out of ten years you’re going to hate me as your manager. But in that one year, or two years that there are big positive trends or big negative trends, that’s where we really tend to shine historically.” That’s the sort of thing that I think sounds good on paper, but when it actually comes to building a portfolio and sticking with a portfolio, and having a sustainable portfolio, it can often be really, really tough for people to incorporate.
Tobias: That little bit of underperformance makes it feel like it isn’t working to the point where it actually starts working, and you need them in there at the time.
Corey: Exactly, and part of the problem is you just don’t know when it’s going to kick in, if it were wholly obvious when the end of the cycle is, I’d say, “Well, don’t hold any trend following. Don’t buy fire insurance until you know your house is going to burn down.” The problem is, it’s normally an unexpected event, if everyone sort of knew, it’s all self-fulfilling, if everyone knew the market was going to sell off next year everyone would try to exit before that, it would cause the market to sell off, blah blah blah. The reality is, nobody knows when the next economic crisis is going to unfold, and I’ll go so far as to say there doesn’t have to be one, as crazy as that sounds, I know everyone’s sort of talking end of cycle and I think there’s all sorts of potential catalysts for an end of cycle, but there have been periods in markets where you can go 20, 30 years benign to positive performance. And without any really significant drawbacks, in which case you’re going to look back in 20 years and go, “Wow, I really wish I hadn’t paid that insurance cost.”
Corey: But you’re not going to know until after, and I think that sort of thinking isn’t prudent necessarily going forward, it’s sort of like saying, “Well, I’m not going to put fire insurance on my house because I don’t think the house is going to burn down.” Well yeah, you might get lucky but was that really better thinking? Was that really the right thing to do? Probably not.
Tobias: So, how do you take these risk bases philosophies and this sort of diversification agnostic … and I mean, sorry I mean diversification in the sense of strategies, diversification in the sense of rebalancing, risk-focused, how does that then manifest as strategies in your firm?
Corey: Yeah so maybe the simplest example would be sort of our flagship strategy which is our risk-managed U.S. sector strategy. And what we’re going to do there again is this is what we call a trend equity strategy, where we’re going to try to participate as fully as possible in the economic growth of U.S. equities, U.S. large-cap equities, and then we’re going to try to use trend following to avoid those significant and prolonged market declines. I like to emphasize the significant and prolonged trend following as a category, as an investment style, doesn’t tend to do well in market environments like 1987, those significant one-off drops. What we’re really talking about tend to be economically induced sell-offs, so your things like your 2000s, your 2008s where the market takes time to digest some, some new trends emerge, they tend to be six, 12, to 18 months in length and that tends to be where these sort of strategies tend to do well.
Corey: So how’s it really manifest, well again, what I like to think about is how can we add diversification in the portfolio, diversification to what we’re investing in, so the sort of correlation based diversification. Diversification to how we’re investing, how we’re making these decisions about the trends, and then diversification in when we’re looking at trends, the opportunities we’re seeing. So under the hood instead of just investing in large-cap U.S. equities, we’re going to look at different sectors, and the reason we like to split it up among different sectors is because despite all the problems with sector-based classification, you do tend to find that they act as individual groups. That if I tell you a company is an energy company and I tell you how energy stocks did on a given day, you tend to have a good sense of how that stock did, the groups of fish tend to swim together. And yet, as a whole, they make up the market and so they tend to have a least a positive to very positive correlation, but there’s still a bit of diversification opportunity there.
Corey: Then what we’re going to do, is we’re going to run trend following models on all these different sectors, and it depends on which model we’re delivering to a client or the fund that anyone’s investing in, but what we’re going to do is we’re going to look at each of those sectors and apply a trend following model and determine whether we’re in or out of that given sector. When there’s positive trends, we tend to increase our exposure, and when there’s negative trends, we want to decrease our exposure if not outright remove a sector. So if we can’t find any positive trends on call it a three to 12-month horizon with a variety of different models, you’re probably going to see that that sector is completely removed from the portfolio. We’re going to typically take that capital and try to reinvest it, with the idea that over the long run you expect to earn a premium for holding equities, that equity risk premium. And so we don’t want to just necessarily go to cash, we want to stay invested as long as possible, but what will happen is that once we start to see a large number of those sectors fall out of the portfolio, we’ll put a hard cap, so typically around 20% again, depending on the mandate that we deliver to a client, how risk averse they are.
