Reasons for failur in automatic trading systems

Discussion in 'Automated Trading' started by cohvi, May 10, 2006.

  1. Don, you have had this anti-ATS attitude for as long as I've read you on these boards. If an ATS has to be tweaked or even intervened with occasionally I don't understand why this invalidates the system. Surely your discretionary trading has evolved over time, no? Old edges go away, new ones appear. Why do you view this as OK for discretionary trades but not automated ones?

    For me and the guys I know with automated systems it is simply an extension of our manual trading. If I can take a manual trade setup and express it with rules I can program into a computer I've now greatly increased # of trades I can execute. Don't you run an automated program for exiting your opening only trades? Doesn't it allow you to trade more names on the open if you chose to?

    I bet if you could put your computer at a Vegas blackjack table and have it play the dealer you'd do it in a heartbeat. Bet you'd do it even if you had to tweak the computer once in a while as the house changed the rules. Why not the stock market?

    Just curious about the contradiction.
     
    #171     Mar 26, 2007
  2. Actually, we may agree more than we disagree....perhaps the definition of ATS is where the difference lies. I use an automated program for my opening only orders. It submits orders, and when filled, submits retracement orders for (any variable) half (is what I use currently) the number of shares with a predetermined price gap from the opening. This is simply automating what I do manually, and I love it.

    The various scalper programs simply automate our "outside enveloping" strategies with greater accuracy. The various Pairs Portfolio Watcher programs allow us to engage in many more pairs than if we were to attempt to monitor manually.

    My experiences with those who have elaborate algo's, based on backtesting, have not been very good overall.

    So, I like automating good, working strategies (with tweaking, adapting, as you say we all have to do). I am not a big fan of the elaborate algo's that tend to look good on paper, but don't seem to work in real life.

    And, yes, my mind is always open to anything that makes good, consistent money.

    All the best,

    Don
     
    #172     Mar 26, 2007
  3. MarkBrown

    MarkBrown

    dont know many traders who could pull this move off ahead of time - but i know a system that was fearless and cold blooded - in the face of the unknown.

    i bet all the weak minded human traders were setting around the tv with flat positions wondering what the news would be. or they were over there in the chocolate bar daily trading journal drawing trend lines on the left side of the chart dreaming of perfect interpretation.

    [​IMG]

    mark brown the mechanical trader

    and author of the worlds most successful free fully mechanical trading systems published in numerous medias world wide. just do a search on et forum here for oddball system for one example. others are on this forum as well.

    http://i3.tinypic.com/2rqeu1e.gif
    http://i10.tinypic.com/44vqwyg.gif
    http://i7.tinypic.com/2wlyef8.gif
     
    #173     Mar 26, 2007
  4. I know that the intention of posting those charts were not to show off the system's performance itself but to show that systems can make trades that humans won't be able to. Though, I'll bite...

    I doubt that the system gets filled as the TS signals show. At least the signals can cover slippage, though. (1 lot, $10,000 in 300 trades... $33.33/trade)

    ===============================

    Anyways, people try to duplicate a human trading with computers, way too much. You have to let a computer do, what a computer can do. You don't need 2 traders doing exactly the same thing anyways, and laziness never gets rewarded in business.
     
    #174     Mar 26, 2007
  5. joesan

    joesan

    is the underlying strategy in the chart Oddball ? Very impressive. Mark I remember you had a homepage, where can I find it now ?
     
    #175     Mar 26, 2007
  6. It's not Oddball.
     
    #176     Mar 26, 2007
  7. MarkBrown

    MarkBrown

    oddball as i have it today. killin it i might add.

    http://i7.tinypic.com/3yq130h.gif

    is not the daytraing system i posted, different project completely.

    mark brown
     
    #177     Mar 26, 2007
  8. MarkBrown

    MarkBrown

    exactly the point - this was mainly posted for guys like bright to ponder why a human cant do that.

    good phrase let systems do what they do. you either build it right and it works or it dont.

    mark brown

    ps everything i have goes to market. - listen to this - goes to market when the system is one tick higher or lower than the order. so say i want to buy at even - the system waits till the price is .25 below even and then buys at market. this is how i do it.
     
    #178     Mar 26, 2007