What correlation method should I use to analyze financial timeseries? I need to check if two price lists move togheter. I three methods:


*

*Pearson

*Spearman

*Kendall


Could someone explain me why I should use a method instead of other to check if those price lists move in the same direction? I know that Pearson is often used for this kind of analysis... Is that correct?
Thank you
 A: Pearson correlation is used if you suspect you have a linear relation in your data (if you think you can do simple linear regression on your data, you can also use pearson correlation). If you only assume any relation between two variables, you should use either kendall or spearman correlation.  I personally prefer spearman correlation, since I have a firmer grasp of it. It is essentially pearson correlation on the ranks of observations instead of the numerical values.
A useful tool to see if pearson correlation is appropriate, at least if you only have a few variables (say 10 or less): Plot them all pairwise, which most statistics software should have a function to do, and inspect if some obvious linear pattern is present.
The level of your data (ratio, ordinal, nominal etc.) should not guide your choice of correlation measure other than to clue you in that you probably don't have a linear relation.
A: The Pearson correlation is for interval/ ratio data and it is quite suitable for your problem. 
The Kendall correlation is used for non-parametric data ( If assumption do not full fill in your data)
The Spearman correlation is for ordinal data and I am sure this not a case with your data.
