In Python: Backtesting An Intraday Mean Reversion Pairs Strategy Between SPY And IWM

From the NYC Contact so thanks to him:

Backtesting An Intraday Mean Reversion Pairs Strategy Between SPY And IWM

Join our latest newsletter to see how we do stuff like this



  • seberbach

    Reply Reply February 20, 2014

    IMO, many of Michael Halls Moore’s articles are an excellent resource for shared experience in the field.

    Note his reasons for liking Python:

    Scripting language capability!
    General purpose language capability
    rapidly expanding community with scientific/academic participation
    Interfacing and software integration with Linux Microsoft OS, other languages, data formats, etc. etc.

    Free download of Canopy, semi proprietary upcoming alternative to Matlab…

  • seberbach

    Reply Reply February 20, 2014

    Was following up on thread of your question: what kind of “apps” would double my profit?

    My answer is “software apps for diverse software and data integration”.

    Your current first answer is Matlab is the best, but is not affordable for everybody when needing the required toolkits.

    Python may be alternative starter path which is MUCH less expensive, and offers further paths to the likes of Matlab. For you to render further investigations and opinions thereon to your members?

    What packages are easy to add to Canopy Express to move it past a scripting app with primitive data flow canvas? (a for example question you could answer?)

    • caustic

      Reply Reply February 20, 2014

      In terms of expense of Matlab for my members, I have shown yesterday self containing programmable or self containing applications can be generated from Matlab thanks to the Coder and Builder NE toolboxes. Those cam be distributed royalty free to my Members which I plan to do. This could include DLLs, Excel add ins, or dare I say Java JAR files. I have already done the 6 month experiment with R which lead me to some interesting results but does not play nice with critical trading custom platform. From what I hear, Python is the same problem. Many of the R packages don’t play nice or aren’t properly vetted for live trading. My potential investors don’t like that. Also, I have heard various historical Python upgrades are not backward compatible which is another huge issue. I could go on and on about it but why bother. To answer your last question, I just plan to stick with Matlab generated components to be used in my custom trading platform done in .NET with C#. I just get stuff done faster and easier with this combo. Hope this helps/

  • seberbach

    Reply Reply February 20, 2014

    Another example idea just slightly out of that box:

    Is academic license “non commercial” for Canopy or Matlab?

    You have founded an Academy, online. These are becoming the new era college.

    Can you arrange for Matlab academic license for your students?

    • caustic

      Reply Reply February 20, 2014

      Thanks but no Mathworks does NOT discount at all. I tried and well was pretty laughed at. As for the expense of Microsoft product, I would look into BizSpark which gives you free developer editions for 6 months. Pretty sweet.

Leave A Response


* Denotes Required Field