How Long Before A successful AT system

Discussion in 'Automated Trading' started by August, Jan 22, 2007.

  1. August

    August

    I've been discussing my efforts for building a trade system with a couple people more than others and it dawned on me that both people say that it took about five years before their program worked so that they were very happy. It was about 3 years before it was useable.

    I'm wondering if that is similar to most people's experience?

    The question here would be (for people who are trading real live systems and consistently making money from them only please):

    How long did it take before you had coded an automatic trading system (bot) which earned consistent income? (Not when it was perfected.)

    thank you!
     
  2. asap

    asap

    How long did it take before you had coded an automatic trading system (bot) which earned consistent income? (Not when it was perfected.)



    three months is my usual development cycle for any of my hi freq bots. that includes planning, programming and testing the concept against millions of historical price records and also against computer simulated price data. thats a full time job for three persons with considerable computer power. low freq bots could be programmed in a couple of hours but i guess those dont deserve being mentioned because they're just based on typical TA analysis, such as support and resistance, etc.

    it all depends on your gear and your skills.
     
  3. jonnyman

    jonnyman

    It took us about 6 months for our first one, and our second one only took 2 months. After you've developed the infrastructure it's a lot easier.
     
  4. Bowgett

    Bowgett

    I second to that. About 6 month to prepare infrastructure from scratch. Then you just build modules on top of that.
     
  5. t1ck3r

    t1ck3r

    The trading strategy itself should only be the very top level application. Business opportunities come and go as the markets present them. Sometimes only lasting 3 months. I find the successful traders often stick to a particular style of trading and vary the technical triggers as the market changes. The ability to keep order status, process quote feeds, and manage a presentation layer (user interface) is a commodity in our business that companies like mine offer up for subscription. Some strategies within our system make up only 200 lines of code. And took less than 3 days to code. Others took much longer but mostly due to logic design not the coding. When in production our traders need at least 30 trades in varying market conditions to get a good idea if its going to be a revenue producer. Tweaking enty and exit parameters as market volatility changes. Hope this helps with your business plan....greg
     
  6. August

    August

    All of this information is helpful Much more optomistic than three to five years as well.

    Are you guys (and anyone else who answers this) mostly writing your own code from scratch (or from open source) or using popular applications like amibroker, tradestation, strategy runner, or neoticker to develop in?


    thanks!
     
  7. jonnyman

    jonnyman

    In between - we're using an API and building our machine based around that. So not completely from scratch, but not an existing 3rd party app either.

    Three to five years is laughable. Unless you're just coding in your spare time, if something takes you that long you should probably learn to code better or have someone else do it. Strategies should also not take very long - if it takes you more than a few months to perfect it you risk that fundamental market changes will happen between when you conceived the idea and it's implementation.
     
  8. asap

    asap

    Are you guys (and anyone else who answers this) mostly writing your own code from scratch (or from open source) or using popular applications like amibroker, tradestation, strategy runner, or neoticker to develop in?

    More effective to build on top of an existent platform. trading algorithms could be extremly complex but its implementation should be simple and transparent. even though custom built will give the best flexibility, the associated costs, eg maintenance and upgrade, etc, etc don't pay off in the long run. either you are in the business of trading or in the business of producing applications... as they say "jack of all trades master of none".
     
  9. moise

    moise

    The problem is that most traders will not give enough time for a system to work ..........

    Example ; after having created all bunch of trading systems with lots of BS settings ........... I came to a conclusion that your basic moving averages are the only things you need.....

    Pattern recognition works (sometimes)
    Support and resistance work (sometimes)

    A simple rule .........like a long term moving average (exponential), works all the time .........

    If I had to tell you that you had to stay out of the markets for 6 months because you had no buy signal ; can you handle staying out for 6 months ? I doubt it.

    Yet, if you do all the testing, you will quickly notice the huge gains in just using a simple rule ..........

    Do not waiste your hard earned money on bogus "system vendors" out there ..............

    Use a simple rule ; and try also using a simple rule regarding the fundamentals of what you plan to trade ;
    ie ; I prefer to buy a stock with a very low PE (giving me a buy signal)
     
  10. Kohanz

    Kohanz

    Only true if you go it alone.
     
    #10     Jan 23, 2007