SERVING THE QUANTITATIVE FINANCE COMMUNITY

Topic Author
Posts: 78
Joined: July 13th, 2011, 1:01 pm

Dear all, The pair trading for two stock is easy to understand. For instance, the two stocks have very high correlation(they rise and fall together), so we can do the pair trading for these two stocks. But if I have three stocks in the same sector, and they have very high correlation, how can I do the "pair trading" for these three stocks ? Or if I have n stocks in the same sector, I want to do the same thing, how? Many thanks

Jeriot
Posts: 97
Joined: January 16th, 2007, 11:54 am

If you just want to work in correlation, you can literally take one time series and do an ordinary least squares regression on the other time series, which will give you the weights of the individual assets. But that is just a variance reduction technique, which increases transaction costs and reduces trading opportunities. You might want to look up cointegration, which takes time dependency into account, which should increase trading opportunities.

blueraincap
Posts: 80
Joined: February 21st, 2013, 3:04 pm

generalized regression

chocolatemoney
Posts: 322
Joined: October 8th, 2008, 6:50 am

QuoteOriginally posted by: JeriotIf you just want to work in correlation, you can literally take one time series and do an ordinary least squares regression on the other time series, which will give you the weights of the individual assets. But that is just a variance reduction technique, which increases transaction costs and reduces trading opportunities. .. but bear in mind the impact on asymptotics of unit roots.

AnalyticalVega
Posts: 2260
Joined: January 16th, 2013, 5:03 am

QuoteOriginally posted by: JeriotIf you just want to work in correlation, you can literally take one time series and do an ordinary least squares regression on the other time series, which will give you the weights of the individual assets. But that is just a variance reduction technique, which increases transaction costs and reduces trading opportunities. You might want to look up cointegration, which takes time dependency into account, which should increase trading opportunities.Good Luck! OLS is a linear estimator and markets are in general non-linear. Cointegration Arbitrage P&L generally does NOT do so well in out of sample testing.

Topic Author
Posts: 78
Joined: July 13th, 2011, 1:01 pm

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...

 JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...

GZIP: On