The following commentary comes from an independent investor or market observer as part of TheStreet's guest contributor program, which is separate from the company's news coverage.NEW YORK ( MagicDiligence) -- Today I'm using the Magic Formula Investing (MFI) method to test a range of stocks. Joel Greenblatt, the founder of MFI, explained the strategy in broad terms in his book The Little Book that Beats the Market, but the exact calculations have always been a bit of a mystery. We know the spirit of the two statistics. Earnings yield is defined as (EBIT/enterprise value). Return on capital is defined as (EBIT/tangible invested capital). Many have tried to duplicate it, and MagicDiligence has tools to calculate MFI statistics for any stock. Using these tools, I've found what looks like a few discrepancies between my calculations and those on the official site. So let's cut three of the "official" stocks that don't make sense and replace them with some high-rankers that, for unknown reasons, don't screen officially. The screen we're working with here is the top 50 stocks over $50 million market cap. Cull These "Official" Picks 1. H&R Block ( HRB). H&R Block is not an expensive stock, but doesn't look either particularly cheap or super-efficient. By my calculations, MFI earnings yield is 8.7% and return on capital is 22%, neither of which should screen the stock into the top 50. I am not exactly sure what Greenblatt is doing with H&R Block's roughly $500 million in mortgage loans held for investment. Since Block includes interest income as part of operating earnings, they should be added to invested capital. Even taking them out does not dramatically improve return on capital. I'm throwing Block out of the screen. 2. China North East Petroleum ( NEP). We know from Greenblatt himself that he has specifically filtered out Chinese reverese take-overs from MFI after they littered the screen in 2009. So why does NEP remain? It has a history of SEC filing delays and a rotating door at CFO, putting accounting into question. There are so many other cheap stocks without basic accounting concerns, so this one gets chopped as well. 3. Korn Ferry ( KFY). Korn Ferry is cheap with a 17.5% earnings yield, but return on tangible capital at 27% is nothing special, and certainly shouldn't screen the stock into the top 50 (over $50 million). I've looked over the numbers many times on this one and cannot understand why it is an official MFI stock -- there is little to misinterpret. Perhaps the data set is just wrong. Replace With These "Unofficial" Picks 1. Incredimail ( MAIL). At a nearly 40% MFI earnings yield and over 400% pre-tax return on tangible capital, how can MAIL not be screened? The stock has also consistently paid a monster dividend of about 15% for the past three years. It is an Israeli-based company, but so was MIND CTI ( MNDO) and Cellcom Israel ( CEL) when they were on the official screen. (Israeli companies all seem to pay huge dividends) The company makes software to spruce up emails, has 55% of its market cap in cash and carries no debt. 2. Research In Motion ( RIMM). RIMM has been in the official screen before, but is not today and hasn't been for some time. At an earnings yield of 35.5% and a return on capital of 93%, it is, by far, the highest-ranking stock with a market cap over $10 billion. It certainly should rank over official stocks like Dell ( DELL) (16.1%; 85%) and Microsoft ( MSFT) (15.1%; 133%). This one is a mystery. 3. Amtech ( ASYS). MFI has a number of chip-equipment stocks, but why not Amtech? My calculations show an earnings yield of 88.6% and return on capital of 94%, putting it above "official" competitors like LTX-Credence ( LTXC) (27%; 103%) and KLA-Tencor ( KLAC) (24%; 87%). Amtech is attractive, too, with a history of rapid growth, a stellar balance sheet ($60 million in cash with no debt), and it's trading near a 52-week low. The writer owns DELL, MSFT, MNDO.