Designing and Building a Profitable Automated Trading System

Discussion in 'Automated Trading' started by ScottD, Dec 9, 2008.

  1. ScottD

    ScottD

    Signal-to-Noise ratio tool

    Technology enables some effective tools.

    Here's a tool that improves ET's Signal-to-Noise ratio:

    S/N Tool
     
    #81     Dec 11, 2008
  2. ScottD

    ScottD

    As suspected, the volume suppressor, which first appeared in v0.2, did not allow an entry at that time.
     
    #82     Dec 11, 2008
  3. ScottD

    ScottD

    Updated Trade Log with Pace Shift and Volume Direction logic.
     
    #83     Dec 11, 2008
  4. Nice spreadsheet but you didn't total the profit column. It appears to be a $913 loss. Seems kind of important.
     
    #84     Dec 11, 2008
  5. Neoxx

    Neoxx

    Keep up the good work.
     
    #85     Dec 12, 2008
  6. ScottD:

    You may want to change this so you enter on the open of the following bar. You can't enter on the close of a bar based on signals derived from the closing bar price (because the bar has already closed).
     
    #86     Dec 12, 2008
  7. ScottD

    ScottD

    Attached is the updated Trade Log with cleaner notation. You can delete prior logs if you like.

    The log contains the following comment:

    In Stochastic(14,1,3) formula notation, we see the "Middle Parameter" is 1. In Technical Analysis from A to Z by Steven B. Achelis the Middle Parameter is the smoothing length for SlowK. However, the MultiCharts formula appears to be a non-standard implementation because it treats the Middle Parameter as the smoothing length for FastD, instead of for SlowK. There is related discussion above in my Stochastic post that starts out "Topic for my own clarification..."

    In the attached table, generated from MultiChart's formulas, this is why Stoch14 FastD equals Stoch14 FastK. In this case, FastD is "smoothed" by a moving average of 1 period (i.e. it is not smoothed at all).

    So, the question is: In Stochastic(5,2,3) notation do we intend for Stoch5 FastD to be smoothed by a period of 2 according to MultiChart's non-standard implementation...or by a period of 3 according to my understanding of the standard notation as described by Achellis?

    We can use the column data to validate the data and determine our smoothing preference.

    Four columns are grayed out because the data is not in use at the moment.
     
    #87     Dec 12, 2008
  8. ScottD

    ScottD

    Fair comment, and I see what you're saying.

    At a future stage of development and testing, we plan to move from 5min close-of-bar decisions/actions to intrabar (tick-by-tick).

    That's why I have the code setup to operate on a "this bar" basis. I believe I need to maintain a "this bar" construct in order to preserve the (future) ability to make tick-by-tick decisions/actions. So, I have it setup for "close of this bar". Fortunately, the close to open gap is usually just a tick (favorable or unfavorable), if any. In reality, it probably nets out to nil.

    I appreciate the thought provoking comment.
     
    #88     Dec 12, 2008
  9. ^
    I got your Hershey 5.0 script to work on 15 minute bars in ES, but it had horrible results just for 1 contract. About $3000 made for risking $11,000 over a period of time.

    I'd just quit here.
     
    #89     Dec 12, 2008
  10. Good stuff ScottD. If I might suggest, whenever you require a clarification about some technical point just ask Trader666. LOL.

    On the off chance you didn't find that approach to be fruitful, this summary is from an earlier thread on the use of these oscillators which you may find useful (or redundant or superfluous).
    http://www.elitetrader.com/vb/attachment.php?s=&postid=1781944

    lj
     
    #90     Dec 12, 2008