View detailed analysis

StrataSearch has many powerful features, with many techniques and approaches to explore. Here we discuss the things we've done, and the things we'd like to do.

View detailed analysis

Postby eleven » Tue Jan 29, 2013 12:44 am

I am working on a strategy that (after processing for a while) gives a result of "1portAPR" at 78.24%. When I run and and view the detailed analysis, the performance tab does not match up. I looked at "fixed", "variable", and "trades" report under the performance tab and nothing matches the 1portAPR of 78.24%. The "5portAPR" (or the rest) doesn't match as well. I've included a couple pictures to illustrate this.

I have run the Moving Average Crossover Example strategy and everything matches up fine. Please let me know what I'm doing wrong.

Thanks.
Attachments
detailed.png
detailed.png (16.18 KiB) Viewed 2753 times
results.png
results.png (6.17 KiB) Viewed 2756 times
eleven
 
Posts: 5
Joined: Sun Jan 27, 2013 6:34 pm

Re: View detailed analysis

Postby Overload » Tue Jan 29, 2013 9:36 am

The Combination Results xPortAPR values should match the Trades Report, but not necessarily the Equity Reports. In the top left corner of the Detailed Analysis, change the Report selection to Trades Report. Assuming the trade amounts and other settings were identical between the 2 runs, the Average Annual Return on the Performance tab and the xPortAPR on the Combination Results should be nearly identical for the selected Portfolio Size.

For more information on the different reports (Fixed, Variable and Trades), click the Help button on the Detailed Analysis window.

Hopefully that helps, but let me know if you have additional questions on this.

Pete
Overload
 
Posts: 2246
Joined: Wed Nov 30, 2005 12:14 pm

Re: View detailed analysis

Postby eleven » Tue Jan 29, 2013 11:47 am

Hi Pete,

Thanks for the quick reply. Even on the trades report the average annual return does not match. I have included a picture. Please advise.
Attachments
results.png
results.png (6.17 KiB) Viewed 2741 times
tradesreport.png
tradesreport.png (14.75 KiB) Viewed 2741 times
eleven
 
Posts: 5
Joined: Sun Jan 27, 2013 6:34 pm

Re: View detailed analysis

Postby Overload » Tue Jan 29, 2013 1:07 pm

Usually you'll see this when the Combination Result and the Detailed Analysis have been run with different settings. But even importing new prices between the runs can cause differences. To verify that you're using the same settings and prices for each, you should rerun both of them using these steps:

1) On the result in the Combination Results Listing, right-click and select "Rerun Combination".

2) When complete, on the result in the Combination Results Listing, right-click and select "Run Detailed Analysis".

When both of the reruns have completed, compare the result in the Combination Results Listing with the Performance tab of the Trades Report again, and let me know if that takes care of it.

Pete
Overload
 
Posts: 2246
Joined: Wed Nov 30, 2005 12:14 pm

Re: View detailed analysis

Postby eleven » Wed Jan 30, 2013 12:15 am

Pete,

I followed the steps you have outlined but the values still do not match up (combination result listing vs. detailed analysis - under 'trades" tab). As a mater of fact, when I "rerun combination" I get a different result every time. It appears to be somewhat random.

Can I email you my strategy setup for review? I cannot determine what could wrong with it.

Thanks,
eleven
eleven
 
Posts: 5
Joined: Sun Jan 27, 2013 6:34 pm

Re: View detailed analysis

Postby Overload » Wed Jan 30, 2013 9:31 am

As a mater of fact, when I "rerun combination" I get a different result every time. It appears to be somewhat random.

The only reason I can think of for this is that you've for some reason programmed a random number generation into your strategy using the random() formula. To check this, double-click on the result in the Combination Results Listing. This brings up a View Combination window that lets you see the specific formula being used. Check the General, Rank Selection, and Order Types tabs and see if you have a random() formula somewhere in there.

If that's not the issue, you can send me a copy by right-clicking on the result in the Combination Results Listing and selecting "Export Selected Results". Enter a valid filename for the export, and then click OK to create the export file. You can then send me a copy at support@stratasearch.com and I will have a look.

Pete
Overload
 
Posts: 2246
Joined: Wed Nov 30, 2005 12:14 pm

Re: View detailed analysis

Postby Overload » Wed Jan 30, 2013 2:22 pm

Thanks, I got the file. But when I rerun either the combination result or the detailed analysis multiple times, I get the exact same result every time. Could you please try that again to confirm? If you rerun the combination result multiple times, you should get the same result each time. Likewise with the Detailed Analysis.

However, I've been able to recreate your scenario where the combination result does not match the detailed analysis. I'm still looking into that, but will let you know.

Pete
Overload
 
Posts: 2246
Joined: Wed Nov 30, 2005 12:14 pm

Re: View detailed analysis

Postby eleven » Wed Jan 30, 2013 11:36 pm

Pete,

I can confirm that when I rerun the "run detailed analysis" I get the same result. However, when I "rerun combination" it is different every time. Please let me know what you find.

Thanks,
eleven
eleven
 
Posts: 5
Joined: Sun Jan 27, 2013 6:34 pm

Re: View detailed analysis

Postby Overload » Thu Jan 31, 2013 12:12 pm

I believe I've found the problem. In your Phase() custom formula, the rcalc and icalc variables are uninitialized. Normally this isn't a problem if your first use of the variables is like this:

rcalc = ...
icalc = ...

But instead your first use of the variables is like this:

rcalc += ...
icalc += ...

Because those variables are not initialized to zero, their starting values will be random depending on what happened to be in their memory locations when they were first allocated.

To fix this, place these 2 lines immediately after your twopi initialization in your Phase() custom formula:

rcalc = 0;
icalc = 0;

After adding these lines, everything matches up, and I get the same values after every rerun. Please let me know if this takes care of it for you.

Pete
Overload
 
Posts: 2246
Joined: Wed Nov 30, 2005 12:14 pm

Re: View detailed analysis

Postby eleven » Fri Feb 01, 2013 12:25 am

Pete,

That worked. Thanks for your help.

eleven
eleven
 
Posts: 5
Joined: Sun Jan 27, 2013 6:34 pm


Return to General Discussion

cron