The Art of the Pairs Trade

About the kindest thing you can say about the Nasdaq right now is that it's in a trading range -- a big, broad one bounded by 3000 on the downside and 4200 on the upside. In an environment like that, the momentum plays just don't work. This is a whipsaw world where there's not enough room in either direction for stocks to run enough to make momentum trading worthwhile.

In times like these you need to play for inches rather than yards, and think about using trading styles that have built-in hedge characteristics. One of my favorites is the "pairs trade."

In a pairs trade, you match a long position with a short position in the same sector. That means that you're hedged against the overall market, and hedged against the sector that the two stocks are in. If the market or the sector moves in one direction or the other, the gain on one side is offset by a loss on the other.

So what's left is a pure bet on stocks: The "good stock" that you buy vs. the "bad stock" that you short. It's the ultimate strategy for pure stock pickers, because stock picking is all that counts in a pairs trade.

The classic pairs trade is between two companies that compete with one another -- you try to pick the winner. For example, we put one on Wednesday, between long F5 Networks (FFIV Quote) and short Cacheflow (CFLO Quote).

Both companies make equipment designed to speed up Web sites. Cacheflow makes a box that optimally caches, or stores, frequently requested content so that the site's application server doesn't have to start from scratch every time it's requested. F5 makes load balancers that distribute Web traffic between multiple Web servers, application servers and hosting centers based on which one has excess capacity at the moment the traffic arrives. I've had experience with both companies' products while running the Web site, MetaMarkets.com.

The reason for the pairs trade is that I think F5 has a chance of horning into Cacheflow's business big time. F5's new Edge-FX Cache box costs less than Cacheflow's -- it's about $10,000 compared to $30,000. The F5 box is more dense, which means it takes up less precious space in the collocation facility. Both F5's and Cacheflow's boxes are Akamai (AKAM Quote)-compliant, so they conform to the industry standard for caching.

F5 has a good chance of leveraging all of these advantages into a Cacheflow killer, building on its entrenched position as one of the leaders in load balancing. It should be fairly easy to up-sell or cross-sell. And Network Appliance (NTAP Quote) is trying to make inroads in caching, too.

Now Cacheflow is introducing a new denser, cheaper box that I hear is just starting to be delivered. But if nothing else, F5's competitive thrust into the core of Cacheflow's space will probably hurt Cacheflow's margins because it has forced the company to cut prices in order to compete. With F5 trading at seven times its sales, and Cacheflow trading at 60 times, Cacheflow's stock probably can't afford much margin erosion.

Anatomy of a Trade

Here's how I did the trade. On Wednesday, I shorted Cacheflow at $121.29, and we bought F5 at $32.93. I scaled the number of shares on each side of the trade so that the dollar value of the long was approximately equal to that of the short.

You don't necessarily want to do every pairs trade in equal dollar weights as I did in this case. Sometimes the stock on one side of the trade is far more volatile than the other, so the dollar values should be adjusted accordingly. For example, if the long is expected to be twice as volatile as the short, you might construct the trade so that you only have half as much money on the long side. It's also possible that as a stock picker you may have long convictions that exceed your short convictions -- after all, a hot stock in the right market can double in no time, but how many shorts ever actually go to zero?

But when the dollar values are equal, the pairs trade has a very seductive twist: It is self-funding. The proceeds from the short sale can be thought of, at least notionally, as paying for the cost of the long. That means any profits on the trade are earned on a notional cost basis of zero.

What are the risks in a pairs trade? Simply the risk of any investment: That you might be wrong. And in a pairs trade, there's a chance to be wrong twice. Imagine how we'd feel if Cacheflow goes up and F5 goes down at the same time?

That can happen. But most of the time a well-constructed pairs trade works as an effective hedge. In fact, there was a time when so-called hedge funds actually hedged, and this is how many of them did it.