The shares set to bounce back

On 1 October each year, buy the 10 worst-performing stocks in the S&P 500 index using their price performance over the previous three years. Hold these shares for only three months until 31 December and then sell them. And that's it. You don't even have to pore over the finer details of fundamental analysis used by stockbrokers and equity analysts when deciding on the 10 shares to hold. So, forget about dividend yields, price-to-earnings multiples and price-to-book values (the ratio of a company's market value to its net asset value). This strategy does not rely on any of these valuation measures to work.

And work it certainly does: if you had followed this strategy in the past decade, you would have turned in an average quarterly gain of 23 per cent. That's over 15 percentage points more than a S&P 500 index tracker made in the final three months of the year.

So why does this trading strategy of buying the worst-performing stocks in the S&P 500 index work so well?

Shares overreact to news

In a now-famous paper* published in the 1980s, academics Richard Thaler and Werner de Bondt found that portfolios consisting of the 35 worst-performing stocks in the S&P 500 (using price data over the previous three years) outperformed the 35 best-performing stocks by an average of 25 per cent over the subsequent three years for each three-year period between 1933 and 1979. They noted at the time: "Most people overreact to unexpected and dramatic news events. And you can make big money by exploiting this."

For example, some companies get a bad reputation for perennially disappointing and, as a result, both shareholders and potential new investors are more inclined to ignore the few merits the company and its management have. In the most extreme cases where share prices of the worst performers in the S&P 500 have fallen by over 90 per cent over a three-year period, as has been the case for the majority of the 10 stocks included in these dog portfolios in the past, this savage derating can take valuations way below fair value. So why does a policy of buying on 1 October do so well? Why not pick any other date? Fortunately, there are sound reasons why the policy of buying the worst-performing stocks on this specific date works so well.

Window-dressing

The reasons why the dogs of the S&P 500 start to bounce back on 1 October is easy to explain: the US fiscal year ends on 30 September. At this time, US fund managers must send reports to their investors detailing their performance during the year. However, the last thing they want to put in their reports is that fact that they are holding some of the worst-performing shares in the S&P 500. It would hardly inspire confidence in their stock-picking ability if shareholders in their funds found out that they had taken big hits on some of the Rottweilers in the index.

As a result, in an effort to hoodwink their own shareholders that they're better stock-pickers than they really are, the asset managers sell these dog stocks before the fiscal year-end. Other fund managers, for the same motives, are reluctant to buy them.

The upshot is that loser stocks are likely to be especially undervalued at the end of September and so are ripe for bouncing back. That's when the 'buy-the-dog' investment strategy kicks in.

There is certainly some merit in this explanation, although it does assume that investors are stupid enough to be taken in by this window dressing ruse. Moreover, it also assumes that less savvy investors haven't learned that stocks overreact on the downside in this way. Otherwise they would simply buy the loser stocks, which would push their prices up, and so make it impossible for other later investors to make money from them.

*Source: 'Further Evidence on Investor Overreaction and Stock Market Seasonality', Werner F M De Bondt and Richard H Thaler, The Journal of Finance, Vol. 42, No. 3, December 28-30, 1986 (July, 1987), pp. 557-581 and "One Step Plan", Simon Thompson and Chris Dillow, Investors Chronicle, 3 October 2003.

So, even if we accept that the window-dressing ruse exacerbates the downward pressure on share prices in the months leading up to 30 September, it is very unlikely that it can be the only reason why stocks then perform so strongly in the following three months. Instead, there's an another explanation. It's all to do with risk. Stocks that have fallen by 90 per cent or more in the past three years carry lots of risk. There are five types of risk:

• Volatility. Stocks that have fallen by 90 per cent or more can fall by another 90 per cent - that's basic mathematics. And the fact that they have fallen that far is evidence that they are more volatile than most stocks, and so have more chance of falling another 90 per cent - that's basic statistics. So these dog stocks are likely to be more volatile than the average constituent of the S&P 500. Bear in mind, too, that shares can be just as volatile on the upside when bouncing back as they are when falling.

• Liquidity risk. The 10-worst dog stocks in the S&P 500 have low absolute prices. That often means they have bigger bid-offer spreads (the difference between the prices market makers offer to buy or sell at) than most shares. As a consequence, it costs more to trade them. This in turn means that, if things go wrong and investors are forced to sell, this could be very expensive indeed. But liquidity risk works both ways. On the downside, it depresses share prices and sometimes forces them below fair value. But, to the upside, liquidity risk falls as prices rise thereby offering scope for above-average price rises when these stocks start to bounce back.

• Distress risk. Dog stocks that plunge by 90 per cent or more run a far greater risk of going bust. For one thing, they usually carry much higher levels of balance sheet gearing - the ratio of net borrowings to the net assets of the company - than the average constituent in the S&P 500. In some cases, the bank covenants on the debt will be related to the market capitalisation of the company, so the further the stock falls the greater the risk of a breach of these covenants. Moreover, investors clearly sense this as distress risk tends to be an increasing factor in the downward share-price momentum seen in poorly-performing stocks. However, if investor sentiment improves and the perception of a company going bust or breaching its bank covenants diminishes - which is likely to be the case if the stocks start to rise strongly - then distress risk falls, which in turn helps the bounce back in the share price.

