Creating value through smart-beta ETFs: Myth or reality?

Smart beta ETFs have evolved since the first single-factor ETF appeared back in 2003. Investors today can choose from exposure to a single-factor or multifactor approach that combines factors in an equal- or custom-weighted way. We believe that a transparent, rule-based, multifactor, smart-beta strategy with custom factor weightings may provide improved long-term, risk-adjusted returns with better downside protection.

The importance of factor weights and factor measures

As we can see in Chart 1, individual factor performance over the past decade has varied considerably from year to year. Take momentum for example which was the best performing factor in 2007 and then the worst performing factor in 2008 during the financial crisis. Given this variability in performance, it may be difficult for investors to predict which factors would be in favour in a particular period.

Chart 1: Calendar year returns (%) of MSCI Factor Indexes compared to MSCI ACWI Index 2007–2016
Chart 1: Calendar year returns (%) of MSCI Factor Indexes compared to MSCI ACWI Index 2007–2016
Source: MSCI
Past performance is not an indicator or a guarantee of future performance. Factor index performance is derived from backtested pre-inception performance and is not representative of any
ETFs’ performance. Performance for the MSCI ACWI Index represents actual performance.1

Our Systematic Modelling Team has conducted proprietary research that identified a mix of factors that could serve as an attractive component for an investor’s core portfolio. In summary, our smart-beta methodology focuses on both factor measures and factor weights.

1. Reconsidering factor definitions: Custom measures

While many smart-beta strategies utilise factors such as quality, value, momentum and low volatility, many also employ standard, widely accepted approaches to measure these factor exposures. We sought to develop custom factor measures, which we believe could provide a more comprehensive evaluation of a stock’s exposure to each factor. Chart 2 shows our factor metrics.

Chart 2: Factor metrics
Chart 2: Factor metrics
Source: Franklin Templeton
2. Reconsidering factor weights: Not all factors are weighted equal

As we considered the relative weightings of quality and value, we noted that our custom approach to the quality factor clearly produced higher risk-adjusted returns than the value factor, with improved information ratios of 0.45 and 0.13, respectively. Consequently, we have assigned a 50% weighting to the quality factor and a 30% weighting to the value factor.

Momentum and low volatility also play important – if smaller – roles in Franklin Templeton’s factor weights, with each assigned a 10% weight. Momentum may help identify investment trends and avoid value traps, while low volatility may help provide a defensive measure against market downturns.

We then examined performance characteristics of two different scenarios that use Franklin Templeton’s custom factor measures – one that combined equally weighted factors, and a second that combined factors based on the strategic weights outlined above – and compared them to the MSCI ACWI Index. The results of our analysis presented in Table 1 yield two key insights for the ten-year period:

  • Combining Franklin Templeton’s custom factors in equal weights improved hypothetical risk-adjusted returns (as measured by the Sharpe ratio) versus the MSCI ACWI Index
  • Combining Franklin Templeton’s custom factors using the strategic weights yielded even stronger hypothetical absolute and risk-adjusted returns

Leveraging multifactor solutions enables investors to minimise the cyclicality (and avoid trying to time the market) of single-factor investing. We believe that customised factor definitions and weights may be combined to seek better risk-adjusted returns over the long term.

1 MSCI ACWI Quality Index was incepted on 18/12/2012. MSCI ACWI Momentum Index was incepted on 15/2/2013. MSCI ACWI Minimum Volatility Index was incepted on 30/11/2009. MSCI ACWI Value Index was incepted on 8/12/1997. While the information is based on hypothetical pre-inception index returns for MSCI ACWI Quality Index, MSCI ACWI Momentum Index, MSCI ACWI Minimum Volatility Index and MSCI ACWI Value Index, they do not represent any ETFs’ actual performance. MSCI factor index performance includes pre-inception index returns, based on criteria applied retroactively prior to the index inception date and, as such, is hypothetical and for illustrative purposes only. They provide a general indication of the risk/return profile of the respective MSCI single factor indexes. Indexes are unmanaged, and one cannot invest directly in an index. They do not reflect any fees, expenses or sales charges.

Table 1: Performance statistics, 10 years, annualised

December 2006 to December 2016

Annualised return (%)

Sharpe ratio

Tracking error

Information ratio

Up capture ratio (%)

Down capture ratio (%)

Cumulative return (%)

Franklin multifactor strategic weights 50% quality, 30% value, 10% momentum, 10% low volatility








Franklin multifactor equal weights 25% quality, 25% value, 25% momentum, 25% low volatility
















Source: MSCI and Franklin Templeton. Performance for the MSCI ACWI Index represents actual performance. Franklin Templeton’s combined-factor analysis (equal weights and strategic weights) represents hypothetical, back-tested performance calculated by Franklin Templeton for factor methodology research purposes. The actual performance of any exchange-traded product may vary significantly from the pre-inception data presented due to assumptions regarding fees, transaction costs, liquidity or other market factors.

Commerzbank Disclaimer
The views expressed in this article are those of the author and may differ from the published views of Commerzbank Corporate Clients Research Department, the communication has been prepared separately of such department. No representations, guarantees or warranties are made by Commerzbank with regard to the accuracy, completeness or suitability of the data.