Building an ATS - Logbook

Discussion in 'Automated Trading' started by tiagor, Dec 5, 2012.

  1. tiagor

    tiagor

    Sorry, I'm afraid its not open source for the time being. I haven't even considered that option until you mentioned so I may give it more thought and reconsider in the future. Your interpretation on the concepts of SHELL and GHOST is correct though.
    Python is indeed a nice language to quickly build a working prototype and its data structures lend well to most kinds of mathematical manipulations. Do you have experience coding ATSs in Python?
     
    #11     Dec 5, 2012
  2. Will you be observing other's systems in real time?

    MBT has a way for you to observe other accounts in real time. It could generate ideas for your different ghosts' useage.
     
    #12     Dec 5, 2012
  3. 2rosy

    2rosy

    Yes, it takes about a few days ... or less
     
    #13     Dec 5, 2012
  4. CT10Gov

    CT10Gov

    The shell is uninteresting. Most people on ET spend faaar too much time talking about it. I suspect it's because ET guys are more software development in background rather than quant/finance.
     
    #14     Dec 5, 2012
  5. quatron

    quatron

    I guess this is true for any language if you have proper infrastructure. At least it takes similar time with Java for me.
     
    #15     Dec 6, 2012
  6. The shell (OMS) is the difficult part, especially if your ATS generates complex multi-leg, algorithmic, and/or contingent orders that, once filled, need to be dynamically hedged. Dealing with partial fills, hanging legs, etc can be convoluted and error prone. I have never seen a non-pro OMS that gets these things right.

    Frankly, I am skeptical of claims that an ATS, including functional OMS, can be coded up in a matter of days.
     
    #16     Dec 6, 2012
  7. Exactly.

    It's a never ending story of building f**king boring oms infrastructure.

    Zero edge, with lots of fancy "software architecture".

    Yawn...
     
    #17     Dec 6, 2012
  8. I kind of agree, in so far as managing a stack of orders is tricky, especially skewing and re-sizing without losing priority -- but -- none of that is sufficient to make any money.

    If you can't do the quant modelling work to figure out a fair value to quote around, then don't even bother to build the oms tech masterpiece.
     
    #18     Dec 6, 2012
  9. CT10Gov

    CT10Gov

    OMS is technically challenging to do right, especially on a fund level. It *can* be trivially easy if it's done on a small scale and just enough that everything works for the strategy at hand.

    The GHOST is intellectually hard.

     
    #19     Dec 6, 2012
  10. 2rosy

    2rosy

    thats what I was getting at. I meant that for personal use infrastructure (especially in python) is not difficult and can be done in a few days. Nothing needs to be fast considering you're going over the internet and most likely have a slow feed.
     
    #20     Dec 6, 2012