The following is a guest article written by Estimize Founder and CEO Leigh Drogen. This is the second of a three-part series examining the rise of quantitative investing and how it is transforming discretionary asset management. The first article in the series can be found here.
We are in the beginning stages of a massive transformation taking place within the discretionary institutional management industry. Discretionary asset managers will need to either adapt or die. In this installment, let’s look more deeply at the forces that are creating this huge dislocation in the industry: the rise of systematic investing, the new Quant arms race and how the most successful discretionary managers are beginning to respond.
- Assets Are Flowing From Discretionary to Systematic
You don’t have to look too deeply to see this massive trend in strategy allocations playing out. Within Millennium, we’ve seen its affiliate WorldQuant blow the doors off the barn with returns and inflows of capital. Within Point72 (SAC) we’ve seen Cubist outpace the discretionary side of the firm by a wide margin with now over 40 systematic PMs. Balyasny has quickly shifted focus and is building a stable of systematic managers to effectively do something with their huge AUM growth. Other multi-manager platforms like Schonfeld, Paloma, AHL, Engineer’s Gate and GSA have added significant assets. Paul Tudor Jones is attempting to remake his firm by hiring a bunch of systematic managers, and others are following suit. And let’s not even get started with the continued dominance of firms like Renaissance, AQR and Two Sigma, where you probably can’t even give them your money if you tried.
I would say that the nerds are the new kings of Wall Street (Midtown), but frankly they (myself included) would cringe given their propensity to run in very different circles than the rest of the money manager crowd. This group is mostly made up of unassuming nerdy PhD types that you would probably take for accountants on the subway. They have serious mathematical and scientific training and have usually honed their craft on other data sets before coming to the financial world.
The fact of the matter is that there’s simply more efficacy to what these managers are doing than the vast majority of the discretionary trading world, and they’ve (mostly) put up the numbers to prove it. And I’m not just talking about returns, these groups are producing real alpha. Their strategies are meticulously back tested in and out of sample before going live, and are scaled up over time. In contrast, many discretionary managers launch a book with $500M in play from day one, I can count on one hand the number of systematic funds that have done that in the past 5 years.
And while some systematic funds don’t perform well, you’ll be hard pressed to find any massive blow ups akin to what’s regularly seen on the discretionary side. Pension funds can certainly deal with paying 2 and 20 if they have more confidence that their returns from year 1 through 3 aren’t going to all disappear in year 4.
The flow of capital from discretionary to systematic strategies is going to continue, as it should. That will have its own repercussions, which we’re already starting to see.
- Quants Dig For New Alpha
A 2012 tell-all book from a former Goldman Sachs trader revealed how the Great Vampire Squid often endearingly referred to their unsophisticated clients as “muppets.” While they rightfully got skewered for that comparison, they were certainly onto something when their trading desks were basically taking candy from babies.
However, many of the muppets are gone now and that’s left far less alpha in the market to capture. Relative value and statistical arbitrage strategies rely on capturing asset mis-pricings associated with the irrational behaviors of fear and greed. The muppets aren’t coming back, they’ve wised up. Less alpha overall will lead to a drop in the number of hedge funds and the amount of hedge fund assets that can generate enough alpha to command high fees.
It truly is amazing to watch a data set go from being an alpha to a beta over time. I’ve seen the sell side analyst estimates data set owned by Thomson Reuters IBES travel this path over the past 15 years. Yes, there will always be alpha which is associated with the irrational behavior of humans in markets available to be arbitraged, but most alpha generated by systematic traders is associated with an informational advantage.
About five years ago many of the classic stat-arb strategies stopped working due to an influx of competitors. There simply wasn’t enough alpha to go around. This precipitated the smartest firms to search for new data sets with predictive power, or reflexivity. Fast-forward a few years and an all out arms race is now under way.
I love to use the example of the company that is selling data captured from new car insurance registrations. They get this data daily, and it’s incredibly accurate at calling new car sales. So instead of waiting until the end of the quarter to find out how many vehicles GM sold, you can basically get a running count of growth on a daily basis. Obviously that’s going to give you an advantage in trading those auto names — until everyone else is using that data. At that point, the data set goes from providing alpha you can capture, to a data set that you must be looking at in order to avoid an informational disadvantage. In a sense, it becomes beta.