Corey: But typically around 20%, which means that once you get four or fewer of the ten primary sectors exhibiting positive trends, now we’re going to have cash, short term treasuries in the portfolio, and in fact, if we have no trends across any sectors, we’re completely out. So we can go `100% to cash, now the final element is okay, when are you reevaluating? And again, that’s somewhat mandate-specific, but what we’re trying to do is not just reevaluate all the same model’s end of month or something like that. In all of the mandates that we deliver, what we are trying to do is recognize that the opportunity to look at trends may change dramatically whether we are looking end of month, mid-month, the fourth day of the month, the 17th day of the month. So again, depending on our ability to control trading, we might under the hood rebalance almost every single day, where we’re just going to make a little change to the portfolio, sort of dollar cost average all of our trend changes a little bit, with other mandates it might be a we’re going to look mid-month and end-of-month, and at least then we’re getting sort of this bi-monthly view and we’ll rebalance half the portfolio mid-month and half at end-of-month.
Corey: But the idea there again is, we’re never so concentrated on a single point in time, to rebalance our entire portfolio.
Tobias: So that’s determining whether you’re in or out of a given sector, but then how do you use value, how do you use momentum, how do you use carriers that then for the selection of the individual securities that go into the portfolio?
Corey: So that all depends again on the mandate, this is sort of top level, our flagship, most of vanilla trend equity strategy is just pure trend, we’re not including any of those other elements. Those other elements come into play in other portfolios that we offer, but I think from our most popular offering perspective, it really is delivering that very pure trend following objective, because it doesn’t have the risk of getting diluted with any of the other styles that can come in and again, we are delivering that very pure convex style of payoff versus those other styles can then introduce the concavity. So, what we’re actually implementing with most of these portfolios is very low cost, very cost efficient ETFs, so we’ll use ETFs to implement all the different sectors and most of those are just pure data.
Tobias: Right, so I’m a value guy, and I know you have a value strategy, can we talk about your value strategy a little bit? What sort of signals do you use, what’s the recipe, what’s the magic pudding look like?
Corey: Yeah, so this is one that’s pure in-house, so this is not an offered strategy that we really have, but it’s one that we’ve been running for quite a while, for the partner at the firm. And it sort of came about just simply asking the question, if we who are typically sort of a more top-down tactical shop, were to build a bottom-up security selection portfolio, how would we think about doing it? And a lot of it came back to this idea of what we were seeing, was a lot of these value portfolios, especially systematic smart beta value portfolios in the market, where you balance once a year. And they rebalance to me in a way that makes them incredibly sensitive to this sort of timing luck, there’s a great paper out there that you can link to in the show notes if you’re going to do show notes, written that talks about the timing luck that the research affiliates fundamental indexing strategy was subject to in 2009 and the paper shows that if that index, which I believe rebalances in March, was instead rebalanced in September, the performance disparity would have been something like 1000 basis points in that one year.
Corey: And that’s pretty meaningful, now I know Rob Arnott would disagree with my coloring of fundamental indexing as being a value strategy, to me it’s a value strategy, so it tells you at least to me, value is very much affected by this decision. So that was one of the first big things that sort of made us say, “Well, if we’re going to do this, that’s one of the first things we want to address, is this idea of when are we making the decisions, how are we sort of more continuously pursuing this opportunity over time?” Then the second part is again, going back to this, “Okay, which signals are we using?” Are you using the signals across the entire market, so am I looking, for example, at just to use the sort of academic expression, price-to-book across the entire market and picking the cheapest stocks? Or am I looking at other models that I think might be better? Am I looked at within an industry, so I’m going to look at sort stocks by price-to-book within financials, and then I’m going to go look in energy and sort them by price-to-book, all of those are sort of to me again, open questions about which one works best, I think plenty of people at this point would say price-to-book probably isn’t the best metric to use anymore.
Corey: Fine, I can buy that, but do you look cross-market, do you not? I think another open question was, “Does it make sense to be sector neutral in your implementation or does it not?” What you tend to see with sort of these, what I would call generation one value strategies, things like Russell 1000 value versus Russell 1000 growth, is that they are sector unconstrained. Which means that Russell 1000 value is pretty much just a bet on your financial stocks and Russell 1000 growth is really just a bet on technology stocks, and there’s really no impact of value stock picking at all when you compare those indices, it is a pure sector bet. Which defeats the purpose of the value composition, so all of these things are sort of swirling around right, when are we making these decisions, how are we sort of smoothing the decisions over time, what signals are we using and then how do we think about this idea of unconstrained versus constrained?
