What Methods Have Actually Historically Worked To Predict Winning Stocks?
This article is a book review of “what Works on Wall Street” by James P. O’ Shaughnessy. (1997 Revised Edition)
The book tests how various popular investing strategies would have worked by applying the strategies to 46 years of data (1951 through 1996) from Standard & Poor’s Compustat database.
In all cases the strategies were evaluated by creating a portfolio of the 50 stocks that rated the highest on a given strategy and then rebalancing to the new group of 50 best stocks once each year.
Strategies examined include:
- Value based approaches of buying stocks with the lowest Price /Earnings, the lowest Price/Sales, the highest dividend yields, the lowest Price / Book Value, the highest earnings yield.
- Momentum strategies of buying stocks that had risen the most in price during the past year.
- Profitability strategies of buying stocks with the highest return on equity or the lowest price to cash flow or the highest earnings growth in the past five years.
- Strategies of buying the stocks with the highest Price / Earnings
- Contrarian strategies of buying stocks that have fallen the most in the past year.
The conclusion was that (contrary to the efficient market hypothesis) some strategies consistently reward investors while other strategies consistently punish investors through massive under-performance.
The single best strategy found was to combine value, earnings growth and share price momentum.
50% of the portfolio is restricted to stocks with a price / sales ratio of less than 1.5 (a value screen). Stocks must also have year-over-year earnings growth. The stocks with the highest year-over-year share price momentum are then selected from this group.
The other 50% of the portfolio is invested strictly in market leading leading stocks (must be large cap stocks , utilities are excluded, larger than average number of shares outstanding, higher than average sales per share , cash flow at least 1.5 times the compusat mean). The stocks with the highest dividend yield from among these market leaders are selected. (Effectively this appears to be very similar to selecting the highest dividend yielding stocks, excluding utilities, from the largest 100 or so stocks by market capitalization).
This recommended strategy provided stellar returns (17.44% compounded return over 46 year) and minimal risk. It beat a strategy of investing equal dollar amounts in all stocks in the database (which yielded 13.35%) in 8 out of 10 years, in 34 of 40 rolling 5 year periods and in all rolling 10 year periods.
Other strategies that worked very well (though sometimes with higher risks were):
- Require Price to sales ratio less than 1, and choose stocks with the highest 1 year share price momentum. (Highest return but riskier than the recommended strategy).
- Require P/E < 20, choose stocks with highest 1 year share price momentum. (Results very similar to the strategy just above).
- Require Price to book less than 1, choose stocks with the highest 1 year share price momentum.
The worse strategies were those that selected stocks with the most extreme valuations such as highest price to earnings, highest price to book and highest price to sales. The single worse strategy was to select the stocks with the biggest year over year share price declines.
Pure value strategies that ignore share price momentum did reasonably well. Low P/E over all stocks just slightly outperformed the index, but low price to book, low price to cash flow and particularly low price to sales all significantly outperformed the index.
This book argues that a mechanical, screening based approach to stock picking is best. Attempting to apply judgment and bend the rules in certain cases will lead to trouble. A mechanical screening approach insures consistency.
The best stock picking strategies should always include at least one value screen to insure that you don’t pay an unreasonably high price. Positive price momentum is also a key factor that should be included.
Shawn Allen, CFA, CMA, MBA, P.Eng.