What method gives best chance for profit?

Discussion in 'Automated Trading' started by Sam Mcgee, Oct 23, 2007.

  1. horribilicus

    horribilicus

    If "best chance of success" means "largest probability of winning trades", just make sure your stop loss is 10 times farther away than your profit target. Presto, 85% winners.


    I also think you would enjoy reading the poetry of Robert Service, for example: http://www.potw.org/archive/potw22.html

    Now Sam McGee was from Tennessee, where the cotton blooms and blows.
    Why he left his home in the South to roam 'round the Pole, God only knows.
    He was always cold, but the land of gold seemed to hold him like a spell;
    Though he'd often say in his homely way that "he'd sooner live in hell."

    On a Christmas Day we were mushing our way over the Dawson trail.
    Talk of your cold! through the parka's fold it stabbed like a driven nail.
    If our eyes we'd close, then the lashes froze till sometimes we couldn't see;
    It wasn't much fun, but the only one to whimper was Sam McGee.
     
    #11     Oct 31, 2007
    Share
  2. fatrat

    fatrat

    I use tick charts as well, but I feed them into a smooth/transformation kernel. The problem with most brokers is that they'll either send snapshot data, or rebroadcast UDP packets. Not only that, but if a trade executes on a dark pool, it doesn't have to be reported for a full 30 seconds. What this does for Joe Q. Trader is that it can cause the system to ignore events of significance due to error on behalf of the broker, or falsely create events of significance.

    Using a transformation that minimizes the effect of substantial variance on time-series data prevents systems from making faulty trades.

    For example, suppose you're trading a support/resistance system. Then all of a sudden, a million share transaction that happened on a dark pool and then got reported via ISLD or ARCA shows up. What is your system going to do? It's going to be a complete idiot, assign significance to that information, and make a bad trade.

    Trading is a bastard of a profession. Every little detail or corner case is out there waiting to make your anus bleed.
     
    #12     Oct 31, 2007
    Share
  3. Prevail

    Prevail Guest

    mark, what is the best direction to go to nail down a breadth based system? thanx, good seeing you here.

     
    #13     Oct 31, 2007
    Share
  4. thetrendfollowe

    thetrendfollowe

    Thanks for your offer.

    But my systems are always well tested.
    Naturally, one of the (many) tests for robustness is to compare the results against those taken randomly.

    The difference was significant and showed that the system clearly had an edge.

    All the details about how i tested my system are on my blog.

    http://thetrendfollower.blogspot.com/
     
    #14     Oct 31, 2007
    Share