Assessing smart beta strategies

Analysing smart beta performance and risks is not monkey business. For a better understanding of smart beta strategies it is crucial to analyse their construction principles, performance characteristics and risk-factor exposures, including not only value and small-cap factors, but also a variety of other well-documented risk factors such as momentum, profitability, investment, low risk, and possibly others, writes Felix Goltz, head of applied research at the EDHEC-Risk Institute.


In the marketing of smart beta strategies, index providers focus primarily on the ability of these strategies to deliver outperformance over the cap-weighted benchmark. The issue of risk exposure of these indices and performance attribution to well-defined risk factors is rarely addressed by index providers. The existence of so many smart beta strategies coupled with so little information on their sources of performance poses a risk of confusion and overgeneralisations.

There have been claims that smart beta results in value and size tilts, irrespective of the weighting method chosen, that different smart betas produce similar premia, that the added value is the same when the strategy is inverted, and that the performance of smart beta is similar to that of randomly generated portfolios, which are also referred to as “monkey portfolios”. Since the idea that smart beta strategies are as good as random deviations from cap-weighting is at the heart of all the claims above, we collectively refer to these as the “monkey portfolio” argument.


Research refutes argument

In recent research, we reported results to a series of straightforward tests of these claims. In order to test the various claims, we distinguished the four main claims made by monkey portfolio proponents: (i) all smart beta strategies lead to similar performance; (ii) all smart beta strategies have unavoidable value and small cap tilts resulting in performance that is similar across strategies; (iii) smart beta strategies are as good as inverse or “upside-down” strategies; and (iv) smart beta performance is due to a rebalancing effect.

Our results were not supportive of the “monkey portfolio” argument. We found that various smart beta strategies displayed pronounced differences in performance characteristics and factor exposures.

We also obtained a reassuring finding that inverting a portfolio strategy does not, in general, lead to the same performance as the original. Moreover, our tests did not support the idea that rebalancing is a main driver of smart beta performance.

Our findings imply that analysing smart beta performance and risks is not monkey business. For a better understanding of smart beta strategies it is crucial to analyse their construction principles, performance characteristics and risk-factor exposures, including not only value and small-cap factors, but also a variety of other well-documented risk factors such as momentum, profitability, investment, low risk, and possibly others.


Smart beta strategies

Obviously, whether or not the abovementioned arguments hold may depend heavily on which class of strategies one includes.

While their empirical tests are limited to selected strategies, the authors putting forward the monkey portfolio arguments claim that these results apply to smart beta strategies in general, meaning that an analysis of any choice of smart beta strategies should fulfil their claims.

Our selection of strategies focuses mainly on explicit factor-tilted smart beta strategies, which correspond to indices that providers have launched relatively recently.

In fact, the first generation of smart beta indices usually changed the weighting scheme from market cap-weighting, while paying no attention to explicitly controlling the exposures to systematic risk factors. Such strategies provided implicit tilts to systematic factors. More recently, many providers have launched factor-tilted indices to extract factor premia (explicit factor tilts).

The increasing interest in factor-tilted smart beta indices is also due to the success of factor investing, especially since the Norwegian Oil Fund report by Ang, Goetzmann and Schaefer, which showed that the returns relative to a cap-weighted benchmark of the fund’s actively-managed portfolio can be explained by exposure to a set of well-documented alternative risk factors.

Among possible strategies, we include a broad set of smart beta strategies in our tests. First, we include the popular fundamental-weighted portfolio strategy and the equal-weighted strategy based on broad universe.

Given that many monkey portfolio proponents are also promoters of fundamental-weighted indices, it is interesting to first check whether their general claims apply to the type of smart beta strategy they promote.

Second, we include a variety of strategies that seek explicit exposure to a given risk factor by selecting stocks with desired factor exposures. Such smart beta strategies are being offered as “factor indices” by most major index providers.

We include two different approaches in our tests. First, we include indices which select stocks by factor score and apply a diversification-based weighting scheme (diversified factor indices); and second, we include indices which, in addition to selecting stocks with the highest factor exposure, also use a weighting that favours stocks with desired characteristics (concentrated factor indices).

It should be noted that our set of strategies deliberately differs from the strategies tested in some previous research that does not include explicit factor-tilted strategies in the set of test portfolios.

Our research can thus be understood as a test of whether the claims of monkey portfolio proponents apply only to the specifications of smart beta the authors have selected for their empirical tests, or whether they apply more generally to a broader set of commonly used smart beta strategies. In particular, given the recent interest in explicit factor-tilted strategies, it appears to be relevant to test the extent to which the strong claims of monkey portfolio proponents carry over to such strategies.


Insights from testing

While we have not attempted to conduct an exhaustive assessment of the above claims across all possible strategies, our analysis of some commonly employed smart beta strategies suggests that these statements do not hold in general. Our results show that, while some strategies such as fundamental equity indexation may perhaps be mostly driven by a value tilt and may generate similar performance to their upside-down counterpart, many smart beta strategies display exposure to additional factors, as well as pronounced differences in factor exposures across different strategies.

Moreover, and perhaps reassuringly, the inverse of these strategies generates inferior performance.

The differences from the original results underlying the monkey portfolio arguments can be explained by the fact that we consider strategies which explicitly tilt towards a variety of risk factors, while the evidence in favour of monkey portfolio claims had omitted such strategies and instead focused on a particular selection of strategies which may better correspond to such claims.

In particular, if one selects strategies which stay relatively close to equal-weighting, it may not be entirely surprising that the performance of such strategies is extensively driven by value and small-cap exposure, as has been documented for equal-weighted strategies. Moreover, it may not be surprising that strategies which are close to equal weighting can be inverted while maintaining comparable performance benefits.

An important insight from our tests is that one should be careful to not overgeneralise results that have been derived from testing particular strategies.

While the monkey portfolio arguments may apply to particular strategies, they have been invalidated for the explicit factor strategies we chose as our main test portfolios here, and therefore these claims cannot be applied to smart beta strategies in general.

In line with our other rejections of the other aspects of the monkey portfolio argument, we do not find that rebalancing is of great consequence in explaining smart beta performance.

Our findings of important differences across various smart beta strategies imply that care must be taken not to fall into the trap of oversimplification and overgeneralisation.

The differences in factor exposures across smart beta strategies imply that using a particular set of indices corresponds to particular factor selection and factor allocation decisions.

Moreover, the different factor tilts play an important role in shaping the risk–return profile of smart beta strategies. Factor-tilted smart beta strategies perform due to large positive exposure to their respective factors, while their inverted counterparts underperform the originals due to less pronounced or negative exposure to the same factors.

When considering the adoption of smart beta strategies, investors should carefully consider which set of factor exposures is best-aligned with their investment beliefs and objectives. Making a selection from among smart beta strategies is not monkey business after all.


Felix Goltz is head of applied research at the EDHEC-Risk Institute, and research director at ERI Scientific Beta