Discussion in 'Strategy Development' started by clarodina, Feb 16, 2011.

  1. Anyone know how to compute cointegration? Baby steps please. Anyone good element tutorial on this topic with illustration data?
  2. Trader13


    This link may be a good place to start:

    You might be asking this question because cointegration is the basis for a mean-reverting pairs trading strategy. A truly cointegrated pair (not just a pair that exhibits cointegration as a statistical coincidence) is quite elusive. Most of these stat arb strategies are tough to accomplish with any consistency. Suggest you start by trading very small positions until you satisfy yourself that this approach works or not.

    There are a some superstars on this forum who may be able to offer more guidance on this approach. You may attract their attention after conducting some testing and sharing results.

    Keep charging!
  3. How do u interpret the cointegration coefficient? Whether is significant or not significant? What statistical table is use for the significant the t-statistics?
  4. wot


  5. Check-out GRETL.

    Their manual is interesting and contains numerical examples.
  6. This guy got a couple post on his blog:
    He also wrote a book where he has some explanations and matlab code.
  7. Took a look at Ernie Chan's blog, but it doesn't have much detail. Yes, there is a cointegration package in R which can be gotten to programmatically, but I would use GRETL first.

    Chan does reference the book, Alexander, Carol (2001). Market Models: A Guide to Financial Data Analysis. John Wiley & Sons.
    I have this book, and have found it quite valuable. One of these days, I will buy some of her more recent works.