Many quant managers are having a tough year. While one might blame factors in general, their returns do not tell the whole story (or even the bulk of the story). We think one of the major culprits in the US is that a number of factors worked better on the short side and among small-cap names, but even that does not explain all of the unexplained. We highlighted our analysis in a recent paper, which can be found here.
Factor returns have varied widely by region this year. For example, both Profitability Value fell far short of expectations in the US, but fared well in the UK and Europe. With all its ups and downs, the year-to-date return for Medium-Term Momentum is just about zero, with strong performances in Canada and Australia, but a highly negative return in Japan.
In general, however, returns were not enough to explain significant underperformance in our multi-factor “sample factor portfolios,” which we use as a proxy for long-only US large-cap quant manager performance. Whether tightly constrained or not, these portfolios have consistently and significantly lagged the market all year, with active returns between -6% and -8% (on target tracking error of 3% and realized tracking error close to target). Performance was much worse than that of the underlying factors contained in the alpha and also far lower than similar portfolios constructed to target a single factor. For the period ending October 2019, the 12-month drawdowns were the worst or close to it since the portfolios’ inception in 1990. Two-year drawdowns were even uglier.
Sample Multi-Factor Portfolio and Performance of the Underlying Factors
We calculated 60-day rolling factor correlations and found that most factors were either much less correlated than average or had average correlations. Only two factor pairs were much more correlated recently: Momentum and Low Volatility, and Earnings Yield and Value. These higher correlations were offset by lower correlations for other factor pairs (for factors used in deriving alpha), so we do not attribute the sharp underperformance to a sudden jump in factor correlations.
To help us determine why performance faltered in the absence of simultaneous factor failure or an increase in return correlations, we turned to our US “Alternative Factor” portfolios. These portfolios are built to be as much as possible like our “factor-mimicking portfolios (FMPs),” but provide insight into the impact of typical manager constraints, notably the long-only and limited-investment universe constraints. We found that a number of factors fared far worse when a no-shorting (aka long-only) constraint was imposed on the portfolio, suggesting the factors had better performance on the short side. Some also had much better performance across the broad universe of stocks than they did when limited to a large-cap universe. These differences may help explain more of the severe underperformance of the portfolio relative to its underlying factors.
Cumulative Return of “Alternative Factor” Portfolios
Finally, a few other potential culprits, including correlation with economic variables, an increase in specific risk vs. factor risk, concentration of the poor returns in a few names, and significant changes in factor returns versus their long-term history—none of which explained more of the underperformance.
We therefore concluded that at least a portion of the negative active returns could be traced to 1) unusually poor performance in a few factors, particularly Value, 2) a number of factors faring better on the short side, thereby cutting off much of the return possibility for managers unable to short, and 3) other limitations, such as being confined to invest in a large-cap universe, may have had some impact as well, although it actually helped for some factors.