To Cointegrate or Not to Cointegrate

When introduced to the concept of pairs trading, or if you want to impress others, “statistical arbitrage”, many individuals come across a term called cointegration. Many traders, including myself, first assume a good pair is identified through correlation. This is not always the case. For a better explanation with pretty graphs, read this.

A common method to test for cointegration is the Augmented Dickey Fuller and a simple Google search will yield many Matlab and Excel spreadsheets as examples. However are these tests even needed? Let’s assume that we identify a pair that is cointegrated but when we apply our trading rules we find that this pair is actually unprofitable. Taken another step, lets assume we skip the cointegration test all together and identify a number of pairs that back tested profitably, can we assume cointegration?

I would argue that cointegration, like many econometric techniques, are useful in theory but not necessary practically. A back test of pairs over a long enough period that includes earnings, fundamental shocks, different volatility regimes, etc, should be sufficient for the home gamer to develop a portfolio of pairs. If you agree with this line of thinking you should save significant time identifying pairs and can spend more time on portfolio construction and sizing pairs appropriately.

A quick story, when running a stat arb book at the previous hedge fund, Fannie Mae versus Freddie Mac and Bear Stearns versus most of the other brokers were highly profitable for us. It was only due to a DISCRETIONARY call during the financial crisis that saved us from catastrophic loss in those pairs as our models were screaming buy for BSC and shorting GS, JPM, CS etc against it. For myself that was a lesson in learning what drives volatility and movements in stocks. It is ultimately not the numbers or a quantitative variable, it’s people, outliers, and behavior. Quantitative analysis is an extremely useful tool but should not be the complete answer to financial trading.


~ by largecaptrader on February 26, 2011.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: