The investor of today does not profit from yesterday's growth. - Warren Buffett

A form of investment that has always been popular with technical analysts but has recently started getting approval from fundamentalist investors and even the academic establishment is momentum investing.

There are two types of momentum:

This type of investment is the kind that causes the most irritation of all for efficient market theorists. Basically it means that trends exist and that they persist. Buy last year's highest flying stocks and you'll enjoy good results this year as well. Most investors who have been around a while will immediately recognise this as performance chasing and realise it won't work... but somehow it does seem to.

James O'Shaughnessey found evidence of price momentum in What Works on Wall Street, he found that if you bought the top 10% of large capitalisation stocks by relative strength (outperformance of the benchmark) you would have done extremely well, beating the index substantially over the next year.

He found the opposite situation also was true, if you bought the 10% of large cap stocks with the biggest drops in price over the last 12 months you would have been mauled over the next year, so much for bottom picking or contrarianism.

In David Dreman's excellent book on contrarian investment Contrarian Investment Strategies: The Next Generation, Dreman notes that analysts are generally too optimistic overall, but are very much too optimistic for "good" stocks. On the other hand they are frequently far too bearish on "bad" stocks and as a result stocks that have been depressed for some time are usually wonderful value plays.

Dreman notes that contrarian stocks perform well in their first year of being a really oversold contrarian stock, but continue to do well for an extended period of time, frequently outperforming the market for several years.

The reason why contrarian stocks outperform the market for so long, yet "good" stocks lag is because analysts are slow to upgrade or downgrade companies. Perennial favourites are held far too long and the dogs are kept out in the cold long after they have started to become very attractive value stocks. As a result "good" stocks have a tendency to repeatedly disappoint the market with a string of profit announcements below consensus forecasts, yet analysts remain bearish on "bad" stocks when they recover, so these stocks tend to happily surprise the market for an extended period of time when they do pick up, always being better than consensus forecasts.

Dreman discovered momentum by working backward, contrarian stocks exhibit momentum because the market is slow to upgrade its view on those stocks.

Louis Chan, Narasimhan Jegadeesh and Josef Lakonishok tested momentum strategies (see chapter 8, Momentum Strategies in The Psychology of Investing (Lawrence Liffson and Richard Geist (eds), John Wiley and Sons, 1999)) and found also that stocks that had outperformed continued to do so for at least a short period of time following their outperformance.

Examining every stock on the New York (NYSE), American (AMEX) and NASDAQ stock exchanges (but excluding foreign stocks, real estate trusts, listed managed funds and other derivative shares), using COMPUSTAT and the Centre for Research in Security Prices (CRSP), over a period of Jan 1977 to Jan 1993, excluding any other companies for which full data was not available, the researchers showed how stocks performed in a follow-up period from the time of them being in the top decile (10%) of rising prices, comparing the top decile with the lower deciles to see if there was a relationship.

In the next table, I summarise their findings with regard to 6-monthly past price momentum:

Returns

Lowest Highest
Momentum Decile 1          2          3          4          5          6          7          8          9          10        
Past 6 month return %  -30.8 -12.6 -5.5 0.00 5.0 9.9 15.3 21.9 31.9 69.6
Next 6 month return %  6.1 8.6 9.3 9.6 10.2 10.4 10.5 11.1 12.0 14.9
Next 12 month return %  14.3 18.5 19.8 20.8 21.4 22.2 22.3 23.5 24.8 29.7
Next 24 month return %  20.5 20.1 20.5 20.6 20.8 20.8 20.4 20.8 20.7 19.9
Next 36 month return %  19.4 19.6 19.7 19.6 19.9 20.2 20.5 20.1 20.8 20.6

Note that momentum is a relatively short effect, the biggest gains were over the next year, following on from that momentum doesn't seem to affect the stock after a year, the result is that "momentum" investors frequently need a very high turnover in their portfolios to take advantage of the effect. Most momentum funds I know of do indeed have portfolio turnovers of 100% or more, which only adds to their expenses and takes away from returns.

What sort of fundamental factors characterised the stocks that outperformed? Here, on the "purchase" date are the price-to-book ratio (PBR) and price-to-cash flow ratio (PCR) of the stocks that over the previous 6 months had done so poorly, or so well:

Value at "purchase"

Lowest Highest
Momentum Decile    1          2          3          4          5          6          7          8          9          10        
PBR 0.9 1.0 1.0 1.1 1.1 1.1 1.2 1.2 1.3 1.4
PCR 9.0 6.9 6.7 6.6 6.6 6.7 6.7 6.9 7.2 8.7

Even more interesting are the earnings surprises that put those stocks where they are today. It should not be too shocking that the stocks that fell so much become value stocks, one would expect a stock to fall to the point where it started to look like good value and hence ratios would be favourable for the ones that took the biggest hit in price recently.

Earnings surprises

Lowest Highest
Momentum Decile    1          2          3          4          5          6          7          8          9          10        
Earnings surprise last quarter %  -87.9 -33.6 -9.2 4.6 19.6 31.6 43.3 57.0 67.0 82.4
Earnings surprise next quarter %  -105.2 -41.4 -14.7 -3.4 19.2 35.0 47.9 61.3 74.4 91.9

So on average the market did not adjust to the previous earnings surprise, as often the market was surprised twice. If the market had really cottoned on to the fortunes of this company when the first surprise had come along, the second surprise would have been very low, however it wasn't, often the market was given exactly the same quarterly earnings surprise again, and for some reason it was "surprised"!

Abnormal returns as a result of earnings announcements

Lowest Highest
Momentum Decile    1          2          3          4          5          6          7          8          9          10        
Most recent announcement %  -2.7 -1.3 -0.7 -0.3 0.0 0.4 0.7 1.2 1.8 3.5
First announcement after purchase %  -1.1 -0.4 -0.1 0.0 0.2 0.3 0.4 0.6 0.9 1.5
Second announcement after purchase %  -0.2 0.0 0.0 0.1 0.1 0.3 0.3 0.3 0.5 0.8
Third announcement after purchase %  0.2 0.1 0.2 0.1 0.2 0.1 0.3 0.3 0.3 0.5
Fourth announcement after purchase %  0.3 0.1 0.2 0.1 0.1 0.0 0.1 0.2 0.1 0.1

Last, one might look at how the analysts revise their forecasts for stocks.
Lowest Highest
Momentum Decile    1          2          3          4          5          6          7          8          9          10        
Most recent revision %  -2.190 -0.576 -0.401 -0.262 -0.212 -0.127 -0.129 -0.028 -0.003 0.086
Average over next 6 months %  -2.138 -0.578 -0.368 -0.282 -0.220 -0.152 -0.117 -0.068 -0.041 0.004
Average from months 7 to 12 %  -1.843 -0.555 -0.378 -0.318 -0.248 -0.206 -0.191 -0.165 -0.153 -0.180

And what about earnings momentum? The above shows price momentum, the researchers also looked at trends in earnings per share. You can look up the article yourself, either by finding a copy of The Psychology of Investing or by looking at the researcher's original paper in The Journal of Finance 51 (no. 5), December 1996. You can visit a university library and find these articles, a librarian will help you find the article since you'll find whole bookshelves or even whole rooms get filled with thousands of volumes of the larger journals, it took me a while to figure it out in my undergrad days so I would imagine that anyone who hasn't experienced trying to find an article in a big journal should seek professional help.

To save you the trouble though, the pattern is fairly similar, with companies showing strong increases in earnings having good performance over the next 6 months and the next year, but fairly ordinary results two years on. Analysts forecasts were even more pronounced though, it is obvious that analysts are very slow in catching on when a company is really changing from the past norms.

A third set of tables shows the combined effect of price and earnings momentum, with extremely high returns found for stocks with high earnings and high price momentum, serious underperformance for stocks with low earnings and price momentum. Similarly analysts took a while to re-assess stocks correctly, being serially positively surprised every time a stock had high momentum and continuously negatively surprised by low momentum stocks.

