Anyone do any backtesting on "time-based" exits (ie. stops)?

Discussion in 'Trading' started by athlonmank8, Sep 1, 2009.

  1. Was considering incorporating them into my trading. I'm trading profitably so I'm kind of in the attitude "if it aint broke dont fix it". The other side of me is really thinking it could make me a bit more consistent and less emotional.

    I will say this. I use cycles to trade and I really don't see why I shouldn't be using them. Time plays an important role on my entry so I figure if it's a factor on entry, it should be a factor on the other end as well.

    Need some input here......
  2. Figured that
  3. everyone here trades profitably

    the last loser seen was Ted Binkle, who lost 3.45 cents one year.

    Congrats on your success
  4. Nattdog


    Time based exits are one of the most logical approaches to testing entry signals.

    If you backtest a signal and find the results are not signifigantly different than your control group over defined holding periods or "time stops," this is a clear sign that the trading Signal is not valid.

    Some of my best short term trading strategies are managed almost exclusively by a time stop exit. A risk management stop and home run profit target is also used but they get hit infrequently.

    In terms of how they will effect your trading, If you use at time stop only when you are underwater, just remember It is going to give you more losing trades.

    Allot of "fade trade" strategies work better with a time stop.

    For example my short trades in July made money only because I use a time stop exit. If I had held out for some profit target or stop loss exit, I would have been run over.

    You need to look at the logic of your approach and see if the idea fits, and if possible do some backtesting.
  5. MGJ


    They work quite well when used in conjunction with a profit target exit and a stoploss exit. You give the trade a certain fixed amount of time to either hit the profit target or the stoploss; if not, an alarm clock goes off, you exit the trade, and use that portion of your "risk budget" to look for a new opportunity.

    A pretty decent test of long entry signals is:
    (1) profit target exit at (EntryPrice + X)
    (2) stoploss exit at (EntryPrice - X)
    (3) timeout exit at N bars

    By making the profit target and stoploss symmetric (same distance "X" for each), you ensure that profits result from goodness of entries rather than goodness of exits. The timeout prevents you from being trapped forever in a range.

    Short entry testing would be similar except minus and plus are interchanged.
  6. Nice to see a real topic in "Trading"...
  7. There was a time when I developed a nice breakout strategy. It was very time sensitive. From the signal, I only had about 2 seconds to get the trade filled. After that, I had about 10 seconds for the move to happen. Every second after that without a move dramatically increased the probability of loss.

    Not only that, but the loss would be hard to control and the reverse of the perceived breakout would happen and stops were useless.

    The strat carried about an 80% success rate, and I made really good money. Unfortunately it was based on a relationship that no longer exists. In fact it lasted for less than a year.
  8. Are you dense? Your posts are as useless as the paper I use to wipe my ass.

    Find something better to do.
  9. Thanks very much for the reply. I have some work to do I see. Very insightful.

    I appreciate it.
  10. Thanks as well. I never even thought about 1/2 of that to be honest. Glad I asked now.

    THe goal was to keep me out of whipsaws but now you mention it, I may have to use it on profitable trades as well. The idea was to keep the trade as "even" as possible over a period of time IF the stop hadn't been hit. In addition it'd be used to reduce opportunity cost as well.

    Yes it was initially meant to be a 'loss only' approach. Obviously there is a potential flaw there.
    #10     Sep 2, 2009