Risk and Portfolio Theory

Corey Hoffstein’s “Why quants don’t pick stocks” (https://blog.thinknewfound.com/2017/04/quants-dont-pick-stocks/) got us thinking a bit more about risk.  Following on our “Island Economies and Risk” piece from a few months back, we wanted to expand on the discussion about risk as it relates to stock-picking and portfolio construction.

As we’ve discussed, efficient market academia suggests that Mr. Market actually only cares about systematic risk. In this world, idiosyncratic risks are diversified away in the aggregate, and thus stocks are priced as if they don’t have firm‐specific risk.  But what if firm‐specific risks really do matter?

The Argument

An equity long‐short fund, for example, might be short a basket of favourite sell ideas, and hedge the systematic risks away (with, say a long index futures position against the market risk, or other derivatives against other factor risks).  The sell thesis on each of the individual shorts probably makes the assumption that some firm‐specific event will force investors to realize that the security is not worth as much as the market currently thinks (or, similarly, that the security is more risky than the market currently thinks).

The long index hedges eliminate all the ex-ante systematic risks, so factors such as market direction (or momentum or quality or anything) do not impact the performance of this hedge fund portfolio. If the hedge fund manager is correct in her analysis (and not just lucky), then she will generate positive returns regardless of market (or factor) direction, thus those returns won’t come from a factor bet, but from idiosyncratic risk.  

Are those returns not alpha?

Alternatively, could it be that the consensus investor really does care about firm‐specific risks, these risks are already priced into stocks, and that we really are paid to assume them? In this case, imagine a hedge fund manager who likes a deep-value conglomerate from the long side, but is uneasy about two of its underperforming divisions. In this case, the manager might buy the conglomerate, and hedge the position with shorts in highly correlated companies in the same industries as the unwanted divisions of the conglomerate.

The manager is left owning risk of the businesses she likes, with little exposure to the industry exposures she doesn’t like. If so, is this ability to eliminate firm‐specific risk and earn the idiosyncratic risk compensation worth something – especially if the return is uncorrelated to returns from other asset classes?

The Counterargument

In either case above, however, the argument could turn on itself. If there were only one hedge fund out there with the ability to hedge whatever risks the market really cares about (and it had limited capital), then perhaps it would be too small to impact the market and could continue to generate alpha by selling overpriced risks and buying under-priced risks. But there isn’t just one.  

‍The fact is that there are hundreds of well‐capitalized hedge funds out there, and perhaps even more capital in factor portfolios mimicking what used to be called alpha, but now is called smart beta.  Perhaps these investors in aggregate may now be the marginal buyer and seller of stocks. Therefore, the risk that hedge fund managers see in a stock should be the same risk that the market sees in aggregate – and thus there should be no asymmetric risk perception to exploit (aka “no alpha to capture”).

Our Conclusion (Admittedly Biased)

We might be hopeful here, but our guess is that there is still such a significant amount of assets tied up in traditional active pension funds, mutual funds, and other institutional diversified vehicles – that the concentrated manager can still exploit her ability to capture firm‐specific risk.  Behavioural biases abound, en masse, and these can’t go away as long as there are human marginal buyers and sellers.

The next question, which we have already started discussing a bit in “Robots and Alpha” is whether or not - and at what level - systematic active (aka smart beta) and passive strategies start to impact alpha opportunities, either by destroying them, or creating them.  In other words, can dumb robots overreact?  

Again we aren’t terribly objective here, but – from what we have seen in individual instances – we think they can.