[DesertCat]

Joel's a great investor who writes really well. His first book "how to be a stock market genius" has a prominent place on my book shelf. It laid out the math behind diversification and why it's better to have a focused portfolio. It also talked about where the "nooks and crannys" of the market are, where you can find undervalued companies.

I'm still waiting my copy of the new book. I've been to the web site and analysed a few companies it recommended, and I have some (minor) issues with the approach.

First, he's trying to create a formula for people who don't want to do the real work of understanding the businesses behind the stocks. So the book has a rule, sell it after one year. That way you don't have to worry about figuring out the right time to sell. I don't like formulas and don't feel this approach is optimal. I don't doubt it beats the market however. He's done a great deal of backtesting to show it works. The question is if he popularizes it, will the future returns be as good with thousands of investors chasing the same stocks?

Secondly, when I analysed recommendations from his web site, some of them had a ton of hair on them. They were actually very risky, but they happened to fit the arbitrary metrics. They were companies with severe issues that might have had a high ROE previously, but were unlikely to any more, or only had a high ROE because of a one time event (asset sale) that wouldn't be repeated.

So if you follow his formula blindly, you must be very diversified as he recommends (ironic considering his last book) to avoid having too much risk from some of the dogs that will end up in your screen. But I think if you use the screen intelligently (i.e. research each pick carefully), you can separate the wheat from the chaff, sell at better times, and end up with even better returns.

Joel had an interview on CBS marketwatch which said the formula worked well, but since the formula only took into account past performance you would do better if you did analysis of the companies likely future performance.