• Market risk. The fact that the 10 worst-performing dog stocks have fallen so much at the same time - in most cases the falls have been far greater than falls on the S&P 500 - means that they have a high sensitivity to market moves. This is another source of risk and one that helps our dog stocks rise faster than the market when they bounce back.

• Economic risk. Dog stocks are generally in cyclical sectors that have in the past done well during winter months. This is because winter is a dangerous time for the economy. Academics have estimated that half of the ordinary business cycle is the result of seasonal swings in output around Christmas time. Cyclical stocks, including dogs of the S&P 500, offer high returns in winter to compensate investors for this risk.

Common sense tells us that risky shares should outperform other shares eventually, simply to compensate for their greater risk. And the combination of the US fiscal year-end, window dressing by fund managers and the start of a seasonally good time to be holding equities - the S&P 500 index has risen by 4.4 per cent on average in the final three months of the year since 1950 - all helps these risky stocks to outperform in the final quarter.

Risk and Reward

The dog stocks may have their day in the final quarter of the year, but don't expect the recovery to be long lasting. It used to be the case that you could just buy the dog stocks on 1 October and hold them for a year to reap even bigger returns. This certainly worked between 1996 and 2003, as the table below shows.

However, in recent years investors trying to bet on sustained share price recoveries by holding the 10 dog stocks for a full 12 months have been bitterly disappointed. In fact, in the 12 months to end-September 2004, 2005, 2006 and 2007 this strategy would have lost money. Even worse, in each of these years the S&P 500 recorded decent gains - so the 10 shares performed very poorly in a rising market.

However, an analysis of all the dog portfolios since 1996 has one thing in common: there is a clear bias for the best of the gains to come in the four-month period between 1

October and 31 January.

For example, if you had bought the 10 worst-performing stocks in the S&P 500 (over the previous three years) on 1 October 2003 and held them until 31 December, you would have made a very healthy 27.1 per cent return. However, if you had held on another three weeks until 20 January, you would have more than doubled your return to 59.5 per cent (see table below). And, remember, that return was over just 16 weeks. It's worth noting, though, that this was as good as it got as the 10 dogs started to bite and not just bark for the rest of the year and ended 30 September 2004 down 4.9 per cent for the 12 months.

It was a similar story a year later. The 10 dog stocks that year stormed ahead, notching up a hefty 34.1 per cent gain in the three months to end-December 2004. However, by 30 September 2005 the 10 stocks had given up all these gains and ended the year down 2.6 per cent overall, dragged lower by the dire subsequent performances of Delta Airlines and Winn-Dixie Stores, both of which went bust.

In 2005 the 10 dog stocks that had performed so dismally in the previous three years roared ahead once more in the final quarter of that year, buoyed by double-digit gains from the likes of drug giant Merck and insurance group March & McClellan (see table five). The dog portfolio was up by 15.5 per cent in the three months to 31 December 2005, handsomely outperforming the meagre 1.5 per cent gain on the S&P 500 index. But, as in 2004, this was a very good time to bank profits with the 2005 dog portfolio ending the 12 months to 30 September 2006 down 1.4 per cent. This was not helped by Dana Corporation entering Chapter 11 Bankruptcy Protection in March 2006, which again highlights how the bumper gains being made are a reflection of the high risks these portfolios carry.

In 2006, the dogs fared less well, rising 5.5 per cent in the final quarter of the year against a 6.2 per cent rise in the S&P 500 index (see table below). However, this did maintain the strategy's momentum - for 10 consecutive years, the dog portfolios have made a positive return in the final three months of the year.

Trading Strategy

It's worth remembering that dog portfolios only offer the prospect of substantial returns because they carry significant risks, as demonstrated by the bankruptcies of Delta Airlines and Winn-Dixie.

That said, there is a clear bias for the worst-performing stocks in the S&P 500 in the three years to end of September to start to bounce back in the next three months after recording this dire performance. So, if you can stomach the above-average risks associated with this type of trade, buying our 10 dog stocks on 1 October with the intention of banking profits three months later is the advised strategy.

This year, the 10 S&P 500 dog stocks to buy at the start of October are: SLM (TIDM: SLM), American International Group (AIG), MGIC Investment (MTG), National City (NCC), Office Depot (ODP), E*Trade Financial (ETFC), CIT Group (CIT), MBIA (MBI), Advanced Micro Devices (AMD) and Sovereign Bancorp (SOV). These are large-cap corporations listed on the New York Stock Exchange with market values ranging from as low as $900m (£489m) to as high as $67bn. However, they all have one thing in common - in the past three years their share prices have fallen by at least 75 per cent.

Also, don't forget that the reason that our dog portfolios have performed so well in past years is because they carry above-average risk. A quick glance at the 10 stocks above reveals beleaguered insurers, banks holding sub-prime mortgage books and monoline insurers who have significant exposure to default insurance on these loans.

However, what makes this trading strategy of particular interest this year is the fact that the US stock market looks as if it is setting itself up for a strong fourth-quarter performance ('' and '', 26 September 2008). If I am right, then you would expect heavily-sold stocks in the S&P 500 to bounce back more than the general market - which should be good news for the performance of these 10 dog stocks.

One note of caution, though: I don't believe the bear market has hit its final bottom yet. Therefore, any recovery in the stock market, no matter how strong, is only likely to be yet another counter-trend rally. So, in the circumstances it would pay to take profits at the end of December, rather than chancing your arm and running these positions into January, as it is likely that any bumper gains will quickly disappear thereafter.