One may believe that a qualified quantitative strategy has to be complicated, but this belief may not necessarily be true. The main job for quantitative researchers is to discover and innovate, rather than to create sophistication. Discovery is considered as the most difficult part of the research process. Once the truth reveals itself the following implementation and testing is normally straightforward. A recent research paper (The Conservative Formula: Quantitative Investing Made Easy, The Journal of Portfolio Management Summer 2018) perfectly illustrates this thought.
The authors proposed a stock selection methodology labeled as “Conservative Formula”. The first task in the paper is to define what “Conservative” means and the authors used volatility and total net payout yield as the conservative indicators. Stock volatility is calculated from the daily returns for the past 36 months, and the total net payout yield is the dividend plus the share repurchase divided by the company’s total market capitalization. In addition to these two indicators, the conservative formula examines whether a stock has a medium-term catalyst, represented by the return of the stock in the last 12 months (minus the most recent month). The portfolio construction process begins by identifying 1,000 stocks with the largest market capitalization as the underlying universe. Stock volatility is then used to split the stock universe into two halves, each with 500 stocks. The third step is to assign a score of 1 to 500 to the low volatility stocks according to dividend yield and momentum and the resulting average score is calculated. Stocks with the highest 100 ranking will be included as constituents. The stock screening is conducted quarterly to reduce transaction costs. The research article includes the US stock data from 1929 to 2016. The empirical results indicate that the annualized return generated by the “Conservative Formula” is significantly higher than the broad market cap weighted portfolio (15.1% vs. 9.3%).
The interesting result leads one wondering whether the strategy can be applied to the China A-shares market. The structure of the China A-shares is special that the trading volumes are largely generated by retail investors. These investors are widely known to love eye-catching stocks which exhibit high volatility feature. Such preference is an ideal environment for the low volatility strategy. With the U.S. stock universe replaced by the CSI 300 Index universe a similar backtest is performed. The sample period is from August 2008 to August 2018. A few parameter changes are made with regards to the smaller universe. The first step is to divide 300 stocks into two groups of 150 stocks according to the stock volatility. The dividend yield and momentum are again used for further screening. The final portfolio consists of 50 stocks. The backtest results show that the conservative investing is also applicable to the A-share market. The historical annualized return of the portfolio constructed by “Conservative Formula” is 10.95%, while the return of the CSI 300 index is only 1.37% for the sample period. In terms of realized volatility, the magnitude from the conservative formula is lower than that of the CSI 300 Index (25.5% vs. 27.1%), reflecting the usual characteristics of low-volatility strategy. The average turnover rate per quarter was found to be 70% two-way.
Source: Rivermap Quantitative Research. Data from August 2008 to August 2018. Returns shown are before transaction fees.
In an era when data and computer powers are abundant complexity of quantitative models inevitably increases as result. Through the study of a stock selection strategy with a simple and neat concept it reminds us that underlying logic and insights, rather than model complexity, should still be the main elements in any investment strategy.
Reference: Blitz D. and P. van Vliet. “The Conservative Formula: Quantitative Investing Made Easy.” Journal of Portfolio Management, Vol. 44, No. 7(2018), pp. 24-38.
Disclaimer: All information is provided for information purposes only. Every effort is made to ensure that all information given in this research is accurate, but no responsibility or liability can be accepted by any member of Rivermap Company Limited (Hong Kong) nor their respective directors, or employees for any errors or for any loss from use of this research or any of the information or data contained herein. Past performance is no guarantee of future results. Back-tested performance is not actual performance, but is hypothetical.