Here is an interesting turn in the active management, alternatives, factor-investing space: Cliff Asness, of AQR fame, in an article speaks to a point we have made repeatedly in this blog over the past several years: there are no such things as exploitable "factors" (smart or dumb) that are identifiable a priori, sustainable in liquid markets using simplified rule sets and generate alpha.
In our opinion, it's fools' gold but many billions are invested in factor investing presumably to generate some form of alpha. Note that we called out AQR a couple years ago, after analyzing one of their then newest product offerings (QSPIX), concluding that it is offering up over-engineered, re-worked factor ideas and at monstrous prices that are unlikely to deliver net alpha to a representative benchmark. If the outflows from AQR now are indicative of some semblance of investment common sense, that's great.
Now, Blackrock is jumping in with a new ETF offering timing of allocations into various factors (trading symbol DYNF) for .30%! More fool's gold? Let's look at this one year from now.
Which brings us to an actionable point for our readers. We surmise that the nature of current active techniques housed in an ETF or mutual fund wrapper is a general harbinger of inefficiency due to diverse holdings/high security count and in-effective use of underlying liquidity. Separately, there are ETFs that are highly concentrated, eg USO, GLD and others that do not apply.
Let's ran a real live example to help suss out the point.
We use AADR, an AdvisorShares ETF that uses Dorsey Wright's technical analysis-based process to invest in 30+ American Depository Receipts (ADR) and is priced at .88% annually. It is in both our iAWR core portfolio and in our ARBiALT product. So, AADR carries all the potential inefficiencies as noted above.
Our test ran the ARBi learning process on AADR's core constituents and compared the ARBi-based performance to that of AADR. Note that our process involves potentially daily rebalancing but is largely 96% invested (when used on a long-only basis). So, ARBi performance would be gauged two ways: 1) on AADR itself, and 2) on the constituents of AADR. Results 2018/2019 (rounded):
AADR -42% / +14%
ARBi on AADR -9% / -2%
ARBi on AADR constituents +9% / +20%
Although our ARBi process can generate alpha on the AADR itself, what's more salient to our point is how much more it generates when running on the constituents. So, if one is interested in alpha generation, then the real takeaways here are:
Be wary of any basket of securities lumped together that you can trade yourself (mostly US equities)
Cost is not the primary determinant of whether an investment process is of value
Accessing the volatility of the underlying stocks is key for alpha generation
Regardless of the strength of an investment process, ARBi notwithstanding, it will rely on security-level movement for alpha generation. An ETF by general design damps security movement so making alpha on the ETF is a tougher long-term proposition. And, logically, if the ETF is net inefficient then where is the efficiency going? That's right, it's funneled to those market participants that are using the liquidity inherent to the same positions held by the ETF and have an adequate investment process to capture such efficiency. For this reason, a cheap ETF is not a free long-term ride - its actual cost is much greater to the investor than they think but you won't hear that from its purveyors.
So what's the connection between AQR, Blackrock and this AADR test? Simply that systematic investing, stocks, factor/trading rule sets, large baskets applied against a $35T marketplace, unfortunately, appears to be more fool's gold than real edible alpha.