Acadian Interview with Vlad Zdorovtsov

What is your research philosophy?

I firmly believe that although markets are reasonably efficient, it is precisely due to professional active management keeping them so. Quantitative research plays a central role in this process by detecting and rigorously vetting persistent sources of excess returns. Fundamentally, returns are predictable for three reasons:

1. Market participants are emotional and irrational: They succumb to numerous behavioral biases which lead to pricing errors. To me, these biases are akin to optical illusions. Our brains have evolved to take mental shortcuts which generally served us well over millennia, but which can often lead to systematic mistakes with predictable effects. Importantly, investors continue making these mistakes even after becoming aware of them - just as happens with optical illusions we cannot help but see. A key implication here is that by and large we expect to see the same behavioral anomalies and concomitant mispricings manifest pervasively across the global universe.
2. The boundary between risk and alpha is skill dependent: There are systematic drivers of the marginal investor’s portfolio volatility that he or she effectively pays a premium to avoid. The term risk premia is, in my opinion, a bit of a misnomer when it comes to these to the extent they do not necessarily behave as theory would imply risk factors should - e.g. some may at times have positive realizations in bad states of the world. More importantly, what is risk to one investor may not be to a savvier one who is able to model and manage that specific source of portfolio uncertainty, effectively rendering some or all of the associated premium alpha.
3. Market frictions create distortions: Frictions are the proverbial speedbumps on the road of price discovery. These are often the results of man-made rules with temporarily distorting effects on prices (e.g. short selling constraints, lumpy index reconstitutions etc.). We expect to see some regional variation here insofar as these do have regional flavors (e.g. franking credits in Australia or shorting bans in some locations).

Key to being able to separate the wheat from the chaff in unearthing consistent sources of active return is a delicate and evolving balance of economic intuition and empirical evidence. In my opinion, it is striking this balance which separates the best active managers from the rest. Acadian treats this balance very thoughtfully, embracing complexity where appropriate.

  How is Acadian thinking about value in this market environment?

Value has been and will continue to be a significant part of our multi-factor process. As long as markets remain at least partially inefficient, there will be some mispriced securities that are cheaper or more expensive than their fundamentals warrant. The ongoing headwinds experienced by value signals stem from several broad areas:

• Growth & value traps: The expensive stocks that are typically overpriced due to irrationally extrapolated past earnings and subsequently proceed to disappoint have, in a sufficient number of cases, continued to show healthy earnings growth while cheap stocks disappointed. 
• Exogenous crosswinds: The environment of declining interest rates has given a tailwind to growth stocks, which have seen their bond betas rise post GFC.
• Evolving definitions of value: As business models change, the role played by intangibles has continued to grow, altering the very definition of valuation.
• The conventional groups within which stocks are compared have grown more heterogeneous: The standard groups within which one could have made cross-sectional comparisons based on value metrics in the past, have in some cases become significantly less homogeneous, necessitating a more tailored approach.

ESG continues to be of increasing importance to investors. What perspectives do you have on ESG-oriented research?

My view is that E, S, and G tap into the same underlying latent driver – management shortsightedness along two dimensions: the cross-sectional dimension of different stakeholders whose welfare will directly or indirectly pertain to shareholder outcomes and the temporal dimension of whether a given management decision myopically hurts the future for the sake of the near term outcomes. We have a significant number of ESG-related components in our model spanning different facets of this construct. This is an active area of research and there are currently a number of projects underway in various stages. ESG-related signals are held to the same rigorous standards as any other alpha component.

With all the buzz about alternative data, are these sources rapidly crowded and arbitraged away or will the alpha have some longevity?

Active managers have always looked for a competitive edge in new data. Information is the most valuable commodity, as Gordon Gekko put it in Wall Street. The recent proliferation of alternative data is best seen as part of that evolution that has picked up pace as the economy has effectively grown a nervous system, able to sense and retain information on a myriad new metrics containing rich insights on company fundamentals and the behavior of consumers. Importantly, although one might assume that with more data available to market participants, their views will converge to be more accurate, academic evidence suggests otherwise – the more information is thrown at people, the more they disagree about how to interpret it and dig in on their views due to confirmation bias. Furthermore, a number of these new datasets are not yet friendly to systematic managers due to limited histories. Their primary users currently are discretionary fundamental managers – who will often succumb to biases while interpreting that information and distort prices creating opportunities for quants.

What role do you think machine learning should play in quantitative investing?

ML is an area that I feel is still relatively untapped in our industry and one I believe holds significant promise. We know the real world operates in highly non-linear ways with a host of nuanced interaction effects. Teasing out these complex relationships with sufficient confidence takes the right methodology, a significant investment into the requisite tools, and, importantly, an openness to and appetite for complexity. The team has made significant inroads into ML over the years and we continue to invest into this area. There was already strong machine learning expertise on the team when I joined, and we have since further added to this in both deep and reinforcement learning.