When I compare your table of 30 with the results of the same screen Greenblatt magic formula screen in today's article in Investors Chronicle, there are only 6 shares the same. The majority of theirs fail to be included in your screen as they are regarded as outside the top 30 when sorted by the magic number. I will be asking the same question of IC, but it casts doubt on both organisations when the screens come out so differently. Perhaps you can comment on why this might be, as at the minute I'm not sure I trust either.
Hi Igotts. I can't comment on how the Investors Chronicle are compiling their 'Greenblatt' screen. But I do know intimately how we have compiled ours. We're pretty meticulous about using the precise definition used by Greenblatt, to the extent that it is clearly articulated in the book. We discuss how we do this in depth in this detailed article which I recommend reading thoroughly, but in summary, the steps are:
- Calculate Earnings Yield = EBIT / enterprise value.
- Calculate Return on Capital = EBIT / (Net fixed assets + working capital)
- EY Rank: Rank the stocks in descending order based on Earnings Yield and assign a rank number to each.
- ROC Rank: Rank the same stocks in descending order based on Return on Capital and assign a rank number to each.
- Add the rankings and select stocks that have the lowest combined ranking score.
We are aware that some sites use more of an approximation to the Magic Formula by using, for example, ROA rather Greenblatt's highly specific ROC definiton.
For validation of our approach it's worth looking to third party usage of our implementation of this screening system. You can see some interesting discussion on the Magic Formula here by Richard Beddard, chief blogger at Interactive Investor who bases his systematic Money Observer portfolio using our Magic Formula and Piotroski data:
1) http://www.iii.co.uk/news-opinion/richard-beddard/formula-not-so-magic-uk/ (he critiques some other approaches here)
2) http://www.moneyobserver.com/issue/features/nifty-thrifty-portfolio-gets-annual-overhaul (he discusses our approach).
If the Investors Chronicle haven't used the same database as (ours is based on Thomson Reuters data) or used the same precisely defined metrics then they will come out with wildly different results. We also use 'TTM' (trailing twelve month) data whereas they are likely to be using historic data - our dataset is the only TTM database I am aware of in the UK and likely more timely.
It's also worth noting that our screener is completely custom built here at Stockopedia. We calculate all our own ratios and metrics in house giving us the ability to create bespoke definitions of these kinds of fields.
Hopefully this helps. Ultimately one has to do the due dil on who has the more timely, accurate implementation and data. I hope I have helped to settle any nerves with this brief post.