Corey: Because constrained tends to do very well during calm market environments when you don’t have a sector or industry group that’s exhibiting bubble-like characteristics, but on the other hand, it’s great to be unconstrained in the dotcom era, so how do you sort of navigate all of this? So my view has sort of always been if I can’t confidently choose one way or another based on the evidence, diversify, so again if I don’t know whether its better to be constrained or unconstrained and if I don’t know whether it’s better to use any given signal versus another, I should probably use multiple, if I don’t know whether it makes sense to look at them cross-market, or within an industry group, do both. And so that’s ultimately the way a lot of our sort of systematic value strategy works.
Corey: So starting with the S&P 500 because that’s going to be our universe, just as an example here, what we’re going to do is we’re going to rank stocks on a number of value factors, so we’re going to look at things like price-to-book, price-to-earnings, price-to-free cash flow, enterprise value to EBITDA, a number of different factors that have demonstrated success and might touch on different parts of how you would define value, and we’re going to do that both across the entire market, as well as within each industry group and aggregate that up all into different value scores. Then what we’re going to do is once we have all of our value scores, we’re going to try to pick the deepest value stuff, so say the bottom 100 and then we’re going to try to pick them in a way that we end up getting to a place of more sector neutrality. So if I can choose in the bottom 100, a completely sector-neutral portfolio, great, but if for whatever reason there’s a given sector or industry that doesn’t show up at all, it’s not in the bottom 20% of the cheapest stuff, well then I can’t get to sector neutral and that’s sort of my model telling me, “Hey, this is a little bit of a bubble potentially in that sector.” So we’ll avoid it entirely.
Corey: Then once we’ve got sort of the stocks we’re picking, we narrow it down to 50 concentrated stocks, then the question is, “Well, how do we weight them?” We have our 50, we know they should sort of be industry neutral, should we equal weight, should we value tilt them, should we market cap weight them, should we run an optimization? Again, I think there’s ample evidence for all of it, so that’s what we do, we do all of it. We run a model that’s equal weight, we run a model that’s more of a mean-variance optimization, sort of a Sharpe optimization. We run one that’s a value tilt, we run one that’s even a quality tilt, we blend that all together. So there’s a lot of moving pieces, but the whole overarching philosophy here is when I don’t know, diversify, and it doesn’t necessarily dilute, again I’m not diluting how much value I have by using multiple value signals, what I’m trying to do is dilute my specification risk. I’m trying to stick with a style of value without necessarily making a hard choice, yes or no, about a given implementation.
Corey: So at the end of the day, at the end of each month what I have is a value portfolio, then the question becomes, “Well, how often do I rebalance this thing, and how willing am I to kick things out?” So, from my perspective, what I tend to find and I’m perfectly open to arguing this evidence, is that value is sort of a slow-moving phenomenon, that you can buy these value stocks and typically the premium doesn’t mature for three to five years. Now, you might get lucky and it might mature faster but it is a very slow moving decay in the alpha, and so what we say is, “Well, what we want to do is we want to buy that portfolio almost like it’s a PE traunch, and hold it for five years.” So we’re going to have it be one-sixtieth of the portfolio. Then the next month, what we’re going to do is we’re going to take the oldest traunch that we’ve held for five years, kick it out of the portfolio and buy a new traunch.
Corey: And we’re going to keep doing that, so the things that we buy, the deep value stuff has enough time to mature, now, just to make sure we don’t hold onto something way too long, every month we’ll also check, well did anything in the old traunches, were there corporate actions that changed the picture, did something all of a sudden become super overvalued that, “Hey, if matured early, let’s get rid of it.”? In those cases, we do get rid of it, and what therefore you tend to see is, you have this nice decay, that your early traunches tend to be a bigger part of the portfolio, and then over time your traunches get smaller and smaller and smaller as a proportion of capital, until eventually by the end they sort of, before you even kick them out, have sort of decayed their way out because things have had time to fully mature. So, again a lot of moving pieces, but the idea for us is, when we don’t know, let’s try to diversify, let’s try to make sure we’re not overly emphasizing one style, one recipe of construction, we’re not overly emphasizing sector concentration risk versus making sure we’re never taking risk.
