Auto trading newbie

Discussion in 'Automated Trading' started by iancand, Apr 29, 2009.

  1. iancand


    I am finally underway using my first fully automated trading setup with my existing broker TD Ameritrade's Strategy Desk software, I am sure it is not the greatest but it gets the job done for my needs right now. I performed backtesting on 10 pre-canned strategies based on Stoch, EMA, RSI in Strategy Desk on list of US listed ETFs ( about 729 of them) with a 1hr bars over the past 1 week, 2week, 1month, 2 month, 3month, and 1 year. Strategy Desk is painfully slow at the backtesting and drives me crazy! I ending up chosing the strategies and ETFs to be used in my live setup from the those that were near the top from all backetsing timeframes and had a greater the 60% winning rate and a greater than $2500 avg. monthly profit (based on $10k trades).

    Been using this system live for 2 weeks now decent success (at least better than I would have manually done), about about 80% winning rate and ~$3.5k in realized gains thus far trading with $28k used for trading. So it is close to historical performance. I will continue to test every week to evaluate ETFs and strategies over various backtesting timesframes to make adjustments on system.

    1) What is wrong with using the pre-canned strategies if they work?
    2) Does automated trading really have to be so difficult as some make it out to be?
  2. My career was in IT for 35 years before I retired to trade full time. I wrote some of my first trading programs in OS/390 Cobol in the early 1980s. Then moved my data to DB2 in 1990s and added SQL. When Omega Research came out with software in 1990s I was one of their first customers. I have been programming in Easy Language ever since. Now all my strategies are automated on Tradestation and support me.

    Over all these years I learned 2 weeks is nothing in trading a strategy. If you are able to make it work in all kinds of markets for extended periods of time then you have something. The 2 weeks you made money may be a simple extension of a curve fitted optimization. A market change can quickly erase those gains and put you in the hole wondering what you did wrong.

    Pre-canned strategies often do not tell you the rules if price direction changes or volatility changes. This is crucial to understanding what you are doing. You must know under what conditions your strategy makes or loses money!

    Automated trading is difficult because most traders have no idea when a strategy must be stopped for fine tuning or discontinued because markets have changed.
  3. Occam


    1) Nothing is wrong with pre-canned strategies per se, but it's often hard to know if they work unless you use it for a long time. And even then, things can change, so be careful :D

    2) No and yes. "Some people" make it out to be infinitely hard, i.e. impossible; but of course it's not that hard. At the same time, it's certainly not easy, in general; for it if were, the market would be so heavily auto-traded that there would be no more opportunities to profit.
  4. Iancand,
    Good to hear your positive results and Strategy Desk backtesting with several formulas is trying. I posted this in another thread about speaking with the Strategy Desk Techs and wondering if you tested according to what they told me as far as 1minute interval check box and close close for buy and sell?:
    Finding a formula for auto trading isn't easy! Thought I had one that backtested great but found out my using the open open parameter should be close close and 1 minute check intervals used and the results are now horrible. May have to go manual in the real world for now. The tech told me about folks who don't call them first and start auto trading to find hundreds of trades executed in a day or two. Sometimes a hundred in an hour. He said the calls are all painful.
    Seeing how incorrect backtesting parameters can lead to great results and finding out in reality it's the opposite, I feel their pain.
    I now wonder if this quest is a waste of time or if there really is a formula than can be consistently profitable? Now I'm totally off track...back to the drawing board.

    I am trading manually now and haven't backtested since. I am with you on the 60 minute interval time frame and the RSI Stoc as they are the top 2 that I go by manually. Still buying a day or two too soon though but that will hopefully change soon. Perhaps its best to wait until the uptrend resumes rather than trying to catch absolute bottom.
  5. iancand


    -Rabbitone I too am skeptical as it has only been two weeks but we all have to begin somewhere.

    I was used to doing the regular homework for fundamental analysis investing but now with auto trading exclusively focused on the technicals what type of homework should I be doing to optimize the system?

    The first week I used the strategies with alerts and manually approved the auto filled orders, after getting comfortable seeing some success I set it to full auto in this second week.

    I havent found much success backetesting or trying shorter timeframes with the pre-can Strategy Desk strategies. But I certainly have been and will try and find something with a shorter timeframes I see that I am missing out on many opportunities.
  6. Back testing is only part of the answer. I perform extensive forward testing to eliminate curve fitting.

    If forward testing is not available to you than long periods of paper trading may give you an idea of how your strategy is performing in various market conditions. In this paper trading period the strategy must maintain a reasonable expectancy through out.
  7. RScott


    Interesting, i am thinking about doing the same with Strategy Desk. If you are using an intraday system, how much discrepency you do find between the backtested results and the live results?

    I find that my system has to be rewritten for it to work live.
  8. Canned programs are usually as static as you can get and when the environment or volatility change than you are going to see a different sloping curve than you see now.

    I understand you are starting out but this business is all about the hours and mistake you make and if you have capital after that than you are better than 80% of those that try ATS.....good luck and look into trading a portfolio of systems to cut down risk and also learn about correlations of your systems against each other and you are ahead of the game at that point.
  9. iancand


    I am pretty certain that I am giving up on fundamental trading and so rather than paper trading the new ATS I am running it live. And then reviewing results with backtesting on a weekly basis to see reality vs test system. I try various periods and stocks to try an prevent curve fitting but to focus on system optimization.
  10. iancand


    Strategy Desk is definitely not built from the ground up with Automated trading in mind, it seems like an after thought that was thrown in. I find anytime you modify your strategy you you will need to manually approve the trades at least once for each security buy and sale before it is fully automated, really really annoying. But if I can prove that I can be successful on this crummy platform after a few months I might put some money down into a real software platform Ninja trader/tradestation/etc.

    I threw the pre-canned strategy out the window over the weekend and started live on Monday with a new approach I worked on over the weekend as my first attempt to develop something myself. With the new system I will have 2-6 trades/day and after 2 weeks I should have some realistic comparison between backtesting and reality results. But backtesting as well as comparison over the live testing for last two days certainly proved that switching to my home brew approach outperformed the pre-canned approach I was using previously. I have moved on from that and wont be looking back.
    #10     May 5, 2009