In a recent paper researchers from the Federal Reserve and Penn State examine 204 stock market anomalies, which often underpin ‘smart beta’ investment approaches and other quantitative investment techniques. They are not impressed. After transaction costs and other considerations, they argue that, in aggregate, these strategies have not offered anything like the returns many investors expect over the most recent 2006-2020 period. In fact, as a group these strategies have fail to beat the market after costs in recent years.
Dissecting Market Strategies
These are trading strategies such as buying inexpensive stocks (value) or buying stocks that have risen strongly in price over the past year (momentum). Of course, given there are 204 strategies included, so are many different variants of value and momentum techniques as well as many other anomalies that have received academic attention in the past.
They find the returns from these strategies to be uninspiring. There are a few reasons for this. The first is that trading costs eat into returns, factors like bid/ask spreads and price impact can reduce the profitability of a strategy that looks good on paper. In fact, they argue that given trading costs, even the results claimed when some anomalies were published were actually eaten by trading costs, though they argues trading strategy mitigations may help here.
Secondly, after discovery many strategies do less well. This could be due to data mining, or maybe the strategy does work, but enough people copy it to reduce its returns.
Poor Average Performance
Finally, in aggregate these strategies do less well, it’s easy to point to a few strategies that do indeed beat the market, but it’s hard to know that in advance, and when you look at the 204 strategies together, results are meagre.
The paper which can be found here is titled ‘Zeroing in on the Expected Returns of Anomalies’ by Andrew Chen and Mihail Velikov.
The academic debate on market anomalies is a long one. This is thoughtful research brining a useful quantitative angle across a large number of anomalies. The main point of contention is grouping all 204 anomalies together and treating them as a ‘take it or leave it’ strategy. The reason for this, the authors argue, is that its hard to pick winning strategies in advance.
However, there own research does seem to show that strategies with less than lower turnover can beat the market by about 0.04% per month, or almost 0.5% a year. Again that may be statistical noise, but it could show that there still are benefits to certain strategies when trading costs are considered. Also, many strategies, such as value have seen poor performance over the recent decade, and some believe they may rebound.
However, it is interesting to observe that since the discovery of many anomalies, trading costs, at least in terms of bid/ask spreads have fallen massively. The authors’ estimate that bid/ask spreads which may have been 5% for the 75th percentile of stocks in the 1990s are down to under 1% today. That may be one reason why anomalies face a challenging time as implementation costs have sharply declined. Though perhaps they were never a money-maker in the first place.