The Flaw in Trading System Reasoning...?

Discussion in 'Strategy Building' started by Remiraz, Dec 14, 2004.

  1. That's probably true ... but most people will be unable to trade most of them.

    Most systems severely test your manhood at some point and true grit is in short supply.


     
    #41     Dec 19, 2004
  2. But "automated" systems have to be used like football players on a pro team you own. You have to continually measure and evaluate their performance...if after a good period of evaluation and adjustments (to your standards) you have a player that is not meeting your level of expectation in play, then they are out....next! Just like a team of pro players, you might have some systems that are superstars and some that at least get the job done (keep you in the win column every week), but if you have a system that can not be motivated (optimized/curve fitted....LOL) then that system needs to be benched and cut from the team. Also, to keep your team fresh and talented you need to keep doing your scouting work for drafting in the new talent (building and testing new systems...practice squad material). In the end, there will always be a way to keep your team on top, and that is directly related to your level of expectations and required performance for your team.....just my 2. :)
     
    #42     Dec 19, 2004
  3. you got it!
     
    #43     Dec 19, 2004
  4. Remiraz

    Remiraz

    Where you get your figures from? :D

    Because if u have a static set of data in your backtesting, there are only a finite set of rules you can used on it that produce profitable results.

    For example, take the 3x3 Rubik's Cube.
    As long as no new blocks are added and it remains a 3x3, there are finite number of positions that the blocks can form.

    Just like '95-'00 data, there are finite number of systems that can be profitable over that period and that period only.

    In the case of new data however, the new data doesn't change old data like new blocks open up more positions in a rubik's cube.

    Whatever system that works on the new data, have to work on the old data too.
    Unless you remove the old data from the backtest then new systems that did not work on the old data but work on the new data would be consider "profitable".
     
    #44     Dec 19, 2004
  5. YES...exactly...incidentally, isn't the power of compounding marvelous?


     
    #45     Dec 19, 2004
  6. What I want is just one, a profitable one. :confused: :mad:
     
    #46     Dec 20, 2004
  7. Yes, I am looking for the link to post. Just wanted to be forthcoming with my sources as quickly as possible.
     
    #47     Dec 20, 2004
  8. Again, if you assert that the amount of systems is finite, what is the number of finite systems? Clearly, in order to determine that the number is finite, you must have an idea how many profitable systems there are....will you please tell me what that number is.

    I am trying (hard) to see validity in your belief, and I can't. Please give me something to go on...please......

    Maybe you could also make a list of them, so some traders will know if their system isn't on your list, they should just stop working completely. In fact, I want to see your list so I can see if the systems I use are profitable or not.

    Even though I created my systems that I trade, I am sure if they are not on your list my profits must be an accounting error. Please help me.

    But alas, you have no list that contains all the profitable systems that exist. And, to follow that, you don't have the number of profitable trading systems.

    So in that respect, how does it feel to keep making the point that there are a finite amount of profitable systems, when you have no evidence, not even logical reasoning, to support that belief?
     
    #48     Dec 20, 2004
  9. #49     Dec 20, 2004
  10. Remiraz

    Remiraz

    If a tree falls in a forest and no one is around to hear it, does it make a sound?

    Okay, for example :

    You restrict your backtest data to only 10 days and only the opening and closing price of these 10 days.

    Lets say in these 10 days, the open is always lower than the close (uptrend).

    The number of systems you can get to show profit within these 10 days would be finite.

    A Rubik's Cube can have (8! × 38−1) × (12! × 212−1)/2 = 43,252,003,274,489,856,000 different positions (~4.3 × 1019), about 43 quintillion.

    A huge number, but still a finite number.

    Now, did the researchers have an entire list of these positions? Have they even seen all the possible positions to determine the above?

    Its logical reasoning that for a STATIC NON-CHANGING set of data, there is a finite amount of system that can be profitable.

    If you use only '98-'04 ES tick data, there are a HUGE but finite amount of systems you can get to work on it.

    Just like there are only 43 quintillion positions u can twist a rubik's cube into.

    Even if you add new data, the systemss that work on the new data must work on the '98-'04 set too and thus these systems will fall under that finite amount.
     
    #50     Dec 22, 2004