I won't repeat what other reviewers have said. This book is a quick read, with a breezy tone, and in some simple ways helps to explain value investing, but...
A few problems that the author dismisses without any discussion.
1. Backtesting. Most backtested stock market systems don't work in the forward direction for very long. A good example is the Motley Fool's Foolish Four model, based on the Dow Dividend model. Backtested it looked great! But when a large number of people started to follow the model, it's performance approached mediocre. This makes sense. Wall Street is nothing but efficient. Any strategy that works will quickly be copied by tens of thousands of players, and this can quickly ruin a system. That's why hedge funds that use "black box" models don't publish the models.
And since Greenblatt tells the reader that the system only works over a three year period, it would be at least three years before one could tell the system wasn't working.
I would predict that the system will produce diminishing returns over the next ten years, proportionate to how many copies of the book that the author sells. Ironic that the richer that Greenblatt gets, the poorer his followers will get.
2. Trading costs: Greenblatt completely ignores trading costs and taxes in his analysis. If you follow his advice and buy 30 stocks, you would pay $779 in round-trip commissions at E-trade (or $600 if you had more than $50,000). That's about 1.5% a year in trading costs on $50,000 invested, or about 3% a year on $25,000. Or almost 8% on $10,000! That's a big expense drag, especially if the system doesn't outperform by as much as it claims to.
And taxes. If you do this strategy in a taxable account, you would incur another 15% to 25% hit in the form of capital gains taxes and state taxes, depending where you live. Thus the cost of the strategy could add up to as much as 33% a year in a state like New York or California. Again, very hard to make money this way, unless the strategy beats the market by 100% as it claims. (Of course, in a tax-deferred retirement account, this would not be a problem.)
The alternative, investing and holding a diversified by asset type group of index funds, would have a yearly cost of less than 0.3 % a year, and would generate little or no taxes until sold many years later.
3. Being hostage to his website, which is carefully labeled "free for now." Because his formulas don't translate well to free financial sites, the user of this system will have to depend on the generosity and fairness of the author. What if you start to use them system, and two years from now the site becomes an expensive pay site? That just adds another expense, and each expense becomes a drag on performance.
4. Dubious endorsements. So what if the publishers got rave reviews from famous investors? This doesn't mean that the book has any merit. I'd like to ask each of those investors if they plan on replacing their own investing style with the author's system.
5. No analysis of risk-adjusted returns. A basic principle of investing is that you get paid for risk taken. Thus small stocks, which are riskier, tend to return more over time. Riskless assets such as treasury bill will tend to yield low returns.
I'd like to see risk-adjusted performance data. For instance, what is the Sharpe ratio of the stocks picked by his system? What are the standard deviations? What are the beta's? If the system really is good, then it should look good on a risk-adjusted basis. The interested reader might want to take a look at William Bernstein's excellent The Four Pillars of Investing : Lessons for Building a Winning Portfolio if they want a thorough understanding of risk and return.
I'd also like to see a better attitude from Mr. Greenblatt on his website. Amazon reviewers are a serious, thoughtful group for the most part, and perhaps responding to some of the issues raised in an open way would help readers and potential readers of his book to understand it better.
I guess I'd like to conclude by reminding everyone of the standard investment disclaimer..."Past performance does not guarantee future performance in any way."