Designing and Building a Profitable Automated Trading System

Hershey seems to think "chop" exists. Here is his chart from this thread with areas cleary annotated by him as "minor chop".

http://www.elitetrader.com/vb/attachment.php?s=&postid=2213677

 

See attached

The attachment is an annotated code sheet. I annotated with underlined bold italics in paragraphs.

If anyone is coding the work here using the logic illustration and Scott's XLS sheet then you have the references in the once over lightly that I did.

I am not going to do any more followup on this piece of coding since it goes so far astray AND to improve it would take a programmer such as Scott to put the fix in.

Scott when he recovers from the accident will arrive here and we will work further as I have indicated.

 

As spyder indicated in SCT trading there is no chop. This thread came about from Scott's indest in programming using indicators.

As previously seen in the failure of the SPM mechanical signals generated "chop" was a severe handicap that defeated profitability.

Several kinds of chop are documented by CW PA and CW indicator traders. They have needs to avoid deterioration of their P/L.

Scott and I agreed to shovel some coding into these pot holes for such people.

Spyder is correct on not seeing chop in pool extraction of its applications. It is not there because of the algorithms operating principles.

Chop is there for you and it is possible you do not have the workarounds for chop. Thats how it goes.

 

We'll see if you have such a workaround in due time..........

 

First of all, I hope you are feeling better soon after your accident ScottD.

Second of all, thank you for posting the actual dollar amount generated from the Jack Hershey ATS that you're building here.

The system is currently showing a loss of $913 (I rounded up), and along with all of the hind sight charts being posted, I can see that the trend of generating losing trades has continued.

Please keep us updated an informed as to whether the latest batch of curve-fitting and over optimization is having a positive effect on the equty curve.
***
Oh yeah, <strike>I heard it through the grapevine that</strike> bwolinsky redid the code in WealthLab and is currently on his way to generating untold riches from it. His success might have something to do with the "lookback" feature designed into the backtest ... LOL.

I guess the Jack Hershey Spirit of Christmas visited him early.

 

I hope you feel better ScottD.

 

Happy new year all. It’s very good to be back. Still in recovery mode, and improving steadily.

Based on Jack’s guidance in this thread, I’ve taken the ATS code to the next level, as follows.

New functionality:
- Hold logic.
- Non Stationary Data Export Module.

New enhancements:
- Suppression logic.
- Pace calculated in lookup table for non stationary window.
- Awayness calculated in lookup table for non stationary window.
- Trades now execute "next bar at open" instead of "this bar at close."
- Changed last entry time to < 1545.
- Changed Volume Suppression to evaluate current bar 'or' 1 bar ago.
- Better organized code structure for natural reading.

Future planned functionality:
Articulate each formula and incorporate into the proper logic area:
- MACD entwine.
- MACD crossover (C2D) and inflection (D2C).
- MACD convergence and divergence.
- Binary volume vector.
- Sync.
- Bar Overlap.

Attached is the chart from Dec 10, 2008, which is our standard focus date for diagnosis and for comparing before and after effects of code changes.

In this release of the ATS, the Suppressor Logic keeps us out of a lot of Entry/Reverse signals, perhaps too aggressively. The Awayness component of the Suppressor is new and therefore something to examine more closely. The chart shows the 30th percentile Awayness value as at Dec 10. It is no longer tuned to rockets. At the moment Awayness looks at the histo but perhaps it should look at one of the lines.

 

I created my own Pace and Awayness lookup tables for non stationary windows.

Later I discovered Vorzo's very nice version, which is based on Mak's original work.

I enhanced Vorzo's tool to work for Awayness.

Attached is the Awayness table as at 10 Dec 2008.

 

Attached is the Pace table as at 10 Dec 2008.

 

ScottD,

Interesting work - thanks for the report. Are the test results improving as a consequence of these additions?

Does 'awayness' represent MacD histogram > 0?

Your new pace table is in line with the NT 'autopace' results - did you every find out why TN has lower levels?

& congrats on the recovery-

- palinuro