Would you trade this?

Discussion in 'Trading' started by BloodTrader, Dec 2, 2003.

  1. Would you trade a mechanical system with an equity curve like this (see attachment)?

    These are backtested results, except for the last year, which are real time.

    This is actually a combination of two systems, both long-only.

    I've been trading them intermittently for the last year, but I can't seem to bring myself to commit to trading the system! I keep breaking the rules. And so, my results for the last year are very mediocre.

    I can go off and trade discretionary without much angst, but I usually lose. The mechanical systems are going gangbusters and I can't see to stick with them.

    I post here because it definitely seems to be a psychological problem.


    y axis is % gain
  2. What instrument(s) are these systems trading?
    If more than 1 instrument is traded, how many?
    How rigorous was the backtesting done and was it on the traded instrument itself or some "cash" index?
    How many rules?
    Is it a daily system? Intraday?
    How many trades does it do?
    How much optimization was involved?
    Why do you think you can't follow it?
    Likely a sign that you don't really believe in the logic involved, and are afraid the system(s) will crash with the next few trades.

    Also, instead of showing the profit in % terms, show a graph with the actual dollar amounts.
  3. DK_


    Really? did you include slippage and commissions? Did you code up your system haphazardly, with bugs giving you such big results? If not, trade that s*, it looks great.
  4. I would trade it .
    If for some reason you are not comfortable to trade it on your own, find trader ( not broker )who will trade it for you . If you are interested, pm me and I will explain further .
  5. I agree with the above posters. As long as your code is covering all the bases (commisions, is backtesting exiting / entering correctly, not biased, etc).
    For you to ask about that system - that, to me is ludicrous. That curve is great. I believe it comes down to the curve must fit what the systems developer personality is looking for. I look for consistant systems, something that will (seem to ) offer steady return. That is why that curve appeals to me.
    If this is an all out mechnaical system, than I strongly suggest you let it run, putting risk parameters in, and shut the monitors off, so you let it run its life on its own. Thats what I would do.
  6. Both these systems draw signals from a large stock universe. Many stocks, but few signals. Periods of increased acvtivity alternate with periods with few signals/trades.

    I've been backtesting for 6 years, but two years on these two systems. THe systems have very little optimization. I had the idea, selected what seemed to be reasonable parameters and modified them just a little, mainly little tweaks to the entry signal parameters and holding period and exit signal. They are not at all complicated.

    I think I can't follow it because no matter how much backtesting and real-time post backtesting results I see, I always think that it will fail as soon as I start trading it. The system does have draw-downs, not huge overall, but some invdividual trades can have large losses. So money management requires not trading to much on any one trade.

    The logic is simple and no reason it should quit working, but the market is strange and anything can happen. If I could get data going back 40 years, I believe I'd see these systems still perform, but I can't be sure until I can really test them.

    I can show a graph with dollar amount. Will try to get to it tomorrow.

  7. For each round trip trade there is 1% commission and slippage. I've been trading IB and I can usually match or better this. Once in a while I do worse, but not often. I have gone over this code very carefully and have found no bugs. I have experience finding bugs in other systems where things almost always looked to good. But I've found no such things here and the systems are simple to check.

    I would have had a great 2003 if I had just traded these systems. Now after this good run, I expect there may be a flat period in the equity curve. So, I wait.

  8. I'm reluctant to give away the code...even though it is nothing special or complex. What I would really like to do is develop an add on for TWS that would automate all the trading. I'm sure it's doable, just not sure I can do it.

  9. I really wish I could find a way to completely automate the trading. I could generate the list of potential trades in the evening and then enter the symbols into a spreadsheet or some custom interface.

    The code would need to enter a trade based on a limit order x% below yesterday's close. Once the trade was entered, it would need to be monitored. If price closed at least 5% above the entry price then the system would send an MOC order.

    That's pretty much it.

    Should this be doable in TWS API or Excel?

  10. DK_


    What? answer only this: how long is your median hold time?

    "These systems" sounds amibuguous.. make sure your one grand system of systems is clearly defined and scientific.
    #10     Dec 4, 2003