Is this another case of a product based on incomplete research?
ETFs are branching out into ever more obscure parts of financial markets to the extent that it sometimes feels as if ETF picking has become the stock-picking of the 21st century. The problem with so many of these approaches is that they don’t deliver in real life what academic studies promise.
I have written about smart beta ETFs as an example of a type of product that performs much worse in practice than in academic studies here. In practice, many financial products fail to translate backtests into actual performance, but ETFs seem to be particularly prone to that because they take simple rules developed in an academic paper and then apply them to real-life markets.
Don’t get me wrong. I love ETFs. I think a plain vanilla ETF reproducing a liquid and well-diversified index like the S&P 500 is a great investment product. The problem is that such plain vanilla products don’t command particularly high fees. So, the financial industry is constantly looking for an ‘edge’ to justify charging higher fees. After all, we all need to make a living, don’t we?
Traditionally, the edge was active management where skilled fund managers could command higher fees and loyal investors. But as the reputation of active management has suffered due to many fund managers not delivering what investors expected, the financial industry has started to look for other products. ETFs with more and more complex investment processes seem to fit the bill just fine. The great thing about these products is that they tend to be based on academic and peer-reviewed research which gives them the appearance of scientific rigour that a stock picker fund manager never can provide.
Unfortunately, though, the academic literature is full of results that are subject to selection and survivorship bias. Journals tend to publish only positive results which means that if you have an effect that provides random outcomes, but 20 different academics analyse this effect, chances are one of them will find a result that is statistically significantly different from zero at the 5% level. This result will be published while the other 19 never see the light of day. Et voilà, a new effect is born that can be replicated in an ETF.
The other problem is that many product providers try to outsmart academic research and add different twists to their ETF that seem to improve performance in a backtest but reduce returns once the product is launched.
There are a host of other problems and I think the new Nightshares ETFs may have such an issue. These ETFs try to exploit the difference in returns for US stock markets between night and day. There are several papers out there that show that the return of the US stock market from close to the next day’s open is less volatile and in many cases higher than the return of the same stocks from the day’s open to the day’s close. I have written about this effect before and for the US it shows quite a substantial outperformance of nighttime returns over daytime returns.
But note how all the research I have written about was based on US stocks. As I have written here, one of the key things you can do to improve your performance is to test the same methodology for other markets that have different dynamics and different performance histories. So, I calculated the daytime and nighttime returns of the US, UK, and Europe ex UK stock markets over the last 10 years. And as you can see below, while the nighttime returns for the S&P 500 look quite good, the nighttime returns for the European and UK markets are not only quite a bit lower but also roughly the same as the daytime returns.
Now, it may well be that the US market continues to provide exceptionally high nighttime returns and that these new ETFs perform well. I for one wish them luck because if they do while Europe and the UK continue to show mediocre nighttime returns then we have a research puzzle at our hands. And an unexplained puzzle is much more fascinating than yet another effect that disappears after a while.
Daytime and nighttime performance of different stock markets
Source: Liberum, Bloomberg