Corey: We’re not emphasizing any particular rebalance month over another, all of it is trying to diversify wealth, therefore giving us a more pure implementation of that value style.
Tobias: And that’s sort of a feature of value, that it does take a much longer period of time to decay as you say, and I think I’ve seen some research that says that any given portfolio out to about five years, I think that it’s sort of like a rubber band, the bulk of the return comes out in the first few years, but you’re still getting some additional return all the way out to five years. Which is very distinct, very different from momentum which has a much shorter rebalancing period, and I’ve seen some interesting research from you which talks about the relationship between the look-back period for momentum, deciding whether something’s in or out, and then how long it’s held afterward.
Corey: Yeah, so this actually was not original research by me, I think that the piece I wrote was called Momentum’s Magic Number, it was inspired by a research piece I saw from HIMCO, and it candidly blew my mind. So typically there’s this concept that you would say that we think momentum tends to work from a sort of three to 12 months look back horizon, so what does that mean? That means, if I look at the prior 12 months, for example, and rank stocks based on those twelve months, the things that perform best should over the short term continue to perform best. If I rank them again, by six months, those that perform best should in the short term continue to perform best. And the typical intuition has been, your rebalance period should really be a function of how long you’re looking back. Again, acknowledging that this tends to be a very fast decaying signal, so if I’m going to look at 12 months, I might hold on for another month and say, “Okay, I’m going to look at 12 months of returns, I expect this rank ordering to be somewhat stable over the next month but after that, the decay happens really quickly.”
Corey: You might then say, “Hey, six months, I can’t hold on for another 12, I should only hold on for two weeks.” Sort of proportionately decreasing, because I’m using less stable information so I need to adapt more quickly. This HIMCO article completely turned that type of thinking on its head, and what it said was, “Actually, the evidence suggests that the whole thing, the formation period, how long you’re looking back, plus how long you hold just sort of kind of needs to add up to around 14 or 15 months.” So if you look at a three-month formation, which I just said, “Hey, for that to be stable, it’s probably only stable the next couple of weeks.” They’re saying, “No, turns out you could probably hold that for another nine to ten months, and it would be totally fine.” Which I just found to be totally counterintuitive, and they demonstrated, and I believe in that paper with large-cap stocks, so it said, “All right, I’m going to try it with some country indices, I’m going to try it with sector indices, does this really hold up?” And lo and behold it really did.
Corey: Which again, is mind-blowing to me and goes back to maybe the point we were making at the very beginning about backtesting, I have no idea why that works.
Tobias: I was about to say, what’s the driving force of that, that’s an unusual-
Corey: Yeah, it is very odd and I could not come up with a good economic rationale, I could not come up with a good behavioral rationale, it seemed to me like maybe what was happening was trends tend to persist over that period, 15 months was it. So if you bought at 12 months, it meant yeah you were more certain, but you had less time to hold on, versus if you bought at three months, well maybe you don’t want to turn over as quickly, there’s still some signal there because you’re not three of the potential 15, you would then want to say, “Okay, let me hold on for the remainder of the 12 to watch this play out.” That was sort of the best I could come up with but why do trends last 12 months? What behavioral or risk-based phenomenon is driving that? It might be the same thing that says, “Well I 12 months looked back in the first place.” But at the end of the day, trying to … it’s very counter-intuitive as to why you would look at a very short term look back and hold for a very long period and have that be successful.
Corey: So for me, it’s an open question and a very interesting one but I don’t like to build investment strategies for which I cannot come up with a rationale or a reason because to me, then it’s risking some sort of data-mined artifact that I’m being fooled by randomness.
Tobias: I was about to say, it’s a very Nassim Taleb kind of approach, and it’s almost a philosophical question, do you look at the empirical research to make your decisions, or do you formulate the idea and then test that idea. I think whichever way you do it, that you’ve got a risk of finding something that doesn’t work anymore and never really worked.
