An important feature of the financial system is its ability to self-evolve. In order to satisfy the market demand industry practitioners always need to come up new investment solutions. When some of the established investment methods seem to have run their course, there is always a good chance that a new solution will emerge based from a different angle. After more than a decade of development, smart beta strategy has gained recognition from investors for its transparency, low fees and investment intuition. Over time, they gained a deeper understanding of smart beta technology and find that the early versions of Smart Beta have some structural concerns, as a single investment style can be easily affected by the macro environment and may underperform the parent benchmark over time. The emergence of Smart Beta 2.0 is to address these concerns.
To solve the problem of a single investment style, the new generation of Smart Beta strategy adopts the concept of multi-factor investing. To understand what multi-factor investing is, we must first understand what factors are. In simple terms, factors are the characteristics of stocks. For example, the size of a company is a characteristic, and so is the P/E ratio. Theoretically, all characteristics could be used as factors, but in practice, only a limited number of characteristics can be used in a multi-factor investing framework.
Several important criteria must be met for a factor to be used as a factor in a multi-factor investment system. The first is that the factor need to have enough historical data. Sufficient historical data can provide a longer backtesting period, which will increase the reliability of the statistical results. The second is that there must be good intuition behind the factor being used. The third is that the construction of the factor has to be simple, so investors can easily understand the logic and concept of the factor. The fourth is that the factor must have the ability to predict the stock return, meaning that the factor can bring about a medium to long term factor risk premium.
After a period of research and development, a developed framework has emerged for multi-factor investing, which generally consists of six different factors. Although the index provider and the smart beta investment funds have developed their separate frameworks, there is a consensus that the selection of factors should not deviate too far from these choices.
The six factors are value, quality, momentum, dividend yield, low volatility and low size.
Value could be represented by P/E ratio, P/B ratio or other fundamentals. Quality refers to whether the company is profitable, highly leveraged and has sufficient cash flow. Momentum refers to the recent uptrend of the share price. The dividend yield is the relationship between dividends and stock price. Low volatility can be calculated using a simple standard deviation - the scale can be the market capitalization of the company. These six factors represent different perspectives of stocks and are the essence of multi-factor investing.
The table shows the correlation between the excess returns of the factors (excess return is defined as the return of the factor index minus the return of the market capitalization weighted index), using the FTSE single factor index.
For example, the value factor and volatility factor returns have a negative correlation, while the volatility factor has a positive correlation with the quality factor. The lower the correlation, the greater the potential for diversification. The data in the table shows that the correlations between the factors are not high. The overall correlations are taken into account, and multi-factor investment uses these six factors as a basis. This basis is shown to have significant validity.
Different index providers or funds have similar definitions of factors, but the differences in multi-factor strategies lie in the use and combination of factors.
In general, multi-factor strategies can be categorized into two broad areas, with the first type of multi-factor model taking a stock screening approach. The second type of model is the factor tilting method. For example, S&P used a screening method to construct its quality-value index, while FTSE Russell used a factor titling approach to construct its multi-factor Index.
Double screening of stocks result in better quality
Stock screening is a bottom-up approach for portfolio construction. The construction of the S&P China A-Share Quality and Value Index is divided into two steps [Note 1]. The first step is to select the 200 stocks with the highest quality scores in the China A-share index according to S&P's definition of quality, and the second step is to select the 200 stocks with the highest quality scores in the China A-share index. The second step is to select the 100 stocks with the highest value scores out of the 200 stocks to become constituent stocks, and the final weighting of the constituent stocks will be based on the market capitalization of the individual stocks.
Double screening ensures that the constituent stocks have both good quality and low valuation characteristics. Using screening methods may result in over-concentration of individual stocks and sectors, therefore requiring the use of optimization algorithms. The company weight will be adjusted to comply with the restriction that the total industry weighting shall not exceed 40%, or the individual shares shall not exceed 5%.
Contrary to stock screening, factor tilting is a top-down model, where factor exposure is obtained by adjusting for individual stock weights. This is done by using statistical functions to standardize the factor values into a score between 0 and 1. The factor scores then are combined by calculating the average. The average can be either an arithmetic or geometric mean.
If the arithmetic mean is used, the main consideration is to balance the influence between the factors. When one factor scores high and one factor scores low, there will be a hedging impact. The geometric mean takes into account the interaction between the factors. When one factor score is high and one factor score is low, the combined score will be low due to the geometric nature. The final step is to adjust the weighting of the individual stocks according to the total average score, with higher total scores increasing the weighting and lower total scores reducing it.
These two multi-factor construction methods have their own characteristics. Stock screening is more straightforward, easier to understand and has a more concentrated weighting, hence a greater tracking error. It is believed that if the factors are effective in stock selection, the portfolio returns will be greater. Factor tilting focuses on trading and stock weighting control, and the number of stocks is larger, making it more suitable for larger investments.
After ten years of development, Smart Beta has transformed from a single style strategy to a multi-factor nature. In addition, we can observe that some even newer concepts have emerged, and recently there are new products that provide factor selection. Some products use risk parity strategies to allocate factor weights and some use factor rotation based on the macro environment. In addition, some more complicated factors are gradually being added to the factor list. As investors become smarter, it is expected that the smartness of smart beta strategies will increase as well, continuing to evolve itself.
Note 1: The description of the construction is a conceptual introduction. The precise details should be based on the basic rules.