Does any of this mean contrarians are wrong? Does this mean one should go for expensive high flying stocks instead? After all, the stocks with the highest momentum seemed to be in the most expensive bracket when bought and they outperformed following the purchase. Doesn't this directly refute what Dreman says about value stocks outperforming? Well, yes and no. It seems to me that this is another piece of the puzzle, and that both Dreman and the momentum guys are onto something. Dreman talks about stocks that have been dogs for quite a while and earnings expectations are low. Dreman does specifically caution against charging in on value stocks while they may still be falling, suggesting that you get in with patience and only select quality, never buying stocks that the analysts say are likely to make continued losses.

The following is my interpretation of things, presented merely as a comment tying together the results of Chan et al, Dreman and O'Shaughnessey. Chan and his coworkers found that falling stocks are bad to buy, rising stocks are good to buy. Dreman found that cheap stocks are good to buy, expensive stocks are bad to buy. O'Shaughnessey found that cheap stocks are good to buy, expensive stocks are bad to buy, falling stocks are bad, rising stocks are good, and if you buy cheap rising stocks you do even better, whereas if you buy expensive falling stocks you do extremely poorly. I see a pattern!

Many of the stocks that Dreman says you should not buy are the nouveau dogs, companies falling from grace. Companies on their way down are going to become cheap stocks, but it doesn't mean they become value stocks. What we are looking at here is probably a small slice of the whole.

The momentum guys look at all the stocks that fall and find that they are cheap once they fall. Cheap stocks that have recently fallen are poor value. They find that the cheap stocks with negative momentum are bad to buy.

The contrarian guys look at all the cheap stocks, while some cheap stocks have recently fallen and are bad, a great many cheap stocks have been down there for quite some time and their returns are so good as to make a diversified portfolio of value stocks perform well. They find that when you buy expensive stocks you will suffer as a result.

O'Shaughnessey found that you can combine both, and his value plus growth (Low price to sales ratio, high relative strength) was in fact the highest performing strategy of all those that he tested. To me it is not difficult at all to reconcile the different viewpoints and come up with what would seem to be a very odd combination, but one that O'Shaughnessey has proven works superbly.

You've heard of GARP (Growth at Reasonable Price) as the mishmash of growth and value strategies. Well if you put contrarianism with profit chasing in the same cup and give it a good shake you get high momentum contrarianism.

The studies I mentioned above usually were referring to momentum over a one year time frame. How do stocks perform after a more prolonged period of underperformance? There have actually been several studies that have looked at this question.

Werner F.M. DeBondt and Richard Thaler, Professors at the University of Wisconsin and Cornell University, respectively, examined the investment performance of stocks with the worst and best prior investment results in "Does the Stock Market Overreact?", The Journal of Finance, July, 1985.

Rather than one year momentum, they looked at the subsequent performance of portfolios formed out of the 35 worst and 35 best performing companies selected each year based on previous five year performance, reforming the portfolio annually on 31 December from 1932 to 1977.

Their conclusions were different to conclusions reached by the other studies mentioned above. They found that based on five year momentum you are better off buying underperformers than outperformers.

The 35 stock "dog" portfolio outperformed the market benchmark by 12.2%pa, compounded, and the 35 stock "glamour" portfolio underperformed the market by 4.3%pa.

The study was repeated for stocks based on three year past performance and the results were similar. You are clearly better off buying long term bad performers than outperformers, though if you reconcile this with the other research it seems you could do well by buying stocks with a rotten five year past performance and a decent one year past performance.

Always buy good value stocks, but you can improve on plain old value investing by choosing stocks with a rotten five year past performance and sitting back and waiting for them to come up for a few months instead of trying to "fish for the bottom".

Dreman showed us that value stocks outperform for an extended period of time, so you won't miss the boat by waiting for them to start moving up again. Mechanically you might do this by buying stocks that have a miserable long term record but a very good short term record, or you might pick the highest momentum stocks with the cheapest valuations. It seems to me to be an excellent method for buying cheap stocks while staying clear of the ones that richly deserve to be sold.