The “Five Criteria for Good Companies” model portfolio generated solid returns during the backtesting period from January 2006-2017, outperforming the Standard & Poor’s 500 exchange-traded fund benchmark in nine of the past 12 years. As of June 20, the portfolio generated a cumulative return of 261.72%, yielding an average excess return of about 5.72%.
Five criteria for good companies
In the past year, I have researched investing strategies from famous value investors, including Peter Lynch, Warren Buffett (Trades, Portfolio), Charlie Munger (Trades, Portfolio) and Ben Graham. Based on my research, good companies have the following characteristics:
- A GuruFocus business predictability rank of at least four stars.
- A Piotroski F-score of at least 7.
- An operating margin growth rate of at least 2%.
- A gross margin of at least 40%.
- A 10-year revenue growth rate and earnings per share (without nonrecurring items) growth rate between 5% and 20%.
Eleven companies, including Comcast Corp. (CMCSA, Financial), Amgen Inc. (AMGN, Financial), Thermo Fisher Scientific Inc. (TMO, Financial), Jack Henry & Associates Inc. (JKHY, Financial) and Rollins Inc. (ROL, Financial), made the screener as of June 20.
Brief introduction of portfolio statistics measures
Portfolio risk management allows us to analyze the performance of a portfolio compared to a benchmark, like the S&P 500 ETF. The top five statistical measures include the standard deviation (sigma), the beta, the alpha, the Sharpe ratio and the coefficient of determination (R-squared). Other key measures include the tracking error, information ratio and the batting average.
Each statistical measure gives a different perspective on portfolio performance: some measures directly compare the portfolio performance to the benchmark performance while others discuss various portfolio characteristics, including risk.
Portfolio sigma
The sigma (σ) determines the volatility of the portfolio returns around the historical mean return. Based on the standard deviation, investors can compute the predicted range of returns for a given portfolio. Higher standard deviations imply less predictable portfolio returns, increasing the risk of the portfolio.
The “Five Criteria for Good Companies” model portfolio has a sigma of 16.28%, suggesting a wide range of returns. The -30.54% return in 2008 followed by a 33.07% return in 2009 contributed to the high portfolio standard deviation.
Tracking error
The tracking error measures how closely a portfolio follows the benchmark – i.e., the standard deviation of the difference between the annual portfolio and benchmark returns. According to Prudential Financial Inc.’s (PRU, Financial) Global Investment Management Business, the tracking error can provide an “acceptable range of relative performance when evaluating a manager.” Several measures, including style bets, security selection and various transaction costs and fees, can increase portfolio tracking error.
The model portfolio’s tracking error is 5.797%, suggesting that the “Five Criteria for Good Companies” does not closely follow the S&P 500 benchmark. The result is expected as the portfolio ignores the “average” and “below average” companies.
Beta
Unlike the sigma, the beta (β) measures “how sensitive a portfolio is to market movements,” according to Prudential Investments. This measure, which equals the ratio of the covariance between the portfolio returns and the benchmark returns to the variance of the benchmark returns, reports the expected change in portfolio performance for every 1% change in benchmark performance.
The model portfolio has a beta of 0.945, which is slightly less than one. This suggests that the model portfolio is slightly less volatile than the S&P 500 ETF benchmark.
Alpha
Portfolio alpha, one of the most important risk measures, reports the “value added or subtracted by a manager,” according to Prudential. In other words, the alpha (α) determines if the portfolio generated higher or lower returns than the expected return, given the portfolio beta.
American economist Mike Jensen defines the alpha using a variant of the capital asset pricing model (CAPM): Alpha equals the difference between the portfolio return and the risk-free rate of return minus beta times the difference between the benchmark return and the risk-free rate of return. The regression model computes the alpha and beta based on Jensen’s equation.
I computed alpha using a variant of Jensen’s equation: Instead of the annual returns, I computed the annualized returns for the portfolio and the benchmark and used these values in the equation. My model portfolio returned about 11.31% per year while the benchmark returned just 5.79%, yielding an alpha of 5.72%.
Sharpe ratio
The Sharpe ratio, according to Prudential Investments, measures the “amount of return earned per unit of associated risk.” William F. Sharpe’s ratio of investment efficiently equals the portfolio alpha minus the risk-free rate all over the portfolio sigma. Higher Sharpe ratios imply stronger historical risk-adjusted performance.
The model portfolio has a Sharpe ratio of 21.87%, suggesting that the portfolio achieved a return of 21.87% per unit of risk.
Information ratio
Unlike the Sharpe ratio, the information ratio reports the ratio of excess return of a portfolio to the tracking error. According to Prudential Investments, the information ratio “relates the magnitude and consistency with which an investment outperformed its benchmark.” In other words, the information ratio can measure the amount of incremental return for each incremental unit of risk.
Although it has a high Sharpe ratio, my model portfolio has an information ratio of just 0.9865. This suggests that for every incremental unit of risk, the “Five Criteria for Good Companies” portfolio returns just 0.9865.
Batting average
The batting average, according to Prudential, shows how consistently the portfolio outperformed the benchmark. This metric, which is borrowed from baseball, equals the ratio of the number of “outperformance years” to the total number of years. (You can treat the portfolio as the “batter” and the benchmark as the “pitcher.” If the portfolio outperformed the benchmark in a given year, the portfolio scores one hit for that year. Otherwise, the portfolio “strikes out” and scores zero for that year. The batting average is simply the number of hits divided by the total number of years – or innings.)
The model portfolio has a batting average of .750 – i.e., the portfolio outperformed the benchmark in nine of the past 12 years.
Conclusions and see also
Based on these portfolio statistics measures, the “Five Criteria for Good Companies” investing strategy generates good excess returns compared to the S&P 500 ETF benchmark. The stocks offer good growth and value potential to the portfolio.
According to previous articles, GuruFocus introduced several new features to the All-in-one Screener: screening using historical data and screening from a portfolio. These features allow GuruFocus users to interact with their portfolios and run deeper analyses. We expect to add a new feature utilizing these portfolio statistics in the upcoming weeks.
Disclosure: The author has no position in the stocks mentioned.