Corey: Yeah, I think it sounds wonderful and flowery and intuitive to say, “Hey, I want the economic rationale, why would I invest in something without the economic rationale?” But I think the counter to fooled by randomness is fooled by narrative, right? If I discover an anomaly and then come up with a rationale as to why that anomaly works, am I just fooling myself into believing that narrative? If I come up with narrative and then data-mine the facts to fit that narrative, am I any better off necessarily? So I think it is one of these things, as a quant we always try to say, “Hey we’re evidence-based.” The real truth is, the amount of evidence we have at our hands is very, very limited. Financial markets, there’s really just a few driving forces, we’ve only seen a few big regime shifts over the last 50 years, and arguably some of those regime shifts make prior data totally worthless. I mean, is anything prior to World War Two really relevant? The advancements we’ve seen in technology and access to information and our education about financial markets really raises the question of, is some data and some anomaly that works back in 1910, is that relevant?
Corey: We tell ourselves, yes and we’re looking for this robustness across countries and geographies and asset classes and history, and I do think that is a good start, but I think it’s always good to keep in the back of our mind, everyone’s looking at the same data, everyone’s looking at the same anomalies, yeah that does help with the robustness conclusion, and hopefully there is some economic or behavioral risk-based intuition to whatever we’re trying to harvest. But at the end of the day, it is a limited data set that we keep going back to and trying to mine, and we do run the risk of at a certain point, just data-mining it to death.
Tobias: One of the metrics that I find most fascinating is the price-to-book, because it’s the definition of value, the factor, and Fama and French have tracked it for a very long period of time, and the value investor in me says, “This is something that’s probably going to mean revert, it’s bound for some return.” But then there’s some very good research out there by our mutual friends, O’Shaughnessy’s, have looked at book-to-value that make a very compelling argument that the phenomenon of more companies having negative equity has changed the way that the metric analyzes companies, it’s no longer really finding that value effect, you have any thoughts on that?
Corey: Well so, I should caveat all this with, I am not a value investor at heart despite my dabblings and my own personal money. I think it is important to always connect these fundamental metrics back to what they’re trying to extract and tell you about the company. But I do think there’s a really interesting lesson here and I wrote about this, I think it was last summer in a piece called Factor Fimbulwinter, about this sort of corner we box ourselves into as quants. So, you look at something like price-to-book, and you say, “Okay there’s 100 years plus of U.S. evidence of price-to-book that gives us some sort of statistical significance that this is an anomaly and therefore we believe it’s going to continue. Even more so, we see that value as a concept works in other asset classes.” So we like this idea, this sort of theoretical concept of value investing, buying cheap, selling expensive. We find that price-to-book works when you look in other countries, we find that price-to-book works when you compare country indices.
Corey: So there’s all this confirming evidence that price-to-book works and the problem becomes, it’s going to be almost impossible to disconfirm it from a statistical perspective. So if I said to you, “Look, price-to-book has no more positive premium. You can buy cheap stuff, avoid the expensive stuff, all that’s going to happen is you’re going to have noise from here on out, there’s going to be no positive premium.” It would take a really, really long time.
Tobias: How long?
Corey: Well, it depends on your interpretation. So the piece that I wrote basically said, “Okay we’re going to use this sort of Bayesian updating, we’re going to take all of our prior confidence, and then every year we’re going to update our confidences to whether we think it continues to work.” And the way I did that was via simulation right because what happens when a factor breaks is it’s not that all of a sudden it stops working and you get negative results. When a factor breaks, you get random results, I think that’s missed on a lot of people, you can have price-to-book be broken and you could go on a ten-year positive streak. Doesn’t mean it’s not broken-
Tobias: Just through luck?
Corey: Just through luck. And I think that’s one of the things that makes it so hard to now disconfirm this stuff, so what I’ve found was, you get this distribution right of okay depending on how things play out, it could take a couple years or it could take 100 years depending on how luck unfolds, but the median amount of time I think for price-to-book that it took was around 60 to 70 years.
Tobias: That’s crazy, nobody’s got time for that.
Corey: Right, and all of us are-
Tobias: That’s more than a career.
