Factor investing #8: Multi-factor portfolio construction

Since CN&CO is very involved in many aspects of the financial services industry, we often come across useful and informative articles that we believe our friends and partners will be interested in reading. The latest is a series of articles that unpack the concept of factor investing and how it’s used in different situations.

Factor investing is one of the tools used by portfolio managers to maximise returns in an investor’s portfolio. A few definitions are available here. This investment strategy has the ability to empower consultants, multi-managers and advisors to build client portfolios simply and efficiently. From 1) the transparent manner in which factor portfolios are systematically constructed, to 2) the capability of building tailored investment outcomes with greater diversification and predictability, to 3) the low fees, to 4) the reliability in consistently delivering a specific investment philosophy. Factor investing is beginning to revolutionise the investment industry.

While the potential impact of factor investing is transforming in nature, the level of adoption by clients varies by degree of simplicity, from ‘not allocating’ to ‘sophisticated allocation’. In this series we aim to highlight all applications of factor investing across this continuum.

This article is the eighth in our series of articles aimed at discussing practical ways to employ the power of factor investing. In our previous article, we discussed how best to incorporate responsible investing (RI) or ESG principles within the process of factor investing. In this article, we discuss an optimal approach to blending factors exposures, called multi-factor portfolio construction.

Multi-factor portfolio construction

By Jason Swartz

Client level of adoption/allocation:

In part four of this series we discussed the diversification benefits of allocating strategic weights to single factor strategies. We note, however, that in that exercise, we were ‘blending’ factors… essentially combining single factors on a sub-portfolio basis. Another approach to combining factors would be multi-factor portfolio construction.

This approach first aggregates factor rankings on a stock level, and then builds a portfolio based on this overall factor score. Compared to the blended approach, a multi-factor approach improves transfer of information by finding stocks with reasonably positive exposures to multiple factors rather than stocks with very strong exposures that may simultaneously have very poor exposures to another factor (see Figure 1).

Figure 1: Scatter plot of Value vs Momentum z-scores (correlation = -0.25)

In the graph above, the grey stocks (in the blended portfolio but not multi-factor portfolio) have high factor scores despite them being potentially offsetting. On the other hand, the light blue stocks have more balanced positive factor scores.

Another advantage using a multi-factor approach would be trading efficiencies. By not having to execute separately managed single factor portfolios, turnover is reduced by netting offsetting trades against each other.

We decided to empirically test a multi-factor approach in South Africa¹ in order to assess the extent of the improvement in efficiencies versus a blended approach. The results both in overall factor exposures (Figure 2) and risk-adjusted performance since 2001 (Figure 3) are positive. The enhanced factor exposures versus the blended portfolio illustrates the stronger reflection of intended factors, as opposed to a blended approach where exposures may be watered down given opposing factor scores in sub-portfolios.

¹We formulate normalized factor z-scores for each of the 4 underlying factor signals: Momentum, Value, Quality and Low Vol. Multi-factor z-score is simply an equally weighted average of the underlying factor z-scores. At each rebalance period (quarterly, from 2001 to 2016), the score gets optimised against a SWIX benchmark. We apply a tracking error constraint of 5%.

Figure 2: Average Factor exposures of multi-factor versus blended portfolios (2001 to 2016)

Figure 3: Back-tested factor portfolio results (2001 to 2016)

We will, however, note that multi-factor benefits are dependent on a few considerations, including how correlated underlying factors are to each other, the number of factors used as well as the tracking error applied to the multi-factor portfolio.

Ultimately, the value of multi-factor portfolio construction is improved factor efficiency and enhanced risk adjusted returns. Notwithstanding this benefit, blending single factor portfolio has value and application to clients who prioritise transparency of portfolio attribution and flexibility in allocating between factors. The crucial determining factor is whether the client is searching for complete factor solution or wanting to allocate between factors.

For more information on this topic, or the details of our analysis, please feel free to contact us directly.

In our next part in the series, we discuss building outcome orientated multi-factor portfolios

Watch the video below for an optimal approach to blending factors exposures, called multi-factor portfolio construction.