Formula not so magic in UK

Share this

Down but not yet out

TLBTSBTMUS investors using Joel Greenblatt's magic formula (reviewed here) to pick shares have an easy task. His site spits out recommendations for free.

UK investors must improvise, using affordable data. Of the databases I use, Sharescope does not include the variables required to calculate the magic formula, although it might be possible to create a rough facsimile using alternatives. Sharelockholmes offers a Greenblatt ranking which, on the face of it, uses the correct variables. It ranks companies using:

  1. Earnings yield (EBIT:EV). This is a measure of value: the return in profit before interest (Earnings Before Interest and Tax) as a percentage of the purchase price of the whole business including its debt (Enterprise Value). 
  2. Return on Capital (ROC). This is a measure of profitability. It is the return, EBIT again, as a percentage of capital employed. 

The two variables are intended to identify shares in high quality (highly profitable) businesses at low prices.

But there's a problem. The definition of ROC is defined by Greenblatt in the appendix of 'The Little Book That Still Beats The Market' as:

EBIT/(Net Working Capital + Net Fixed Assets)

Sharelockholmes uses the same definition (see the glossary), but differs in the way it calculates net working capital.

I’m sorry, but there is no way to avoid a certain amount of nerdiness here. I think it's significant.

Sharelockholmes' definition of net working capital is:

Inventory + Receivables + Working Cash - Current Liabilities

Greenblatt doesn't provide an equation, but says:

Net working capital was used because a company has to fund its receivables and inventory (excess cash not needed to conduct the business was excluded from the calculation) but does not have to lay out money for its payables, as these are effectively an interest-free loan (short-term interest-bearing debt was excluded from current liabilities for this calculation).

Both count inventory and receivables (money owed to the company) as capital and exclude excess cash (Sharelockholmes counts only 'working cash').

One area of contention is the calculation of working/excess cash. Greenblatt doesn't say how he calculates it so Sharelockholmes has improvised an arbitrary rule that a company requires enough cash to cover any deficit in its net working capital (as defined by Sharelockholmes) plus 5% of its turnover. (For a discussion about excess cash see this thread on Investment Q & A).

Another is the deduction of current liabilities. Greenblatt does not deduct all current liabilities from the capital the company requires, he just deducts payables (money the company owes to suppliers). Although the company uses this money to fund its operations, there's no cost associated with it so we don't have to count it (according to Greenblatt - I've ignored payables to my cost before, but that's another story). Think of it as the suppliers’ capital.

Money borrowed from the bank (interest bearing short term debt) is part of the company's capital yet, because it's part of current liabilities Sharelockholmes deducts it along with payables. Many sites calculating the magic formula do the same thing including the only other one I know covering UK companies (Magic Formula Investing Europe - you can see how it calculates the magic formula here). I don't think these sites have the fine grained data required to decompose current liabilities.

The difference may be significant, it could make companies with large amounts of short term debt look more profitable, and therefore more desirable. When I discovered the discrepancy, I re-calculated ROC for companies identified by my Nifty Thrifty formula, which incorporates Greenblatt's rankings.

For many companies, those that had little short-term debt in relation to overall net working capital, it made little difference. But other companies may be catapulted up the rankings because Sharelockholmes deducts all current liabilities. Cobham's ROC, for example, was 30% when I deducted payables, but 60% when I deducted all current liabilities. De La Rue returned 47% by the former measure and 86% by the latter. The most extreme case was Ultra Electronics, which produced figures of 99% and 3,411%!

These discoveries have dented my enthusiasm for magic formula investing in the UK, but there is an alternative. Where the magic formula variables are not available, Greenblatt recommends substituting Return on Assets for ROC, " best approximate the results of the magic formula".

ROA is a much blunter instrument than ROC. It assumes that all of the assets owned by the company are required to produce a profit including excess cash, and goodwill, which Greenblatt prefers to ignore because it is a historical cost.

Although ROA lacks the finesse of ROC, I feel more comfortable with it, especially since I don’t have the data to screen for high ROC companies according to Greenblatt’s original definition.

Thankfully, Sharelockholmes has implemented ROA, so I can put it to the test (a pretty crude test, admittedly). Expect the results shortly.

I don’t expect the Earnings Yield and ROA formula to perform as well as the authentic magic formula would (and have no way of finding out), but I expect it to beat the market.


Hi Richard

Spooky - you've just published the article that I was about to write!

I've just finished the Little the last couple of days and, following my previous delve into ROE, was wondering how I could screen UK candidates using his rules.

I have an issue with how to determine and remove excess cash in particular - ie - if a trade customer pays one day before a financial year end versus one day after a year end, you could end up with a different result with this definition, whereas if you leave cash in, you end up with either a debtor or cash, but current assets are still the same (which can then be used to fund current liabilities)). That said, he's got the track record, so who am I to argue?!

I reached the same conclusion as you in terms of screening via ROA and P/E, which is (kind of) the pool I'm trying to fish in anyway - good quality companies (ROE) and low valuations (P/E) - with additional considerations on asset values, cash generation and insiders.

I'm planning to screen via Sharelockholmes too and then see how suitable candidates measure up against my Rules (soon to be tweaked). Will be interesting to compare notes with your findings.


If you want a sophisticated back-testing screen/sort system then seems to do a good job. The data covers the US over the last 10 years and they have screens built up to approximate 'gurus' like Buffett etc and you can build some quite details screens and sorts yourself. There's a free months subscription so if you're a tight-wad you can sign up and screen like mad for a month!

