That is a great quote. I know that correlation is not the same as cointegration, I also know that you can have cointegration without correlation...what I don't understand is how you can have correlation without co-integration. Of course a stock is perfectly correlated to itself, but not co-integrated (because the returns would never deviate obviously), but what I mean is for practical purposes - if, say, an ETF is 97% correlated with one of it's component stocks, how can it not be cointegrated? Somebody help me out here!
When people talk about correlation, they usually mean "return" correlation, not "price" correlation. So it's quite possible that returns are correlated but the prices deviate substantially. I find gummy's website a very good reference to understand some basic concepts: http://www.financialwebring.org/gummystuff/cointegration.htm http://www.financialwebring.org/gummystuff/spearman-correlation.htm
Hey Matt, About the hedge ratio: Chan includes an example of how to calculate it in his book using Matlab. I just want to know if I were to use excel to calculate this ratio, would I want to look at the coefficients from the linear regression of the stock prices? Thanks
Hi Matt, I notice that Chan has a few live spreads for pairs of ETF on his website. I wanted to ask you if you have tested out these pairs of ETF through MATLAB and as a trader in the Stat Arb field would you put on the trade especially for the pair of ETFs with a big Z-score. what did you do to test the cointegration for the paired ETFs? I just started using Matlab and I have been inputing the codes that Chan has place in the book in particular example 3_4m but Im getting an error message "??? Error using ==> xlsread at 189 XLSREAD unable to open file IGE. File C:\Users\bt\Documents\MATLAB\IGE.xls not found." If you could help, that'd be great. thanks
You need the file IGE.xls in the folder C:\Users\bt\Documents\MATLAB\ I attached the file. You may need to change the code for the dates (mm/dd/yyyy to mm-dd-yyyy) You can PM for other bugs (to ge it to run but no further!) Good luck
That was quite brutal. I read most of it and yes, there are very good insights. A usefull read but you need some theoretical finance/math background to read it. I now understand how pairs trading can be done intraday with black boxes. Putting the traditional pairs trading (correlation based) as an antique technique. Still usable as an art like technical analysis but not blindly. Chang chapter about cointegration was also of help.
Yes. If you look into the Matlab code, the hedge ratio is simple the slope of the regression line. So in excel you can just add a trendline to the regression and set the intercept to zero. I have verified this with Chan's example of GLD/GDX.
yeah, its a good starting point. finds correlation and back tests based on deviation. I wouldn't recommend using it alone to trade. I use it to run vast numbers of stocks for correlation sometimes