Universal Systems Problem

Discussion in 'Strategy Building' started by pap, Nov 21, 2003.

  1. pap

    pap Guest

    I am somewhat new to system trading/developement and was wondering if anybody has any ideas or insight into the what I call the "Great Close of bar" dilema. Maybe I am the only one that has this problem but I could use some advice. If I am running a condition on a bar and during the open of that bar I get an alert do most of you take the alert or wait for the close and the chance that maybe a signal was never genratate or do you wait for the close and possibilly miss some upside. I realize that all backtesting involves the close of bar so yes you should have to wait for the close of bar but I see some good trades being missed.

    I read this in SFO magazine and I like Edward Valesquez's approach but I realize it is might be a little out of my realm at the moment

    Most traders attempt to build a system around three or four “canned” indicators (indicators that come with software programs). What they soon may realize, however, is that these indicators are lagging. One common problem is having to wait for the bar to close before a signal is valid; for example, a weekly bar isn’t complete until the end of a given week. The market turn may have occurred on Monday, but a trader has to wait until the close on Friday to be sure that his system will give the signal. To
    counteract this problem, Velazquez has moved completely away from the canned indicators. He instead has developed indicators that are price driven and key off of mathematical relationships between various price points.

    Like I said I am not quite there to key off mathmatical levels and I am not sure that is the way I want to trade but i do know the close is causing me problems and lowering my time frame is not the answer.

    Thanks in advance
  2. Not only you have to consider the close but also the open of next period and that also can degrade the theorical performance of your system since when the market has closed you cannot take position until it opens and since market gaps sometimes you see the problem :D.

    As for me I wait for the opening of next period, but unlike many other systems I don't wait for 30 minutes (some systems even wait for 1 or 2 hours) or even 15 minutes I wait only for 1 minute :D. In fact I look less at the close of this 1st min bar than the high or the low of the bar.

  3. Just 2 cents:

    For very short time frames, simply just wait until closing.

    For relatively long time frames, perhaps better take extra actions, imo should be no more than one extra action during a bar, during the current bar based on any pre-defined critical values (due to breakout, reversal, etc.) which could be determined by various ways according to personal preferences and market conditions, and review the extra action's results for further decisions at the closing.

  4. The key is in the architecture (in general I don't talk of mine which is evidently "architectured"): if you want to take into account only one scale there is much chance that you will have some difficulty to fine tune your system.