This [cached]short book elaborates on Graham's principle of value investing and puts it into a concrete form, boiling it all the way down to a simple formula you can use for investing: Buy above average companies at below average prizes.
To be more specific, pick a cut off in market capitalization (say 50 million), then rank the remaining stocks based on p/e ratio and return on assets/capital. Buy those that are best when adding both ranks. Gradually phase these portfolio in in groups of a few stocks, then update whenever a stock has been held for one year. In total you should hold about 30 stocks at any one time.
Greenblatt claims excellent returns for this strategy over the last two decades (20 - 30% annually!), but this seems to have been calculated without taking fees and transaction costs into account - the strategy requires roughly 60 a year. Still, the [cached]formula seems to work.
One potential downside is that the formula can under-perform the market for several years in a row, making it very hard to stick with it.
I'm still on the fence of what to make of this formula, whether it makes sense for individual investing (transaction costs are especially painful if you are buying in small volumes) or if I should just stick with index funds. At least Greenblatt seems to put his money where his mouth is, he has several [cached]"Formula Investing" mutual funds.
Greenblatt didn't just use simple p/e or RoA, but rather:
- Return on capital: EBIT/(Net Working Capital + Net Fixed Assets)
- Instead of p/e: EBIT/ Enterprise Value (market cap + debt)