I have always thought that mechanical Magic Formula Investing is not without its problems. It makes mistakes, and one of the most egregious ones, I think, is Chinese companies. I have written about the problems with Chinese companies a long time ago, and I understand that US implementations of Magic Formula also had problems with Chinese RTOs.
If you are an active stockpicker, then the temptation is to try to weed out what you perceive to be mistakes. This is something that Greenblatt advised against, as people are inclined to kick out the weaker looking stocks, only to find that those are the ones that make a strong comeback due to their cheap price.
There are a number of theoretical problems I have with the Magic Formula:
1. they can incorporate anomalous results. An aberrational earnings figure will distort both the ROC and EY figures
2. I’m not happy with using ROC. It’s not uncommon for the formula to produce ROCs greater than 100%. I just don’t believe those figures. My understanding is that ROC is ultimately trying to produce an incremental return on capital – a good idea – but I’m not sure it succeeds. If you look at AZN (Astrazeneca) for example, it has a Magic Formula Score of A-. Its ROCE (not ROC) is 17.3%, and its PE is 12.6. But, its profits are going down, and fixed assets are going up. If we just take that at face value, then it is earning negative incremental returns. Now, there are some weaknesses with that argument: the decline is most likely due to the decline in returns of assets in place, rather than growth assets, so that doesn’t count. And yet, I’m getting the impression that AZN is trying to make up for declining patents by making acquisitions; bad acquisitions at that. A similar, but not identical, argument can be make for Microsoft. It is on a modest PE, and its ROE is through the roof. But the point is, a lot of that is due its Windows and Office franchises. Incremental return on capital is quite poor. To illustrate: operating income for 12 m/e Jun 2013 was $26.6b. For 12 m/3 Jun 2010 it was $24.2b. That’s an increase of $2.4b. Its capital employed (total assets less current liabilities) was $105.0b on 30 Jun 2013, and on 30 Jun 2010 it was $29.6b. So the capital has increased by $75.4b. Now, admittedly, cash held has increased about $40b during that period, so let’s just strip that out of the capital. Now we get incremental income of $2.4b for incremental capital of $35.4b (75.4-40). So Microsoft’s incremental return on capital has been 6.8% (2.4/35.4). Oh. Suddenly Microsoft’s return on investments doesn’t look all that exciting. So, perhaps we shouldn’t be surprised that you could buy MSFT for $30 per share in 2009 (in fact, it looks like you could have bought it for that as early as 2006), and in 2013, it could still have been bought for that price.
But I digress …
Stockopedia’s Magic Formula screen
(http://www.stockopedia.com/screens/greenblatts-magic-formula-1/#display) actually made a decent return over 1 year: up 23.6% against the Footsie of 10.8%. that’s an outperformance of 12.7%. Not to be sniffed at, for sure – although the outperformance over 2 years is somewhat narrower.
NOW. Supposing you had stripped out the Chinese companies from the Magic Formula? Then you would have left out 4 of the 30 companies, according to my reckoning. The stock symbols, together with their RS1y, are as follows: CAMK -18.5%, CTEK -44.1%, ZHEH -10.4%, CAML 7.6%. Notice how their performances were pretty abysmal. Only CAML actually beat the market, but it still hampered the relative outperformance of the Magic Formula. Now, according to my calculations, those 4 stocks had a combined underperformance of 65.4%, which means that the remaining 24 companies in the portfolio would have had an average outperformance of 18.6%. See what just happened? You gained about another 6% outperformance (18.6% as opposed to 12.7% just by cutting out the Chinese stocks. And an 18.6% outperformance very good to me, even if we stop there and don’t try to figure out what other nasties might be lurking in the Magic Formula portfolio.
Edit 22-Dec-2013 Aswath Damodaran echos much of my thinking on Miscrosoft here .