NA for Alternate Data results

Questions directed to our Support Team. Responses can be expected within 24 hours.

NA for Alternate Data results

Postby timr411 » Thu Apr 16, 2015 1:23 am

I just finished doing a one-click simulation. I have both an Alt1 data period and Alt2 data period set up for out-of-sample testing during the one-click simulations. In looking at the results in the combination results listing, I see that most of the time I get data for Alt1 and Alt2 time periods, but once in a while I'll just get NA for the Alt1 and Alt2 parameters (e.g., CpdAnnReturn). I can't figure out why in these cases I don't get any alternate data results when I do indeed get good data for the target (in-sample) time period.
timr411
 
Posts: 23
Joined: Fri Jan 31, 2014 2:01 am

Re: NA for Alternate Data results

Postby Overload » Thu Apr 16, 2015 8:44 am

Are you getting NA just for CpdAnnReturn, or do you get NA for all columns? This might be important because it could identify whether the source of the problem is with the entire analysis, or just with CpdAnnReturn. If there is a problem with the entire analysis, there may be a line in the File > Event Viewer that identifies the issue. You might see a line in the Event Viewer beginning with "Error in Alternate Data" followed by additional detail. If it is just CpdAnnReturn, it could be an issue just with that column. For example, if the Ending Equity ends up being negative, the CpdAnnReturn will not be produced.

Just so you know, you can right-click on the OneClick Result in the Combination Results Listing and select Run Detailed Analysis at any time. This will reproduce the Detailed Analysis and may recreate information in the Event Viewer if there was an issue.

Hopefully this helps identify it. But if not, let me know and we can dig a little deeper.

Pete
Overload
 
Posts: 2241
Joined: Wed Nov 30, 2005 12:14 pm

Re: NA for Alternate Data results

Postby timr411 » Fri Apr 17, 2015 2:43 am

I reran the detailed analysis, and no errors showed up in the Event Viewer. I'm getting NA for A1CpdAnnReturn, A2CpdAnnReturn, KRatio,and I'm getting 0.00% for A1AvgAnnReturn, A1MaxDrawDPctU, A1PctProfitable, A1SharpeRatio, A1MCAvgAnnReturn (and for the same A2 parameters).

In looking at the Detail tab, it starts with $100k, makes many buys and sells over a 10-year time period, and ends up with a cash balance of $361k. Of course, this is the in-sample time period, but I have to believe that there is some kind of gain in the A1 and A2 out-of-sample time periods as well but the results don't seem to support this.
timr411
 
Posts: 23
Joined: Fri Jan 31, 2014 2:01 am

Re: NA for Alternate Data results

Postby Overload » Fri Apr 17, 2015 8:51 am

Your next step is to test one of your alternate data periods separately, just to confirm it will produce trades. Right-click on the result in the Combination Results Listing, and select Create Multi-System. This will create a completely independent copy of your system that you can run separately. You should enter the original OneClick Setup and verify the dates and sector being used by the alternate data period. Then set the dates in the Setups > Trade Settings, General tab, and run the new Multi-System against the designated sector. The results it produced will give confirmation that the alternate date period should actually have trades.

Pete
Overload
 
Posts: 2241
Joined: Wed Nov 30, 2005 12:14 pm

Re: NA for Alternate Data results

Postby timr411 » Sat Apr 18, 2015 10:46 pm

Okay, I did that, except the "Create Multi-System" was grayed out so I chose Create Strategy and ran that strategy across the Alt1 time period of the original test. After doing a detailed analysis of the results and choosing the Summary tab, I get real trades happening and real numbers for the in-sample period (which was the Alt1 time period in the original test). So, I still don't know why it gave me NA for this time period in the original test. Suggestion for the next thing to try??
timr411
 
Posts: 23
Joined: Fri Jan 31, 2014 2:01 am

Re: NA for Alternate Data results

Postby Overload » Sun Apr 19, 2015 9:02 am

I can't think of any other reasons for this offhand. So the next step will be for you to send me some results and settings so that I can look more closely. Here is what you should do:

In the Combination Results Listing, right-click on the result that is not showing the Alternate Data, and select "Export Selected Results". After doing this, the Export Database window will display and there should be 1 Result checked for export. Next click the Filter button to the right of the OneClick Setups and add check the box for the OneClick Setup that created that result. Click OK so you're back on the Export Database window, and then make sure you've entered a good filename that will contain the export. After clicking OK again, the file will be created, and you should send me a copy at support@stratasearch.com. I can then look a little closer and maybe try to recreate the problem here.

Thanks,
Pete
Overload
 
Posts: 2241
Joined: Wed Nov 30, 2005 12:14 pm

Re: NA for Alternate Data results

Postby timr411 » Mon Apr 20, 2015 1:32 am

Okay, I did the above request, and you should see the exported file in your mail.
timr411
 
Posts: 23
Joined: Fri Jan 31, 2014 2:01 am

Re: NA for Alternate Data results

Postby Overload » Mon Apr 20, 2015 10:56 am

Thanks, I think I see the problem. The result you sent has this trading rule in the Entry String:

Code: Select all
// Klinger Oscillator Divergence - Sector Periods:HB2 (Supporting Entry and Exit)
(numsymbols("S&P 500 Index", diverge(kvo(), mov(close, 3, simple), 6, 9) <> 1) /
totsymbols("S&P 500 Index") > 0.8)


However, your Alternate Data analyses are both set to run against "All Securities". So what happens is the above trading rule is automatically adjusted to test against the sector being run. So when it is run against All Securities for the Alternate Data Analysis, it is adjusted to this:

Code: Select all
// Klinger Oscillator Divergence - Sector Periods:HB2 (Supporting Entry and Exit)
(numsymbols("All Securities", diverge(kvo(), mov(close, 3, simple), 6, 9) <> 1) /
totsymbols("All Securities") > 0.8)


But the reason you're not getting any results for your Alternate Data is because the above line never gets above 0.8. To test this, you can take the following line from the above formula:

Code: Select all
numsymbols("All Securities", diverge(kvo(), mov(close, 3, simple), 6, 9) <> 1) /
totsymbols("All Securities")


and place it in the Display > Calculations window. You can see that when using All Securities it never gets much above 0.02, while when running against the S&P 500 Index sector it crosses above 0.8 pretty regularly.

When running against All Securities, no trades are produced because of the above trading rule. So it appears to be correct that the AvgAnnReturn would display as 0.00% and most of the other values would show as NA.

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

Pete
Overload
 
Posts: 2241
Joined: Wed Nov 30, 2005 12:14 pm

Re: NA for Alternate Data results

Postby timr411 » Tue Apr 21, 2015 7:23 pm

Wow, great detective work, Pete. I guess I should have specified S&P500 Index for the alternate data Sector specification to be consistent with the One-click Sector specification.
timr411
 
Posts: 23
Joined: Fri Jan 31, 2014 2:01 am

Re: NA for Alternate Data results

Postby Overload » Wed Apr 22, 2015 11:37 am

No problem. I actually had to run it through the debugger to figure it out myself. One would think that a system that produces trades when run against the S&P 500 Index Stocks would also produce trades when run against All Securities. But that was a reminder to me as well that that's not always the case. Glad we got it figured out.

Pete
Overload
 
Posts: 2241
Joined: Wed Nov 30, 2005 12:14 pm


Return to Technical Support

cron