Secrets Of Success: Will the magic formula work again?

Click to follow

Of the various stock and fund screening systems that I track on a regular basis, the high-yielding portfolio selected a year ago on the basis of a variant of the "Dogs of the Dow" methodology once again proved a big success in 2006. It certainly worked a lot better than the so-called "magic formula" proposed in the bestselling investment book by the American academic Joel Greenblatt.

Of the 10 UK stocks with the highest scores on the latter formula a year ago, no fewer than seven have seen their share prices fall over the subsequent 12 months. This is hardly a ringing endorsement of the methodology, even for one that is advertised as being a long-term strategy that only works for sure over a period of three to five years.

If you took the top 20 stocks, rather than the top 10, the average price gain was 1.7 per cent, still a poor outcome in a year when the FTSE index rose 12 per cent. Eleven of the top 20 stocks identified by the method made money, but some of those that went down in price went down a lot.

There are, of course, always dangers in trying to transpose methods that appear to work in the United States to another country; and this is particularly so in the case of a methodology whose two criteria - return on capital employed and p/e ratios - are subject to accounting judgement. There is also the possibility that the minimum market capitalisation criterion (£500m) I chose to adopt was too low.

One year's results in any event hardly invalidate a method whose underlying principle - buying established stocks with high returns on capital and low earnings multiples - is fundamentally sound.

It may be no accident that three of the top 10 shares that the method identified a year ago were media stocks, which have had another poor year, but which Anthony Bolton, for one, thinks are now due to follow telecom and technology stocks out of the doldrums in which they have been for several years. His fund is heavily overweight in media stocks.

In other words, the relatively poor performance of the Greenblatt formula in 2006 might be in part a timing issue, as well as a methodological one. The 20 stocks that score highest include a significant number of energy and mining companies, whose return on capital figures has risen steeply on the back of rising commodity prices.

Unlike with most conventional businesses, with their more predictable earnings growth, the value of resource stocks is heavily dependent on uncertain future price assumptions.

2006 was a year when the new bull market in commodities showed clear signs of flagging and clear signs of speculative excess in some specific markets. While I expect the secular bull market in commodities to resume in due course, the sector as a class is unlikely to produce the same vastly superior returns over the next five years as it has done in the past five years.

If you look at a list of best-performing fund sectors over the past five years, natural resources (with a total return of 261 per cent) and gold and mining (254 per cent) are the two runaway leaders, followed by property (139 per cent) and energy (100 per cent).

Readers will remember my customary warning that sector leaders over five years rarely, if ever, feature among the leaders in the subsequent five-year period (though they may continue to do well for another year or two in the mean time). Those with the best 10-year records never do.

While I have been tracking the magic formula results, to see how effective they are at picking timely winners, it is not the way that I put my own money to work. In my view, there are better ways to find the cheap but reliably profitable stocks that are at the heart of Greenblatt's approach. The Dogs of the Dow is one variant on this theme, except that the primary emphasis here is on dividend capacity (a more reliable indicator than earnings).

As more than one reader has pointed out, there is more than one way to apply the theory to the UK market. The table shows the list of qualifying companies as of 1 January, based on applying the following method, using data from Digital Look. First, take the 30 companies in the FTSE 100 index with the largest market capitalisations. Of these, you then take the 10 with the highest dividend yields.

This is the basic method, but it can be refined further by taking only the five stocks in the 10-stock list with the lowest share price (the original O'Higgins method) or (as Jim Slater and Tom Stevenson suggest in their version of this method) the lowest market capitalisation.

The table shows the results of all three, as well as the results from widening the market cap criterion to include the largest 100 UK stocks. Five of the stocks score on both methods.

In recent years, given the strength of value stocks in general, simply scanning the Footsie for the highest-yielding stocks and applying some additional criteria (such as positive cash flow, dividend cover, gearing and p/e ratios) has served investors very well. The five out-of-favour high-yielding stocks that I identified in this way a year ago and have tracked since have had a remarkable year, comfortably outpacing the market as a whole.

It will be interesting, as always, to see how the high-yielding list of shares performs in 2007. It is worth emphasising that no method of this kind can ever work every year. The fact that we are in the later stages of a maturing bull market adds another degree of uncertainty to the exercise.

The list this year includes a number of telecom and bank stocks that I would say qualify as buys on other grounds (size, cheapness, mean reversion); in that sense there are fewer than normal "dogs" in the list.

Whether this is a helpful coincidence - or merely another case of the market laying a nasty surprise in store - remains to be seen.

The great comfort of a high-yielding stock remains that you are being paid to own it, unlike more exciting but riskier alternatives, and if you reinvest the dividends, compounding can do wonders for your longer-term returns.

Looking for credit card or current account deals? Search here