This algorithm seeks to replicate the strategy set forth in Tobias Carlisle's "Deep Value" See his Google Talk for a great summary of the strategy.
In "Deep Value" and in his first book, "Quantitative Value", Tobias demonstrates that the Enterprise Multiple or "Acquirer's Multiple", defined as Enterprise Value/EBIT outperforms all other value ratios. The books and their back-tests show that when ranking stocks based on the Enterprise Multiple, Value stocks (those with the lowest ratio of EV to EBIT) outperform Glamour stocks on the same spectrum.
As with many value strategies, this one holds no more than 30 stocks at a time and there are some additional filters, including a minimum market cap in the 40th percentile. Re-balancing is performed annually using data available as of the previous month, ideally. Hopefully future fundamentals API changes will allow more exact querying of variables on different timescales.
I'm not convinced that I am 100% faithfully replicating the strategy outlined in the book, but the results seem promising. One compromise I made was to allow negative EV/EBIT ratios. I think Tobias ruled these out, but they seem to contribute significantly to the performance.
Feedback is welcome!