There are issues like setting your own approximate slippage/spread and dividends are not accounted for, but otherwise it's a good lesson in how back-test and how to data mine to find a formula that returns 50% PA. Of course what it returns going forward is anyone's guess!

They also did a blog piece on the magic formula: and how to 'improve' it.

But be careful with data-mining as you can make up any daft screen to pick up companies starting with 'fl' and a market cap between 100 and 120 million or something and get 100% annualised returns. That's like saying it rained on Wednesday more often than on any other day, so take an umbrella out on Wednesdays. It just doesn't work like that!

Hi Yorkiem,

Please steam ahead with your observations on the Little Book, I'm sure you have more and whether or not we agree with every fine detail in it the basic idea gets to the heart of investing in my opinion - buying good companies cheaply.

Good luck with your screaning, I wanted to comment on the ROE discussion but I kept getting confused about your model. Maybe one day I'll find time to sit down and scrutinise it!

Hi UKVI. Portfolio 123 looks interesting but I want to test in the UK market so will stick with good old Sharelockholmes, which judging by your research on ROE and Yorkiem's, seems to be becoming the default.

I agree with their verdict on the Magic Formula though: "simple on top; complex under the hood" and the complications they add, particularly the 5 year ROI, are very interesting. Shame they kept calling Greenblatt Greenberg though :-)

Thanks for linking to the article. Very stimulating read. is a good place for screening UK stocks based on Magic Formula principles.

I agree the ROC calculation is a bit uncertain, but I think how Sharelock handles excess cash is correct. ROA and ROE are fine too - I always tell people its the spirit of the strategy, not always the exact details, that will find the attractive stocks.

A MFI screener that details out exactly how the calculations are done is here:

Steve Alexander

does the magic site referred to above analyse uk companies?



Great article, Richard.

I have been using SharelockHolmes to generate some Greenblatt companies. If you set the market caps low enough, you can get some real stomach-churners like RCG.L (shady directors and business deals; share price that falls off a cliff) and HLO.L (suspended from trading recently).

Another company that looks iffy is green energy company EAGA that is in turmoil partly because the government changed its tarriffs. If you had bought at the end of November, you'd have more than doubled your money. If you had bought a year ago, you'd be down 22%. The business risk here was so great that one could hardly call it a "good" company.

I think HMV made it high on the list recently, too. The plight of HMV will undoubtedly be familiar to most readers, which one critic described as being in a "stage four meltdown". Ouch!

This just goes to show how SI ("strategic ignorance" - i.e. not looking under the hood) is a two-edged sword. On the one hand, SI can stop you becoming scared out of cheap and good companies with temporary bad news; on the other, SI can get you buying into some of the truly worst disasters imaginable.

Investing. It's not easy, is it?

Thought I'd add some additional comments ...

One idea I have been toying with, which seems quite reasonable to me, is to look for high ROE, low debt (maybe use a z-score>=3), and low PE.

I have also been considering using dividend yield as a substitue for PE. Yield has the advantage in that it is real, and tends to be sticky. Plus, you actually get to spend it!

Hi Steve, I wished I'd seen your MF calculator earlier! It's very clear how you calculate the formula.

I'm right in thinking, though, that Greenblatt has not revealed how he calculated excess cash? Rumour has it he uses different ratios of cash for different industries, though I've never seen that substantiated.

I like the concept of MagicDiligence.

Blippy, I had a very similar experience looking at Zetar (though it's not as big a basket case as the companies you describe) see:

It was investigating Zetar that alerted me to the risk in ROC, and the UK implementations.

I love your reference to Strategic Ignorance - it might form the basis of a future blog post.


[...] that some of the companies identified by a mechanical trading system turned out to be "real stomach churners": This just goes to show how SI (”strategic ignorance” - i.e. not looking under the hood) [...]

Just noticed today: RCG down 20% on proposed delisting from AIM.

It's always worth checking out the boards on Interactive Investor, particularly the Sell recommendations, as it can sometimes alert you to significant flaws in the company.

It is perhaps worth reiterating one of Whitney Tilson's comments: "when you see a company that looks like a bargain, there's a 90% chance you are wrong".

Looking back, it becomes apparent just how awful some of these companies are, just how ignorant we can be of the true worth of the companies we invest in, and how even the pros can make huge mistakes.

Also found that obtaining suitable screeners for the Uk to follow up suggestion from'The magic formula Book' a job to find with UK shares.
Just compare the Times screener ROC figures with those of Digitakl look or interactive; Only the latter two are compatible.

Hey there!

I am currently building a MF screener using CapitalIQ. I was wondering if anyone could compare their results with me. I have only done an initial year's allocation using the FTSE 350 (excluding financials and utilities), beginning 31-01-2014 and ending on 31-01-2015. Each month end I buy the two highest ranking companies using my MF screen and so on and so forth until I have 24 companies. I average ~7.6% return. The FTSE 350 averaged 1.7% in the same period. I have not included dividends. Can anyone roughly/quickly replicate the above using their models to compare?

Although woefully inadequate as a test to prove my model, I'd like to poke around out here to see if anyone is still active in the UK Magic Formula market and compare notes.



Hi Mark. Not testing MF on its own but in combination with Piotroski's F_Score in my Nifty Thrifty portfolios now five years old. You can read all about it here:

Post new comment

The content of this field is kept private and will not be shown publicly.
Type the characters you see in this picture. (verify using audio)
Type the characters you see in the picture above; if you can't read them, submit the form and a new image will be generated. Not case sensitive.