Hi all,
I have been reviewing pair trading lectre prepared by Delaney Mackenzie and Maxwell Margenot (thank you, it is very informative). I re-performed all codes and data sets in my notebook. When I run coint function from statsmodels.tsa.stattools. my results is completely different from the file results and I know that my p-value is wrong. I run stationary test using adfuller and it make sense. Any idea what should be the problem?
When I run code on Quantopian notebook, the results are same as the "Introduction to Pair Trading" notebook. but when I run same code on my own PC the results of the coint fuction is totally different for same dataset.
I use python 3.6, could it be the reason?
here is the test:
score, pvalue, _ = coint(X,Y)
print (pvalue)
Lecture result : 2.75767345363e-16
My PC results: 0.683824295029396
Thank you for all your help.
M