Trading a portfolio of markets

Discussion in 'Trading' started by Hendrix, Sep 4, 2002.

  1. Hendrix


    Here's a question which I have been struggling with for a time. I'm hoping someone may be able to help me.

    Assume the following: Mechanical day trading (flat at the close) system, trading 4 futures markets.
    All 4 markets have virtually idential characteristics - profit factor, percent profitable, drawdown, etc.
    All 4 markets have the same amount of "equity" allocated to them, and similar absolute performance.

    Because, at any given time, one of these markets tends to be going through a drawdown (which typically lasts about 2 weeks, but there is a big variation) I would like to only trade 3 of them at a time.
    What I can't work out, however is how to choose which three.

    I have tried using moving averages on the equity curves, and I have tried ignoring the market with the current worst drawdown (while continuing to monitor this market until it is no longer the worst), but nothing I have tried improves the overall performance of the "portfolio". I refuse to accept (although I may have to) that there isn't a way of doing this.

    Anyone have any ideas? Any and all suggestion appreciated.
  2. Quiet1


    i tried various ways of doing this some time ago. i started with 150 eod futures markets and sought the best markets on which to trade a simple long term trend following system.

    i was suprised that all the obvious tries, MA of equity, Moving Sharpe ratio etc (best of, worst of etc) did not actually improve results. the only equity curve based indicator i found that improved results was to use the % of MA of the equity curve (instead % of the actual current equity) in determining how many contracts to put for the system's next trade. this smoothed out the curve just after those huge upspike periods that trend following systems are prone to.

    Delighted to know any method you find works :p

  3. tntneo

    tntneo Moderator

    Consider using the drawdown itself as a parameter.
    When drawdown increases near and past average value, reduce size (can be built in the system itself).
    Alternative I used before is reduction based on consecutive losers compared to average number and max number of consecutive losers.

    Full filtering (stopping the system totally quickly) usually yields lower results from the total system. That's why reducing size gradually is better to me. Eventually size goes to zero : that means the system is really dead for the time being. it happens.

    The only we found to improve results is to increase size on winning strategies and reduce size on losing strategies. Then you can get a positive impact on total equity curve.
    Some discretion is good to. When you build your own strategy you know why it works and it what context it might fail (either rationally or thanks to experience). Then discretion can help to reduce / increase size.

  4. Hendrix


    tntneo. Thanks, I'll do some thinking along those lines.

    Quiet. Sorry, I'm a bit thick (but I can lift heavy things!!). I don't understand "% of MA of the equity curve (instead % of the actual current equity)". Could you expand a bit on this please.

    One thing I have found that does seem to improve overall performance is ignoring the market with the worst equity curve over the past 50 days, but I am wary of this as I found it by trial and error, and it smacks of curve fitting.

    I throw out any trading ideas that I don't or can't first conceive logically, so I wouldn't feel particularly comfortable breaking this rule for the equity curve.