So the arms race is in full swing, and there is now a serious lack of qualified talent to analyze all of these different data sets and incorporate them into the existing multi-factor models. While the quantitative research process into the efficacy of a data set hasn’t changed much, firms are struggling to build a process around the testing pipeline. The most efficient firms like WorldQuant have been able to take advantage of that competency to move quickly and decisively to incorporate new alphas.
This brings me to my last point about the systematic testing process. In the next section of this article, I’m going to heavily malign the discretionary buy side for being fairly clueless about how to undertake this entire process. The truth is, even many systematic quants suffer from a severe lack of creativity and original thought when it comes to generating hypotheses around how to take advantage of a given data set. From our experience working with discretionary firms at Estimize, they are two steps even further behind the quants as it relates to incorporating new data sets.
Let’s just go back to the car sales example for a second. Would you know exactly how to take advantage of that data to run an event study and generate alpha? Probably not. You’d likely want to talk with someone who’s been trading autos for 10+ years to get their take on what they think moves auto stocks and how having a good projection of sales would impact those names. A good quantitative research process requires an ex-ante hypothesis for some level of causation and not just correlation. We need to know roughly why something works, not just that it works, or else we won’t know why it stops working, and as history has proven, everything stops working at some point.
Being able to hand over an easily testable clean data set, and a bunch of original thoughts about how to generate alpha is imperative for data firms to succeed at this process.
- Quantamental, Systamental, Factor Aware…Call It What You Want
The rise of the systematic quants and their use of these new data sets contributed to the poor returns of the discretionary world over recent years. First, the HFT guys killed the day traders making it impossible to pick up pennies. Next, the stat-arb guys crushed the swing traders playing in the couple of hours to one week timeframe. Were they the primary factor of poor discretionary returns? Probably not, but significant none of the less.
A few years ago the first big discretionary firms started making attempts to hire data scientists and acquire new data sources. They’ve mostly failed to integrate any of this into an actual investment process. Then about 6-9 months ago some of the more forward thinking discretionary firms gave in to the realization that they needed to make big changes. It’s not as if discretionary PMs weren’t using data driven statistical approaches to gain an edge, or that none of them had quants on the desk to help, they were just very few and far between.
You may have seen Paul Tudor Jones berating his organization in a strange showing of frustration from such a legendary investor. Steve Cohen has been very public about his attempt to shift Point72 in the data driven direction, even commenting that it’s incredibly hard to find good talent these days (we’ll get to this in a minute). The guys who have been successful in this game historically see the writing on the wall. Hell, even the first episode of season two for the show Billions features main character Bobby “Axe” Axelrod giving his team the condensed 3 minute version of this piece, albeit in a much louder tone. So whomever the producers of that show are talking to, this whole thing has seeped into the mainstream buy-side consciousness now.
The shift that needs to happen is similar to the way players were drafted in Michael Lewis’ book, “Moneyball”. Consider how hard the scouts fought against being replaced by algorithms that were far more accurate than they were, and even in the face of all this evidence, refusing to change. Then consider how much money was on the line in baseball, and the astronomically larger amount in our world. You would think that would precipitate a much quicker shift, but in fact, it will only mean a slower one due to the fear of change when dealing with so much money.
As quants, we are taught how to go through the research process to validate the efficacy of a data set or tool. Everything is derived from this process, and there isn’t too much leeway because it is designed as good science. Yes, as mentioned above, you still need a level of creativity in order to do good research. However, discretionary managers don’t even have the framework for understanding how to do that research, or incorporate new things into their decision making process. This is the largest hurdle to making the shift, and I believe less than 20% of managers will clear it.
This shift isn’t just about using new data sets, like Estimize or the car sales example, it’s about fundamentally buying into the notion that PMs need to be putting the odds in their favor by making investment decisions based on statistics, and not just being gunslingers or bottoms up value guys. That’s an affront to their entire way of doing things, just as it was for the baseball scouts.
In the next installment, we look at the people side of the equation. With the right discipline, savvy discretionary PMs should be able to crush the systematic quants. But that will mean a radical change in the types of people you hire and how you structure your teams.
On June 20th, Estimize will be hosting the L2Q (Learn to Quant) Conference, a one day seminar designed for discretionary institutional PMs, analysts, and traders who know they need to move quickly and efficiently towards building quantitative processes. Segments will be taught by preeminent buy side, sell side, and unique data experts with vast quantitative investment experience at Two Sigma, PDT Partners, WorldQuant, Wolfe Research, Deutsche Bank, and others.