One of the consequences of the ever-cleverer advances being made in computer technology is that academics and other researchers and are able to identify ever-greater numbers of anomalies by back-testing ever-larger amounts of data. It is a point we made, here on The Value Perspective, in our ‘Jellybean Trilogy’ series, which, in an anomaly all of its own, happens to run from Part 1 to Part 4.
These articles also highlight how, once the laws of probability are taken into account and the proper adjustments made, all too often these anomalies disappear. Still, there are anomalies that do turn out to have statistical significance and some are significant enough that trading strategies are built around them so the number-crunchers should carry on merrily crunching their numbers, right?
Well, the short answer is – it depends. Any investor thinking about using a trading strategy based on a supposed anomaly drawn from a back-tested series of data needs to ask themselves one extra question – does the trading strategy work because it is taking advantage of a genuine structural anomaly or does the anomaly exist simply by virtue of increasing valuations? If it is the latter, buyer beware.
As the name of this article implies, it is a companion piece to Blind spot 1, where we warned investors not to underestimate the significance of a change in valuation. We also promised to explain why we hold that view and it is because, if the only reason an anomaly that is revealed through extensive back-testing leads to outperformance is because valuation has increased, you have three problems.
First, the reason you are using the trading strategy is because of past returns and yet, if those past returns have resulted merely from an increase in valuation, they are highly unlikely to be repeated. Second, even if valuation now holds steady, you are not going to see those historic returns that have already been enjoyed by somebody else.
Finally, of course, there is a very real risk that valuation does not hold steady but fades back and all the returns simply evaporate overnight. In the end, because investors are human beings and tend only to buy into ideas that have worked in the past, they generally make no returns from such strategies and, over time, the supposed anomaly on which everything was based disappears.
Now take a look at the following chart, which comes from an article by Research Affiliates’ Rob Arnott – whom we also met in Blind spot 1 – called Can ‘smart beta’ get you in trouble? It shows the recent 10-year performance of four common equity ‘factors’ – that is, potential sources of excess return – that are popular foundations for modern smart beta investment strategies.
Source: Mauldin economics 1 Jan 2005 - 31 Dec 2015
“Investors with a short memory (or following an asset manager with a short back-test) would conclude that value investing has become a fool’s errand and that investing in the most profitable companies is the way to beat the market,” writes Arnott about the blue ‘annual return’ bars before directing our attention to the orange ones, which show the 10-year returns that result from changing valuation.
These reveal that, of the four strategies, low beta can attribute a portion of its success to rising valuation while gross profitability actually owes all of its success to underlying assets becoming more expensive. Somewhat ironically, the one factor that has not enjoyed much of the recent valuation tailwind is value but, of course, that does means it is now trading at a big discount relative to pretty much everything else – growth, the market, low beta and so on.
You can see something similar in the next chart, which shows the performance of the same four factors but over a much longer timeframe. Between 1967 and 2015, observes Arnott, “gross profitability had much more modest returns, all of which came over the last decade, and all of which can be attributed to rising valuations”.
Source: Mauldin economics 1 Jan 1967 – 31 Dec 2015
It is the same story for the low-beta – also known as low-volatility – strategies that, in articles such as Sleep loss and Left smarting, have been the object of some concern, here on The Value Perspective – both on account of the huge amounts of money flowing their way in recent years and, given the attendant rise in valuations, how this is likely to play out for their investors.
Arnott takes a similar view, arguing the popularity of these strategies “brought about massive inflows and expensive current valuations. We cannot know whether the rising valuations created the demand for these strategies, or rising popularity created the higher valuations. Either way, new investors in these strategies may be in for a rude surprise.”
Conversely – and precisely because of its poor showing over the last decade – investors can nurture higher hopes for value’s prospects in the years ahead. Indeed, as Arnott notes in another article, How can ‘smart beta’ go horribly wrong?, value is now cheaper than at any time other than the height of the ‘Nifty Fifty’ in 1972/73, the tech bubble of the late 1990s and the 2008/09 global financial crisis.
Still, this potentially happy outlook for value is somewhat incidental – at least to this article. Our main point is that, if someone highlights outperformance stemming from any kind of back-test or anomaly, be sure to question how much of it is down to an increase in valuation. And if the answer – as it is with gross profitability and low beta – is pretty much all of it, you might want to find a new strategy.