Corey: … most of us will be dead, so yeah that’s longer than a career. And you saw it, and this isn’t some theoretical thing, this is something you saw with small caps, that when the small-cap premium was discovered, it has now taken 30 plus years to sort of statistically disconfirm it and it’s on the cusp, it’s not even disconfirmed yet. People are arguing, “Is it real, is it not?” But a lot of people would still say it’s real, there’s plenty that wouldn’t. But-
Tobias: And with good reason, the narratives pretty compelling, these are underfollowed stocks, everybody’s looking at a big universe, you can find these things that are undervalued, or it’s not a valuation question, but they’re just underfollowed and as they get bigger, they’ll attract more investment and [crosstalk 00:57:14]-
Corey: But they have less analysts tracking them, there’s lower liquidity so there’s lower competition for transparency, there’s a liquidity premium potentially. There’s a lot of great narrative around why small caps, there’s more opportunity there of why they might deserve a premium. I think you take that and you multiply it by ten for value, I mean the order of magnitude defending value I think is so much more severe than the narrative and evidence supporting small cap. So I think the problem is, this is where quant sort of put themselves in a corner, which is we have ultimately too much evidence supporting price-to-book that I cannot disconfirm it from a quantitative perspective. That I now, all of a sudden, need to go back to a fundamental argument as to why it’s broken, and a theoretical argument, which is fine, but it is this give and take. It’s sort of the science and art coming together that makes this so difficult.
Tobias: So just to change pace a little bit, my favorite page from your deck is, it’s got some of the things that I’ve heard you say, some of the things that I’ve seen you tweet out, I think you’ve got one of these pinned to the very top of your Twitter profile, and you’re a great follow on Twitter, we’ll get your details out in a moment, but, “No pain, no premium.”
Corey: Yeah, yeah I actually just wrote a piece this week on no pain, no premium.
Tobias: I saw it.
Corey: So this is one of those, it goes back to the old, “No pain, no gain.” It’s just a play on that, but at the end of the day it goes back to this risk-based mindset we were talking about at the beginning, which is a lot of people are `pursuing returns, and I think often forget that the null hypothesis is that expected return comes from bearing some sort of risk. And whenever you’re evaluating a strategy or a trade, or anything, you should be asking, “What is the risk that I’m bearing or conceptually bearing, that someone else wants to get rid of this, I’m willing to buy it, am I being paid enough for that risk if that risk occurs?” And I think all too often, we get caught up in the pursuit of return and the potential alpha something generates, but I think when you look at things through a risk-based lens, you really start to be able to build potentially better portfolios by looking at, “Okay, what risks am I ensuring when I buy equities, what risks am I ensuring when I buy bonds, commodities, what risk am I ensuring when I buy a carry strategy, when I do trend following, and how do these all interplay with each other to create a more consistent return profile?”
Corey: I do think it ultimately, to me, the most interesting philosophical aspect of the “No pain, no premium.” Mentality is that diversification taken to its extreme limit, there still has to be pain left over, if you diversify away all the pain, all you should expect to earn is the painless return, which is the risk free rate, your risk free, U.S. short term government bonds, if you think U.S. government bonds are risk free. So there’s some limits here, theoretical, philosophical limits as to how much pain you can ultimately get rid of, and as tough as it is to say, “Hey, I’m going to subject my portfolio to day-to-day volatility and once in every while there’s going to be some shock.” That is the pain that allows you to harvest the long term return premium, and I think it’s really important to keep that in the back of our minds.
Tobias: That’s great, and that’s the full hour. Corey, if somebody wants to get in touch with you or they want to read some of your research, follow you on Twitter, we’ll have those details in the show notes, but do you just want to let everybody know now, Twitter handle for example?
Corey: Yeah, Twitter handle is Choffstein, so first letter, last name. It’s about 80% finance stuff, 20% coffee and working out and snowboarding and random family stuff. You can also find me, so I publish a weekly research commentary on our blog, which is blog.thinknewfound.com, you can go there and subscribe, most of what we talk about is not what’s going on in the markets, it’s not what’s going on in the economy, it’s more let’s do some deep dives, some quantitative analysis on these different style premia, portfolio construction, risk management, craftsmanship topics, things that are a little more evergreen. It can get pretty wonky I’ll admit, but if you hang out long enough, I’ll hopefully try to distill it down and circle the drain on a lot of these topics from a lot of different angles. And then you can find my personal email on that website as well if you want to reach out.
Tobias: That’s fantastic, thanks very much Corey, really appreciate the time and the thoughtful ideas.
Corey: Thank you for having me, Toby, it’s been fun.
For all the latest news and podcasts, join our free newsletter here.
Don’t forget to check out our FREE Large Cap 1000 – Stock Screener, here at The Acquirer’s Multiple: