One overlooked aspect of this site (judging from the "hits" statistics for various pages) is our statistics regarding "pairs trading".  The idea behind pairs trading is relatively simple:  you find two stocks whose movements are positively correlated.  That is, when one stock goes up, the other tends to follow.  You can use statistical methods like linear regression to best identify these pairs.  When one stock lags the other, a trading opportunity arises and a trader can buy the lagging stock, and/or short the leading stock.

If you open up our downloadable data, you'll see several columns of data related to pairs trading.  First, you see the actual stock we identify as the best "partner" for the stock you're interested in.  This column is labeled "best pair".  Often, the pairs make good intuitive sense.  As we write (10/11/2007), Exxon's movements are highly correlated with Apache's (another oil interest).  IBM's best partner is Hewlett Packard.  General Motors's best partner is Ford.  Other times, the relationships are obtuse, or nonsensical.  We have Coca Cola's best partner as "Deb Shops", an apparel outlet.   Even if you're not particularly enamored with a pairs trading strategy, discovering your favorite stock's best partner can be an entertaining exercise.

The next column of interest is "corr", which is simply the "r value" for a stock and its best partner.  A high "corr" value would mean that the two stocks are indeed tightly correlated.  A low value would mean that the two stocks are not strongly connected...they're simply the best connections we could find.  A stock with a low "corr" value isn't strongly tied to the behavior of any stock in our database.  We call such a stock a "free agent"...such stocks seem to move independently of the market.  Currently, the highest "corr" values are held by Fannie Mae and Freddie Mac, whose movements are, not surprisingly, intimately related.  The title of the "most free spirited" stock goes to "Synplicity", a software maker whose day to day movements seem to utterly ignore what is happening in the broad market.  These "free agents" are interesting stocks to pay attention to, particularly when you're searching for stocks that might resist a market debacle.

The "corr" value should be considered by serious pairs traders.  Obviously, every stock has a best trading partner.  If the "corr" value is low, however, the connection between the two stocks is likely to be a fluke, and it would be unwise to utilize a pairs trading strategy for such a stock.  "Synplicity", then, is a lousy candidate for pairs trading.

The next column is "pair dif".  This is simply the difference in the percentage gains/losses between the two stocks in the most recent market session.  For technical reasons, the loss/gain is limited to 5%, so you'll never see a figure greater than 10% in this column.

The next column, "pair dif20", gives you a slightly different perspective on a stock and its best trading partner.  Instead of looking at how the two stocks performed vs. one another in the last session, we look at how the two stocks have performed over the last 20 trading sessions (about one month).

Multiplying the "corr" value by the "pair dif" value could be useful.  Currently, we don't devote a column to this in our database, but it's a simple exercise for Excel users.  Here, you'd be identifying stocks that simultaneously have a high "corr" value and a big difference between gains/losses in the last session.  Stocks with a particularly high or low values would be prime candidates for a pair trading strategy.  One final exercise for Excel users would be to divide the various calculations above by standard deviation.  The idea here is that a large gap between two dowdy stocks might be more tantalizing than a large gap between two stocks that habitually jump 3% on a daily basis.

Scouring the internet, you'll see numerous resources regarding pairs trading...books, CD's, courses, and more.  These may or may not prove worthwhile.  In any case, though, you can find the raw data right here, and it's updated nearly every day!

Copyright © 2010 MarketSynopsis.com. All rights reserved.