Optimization with Extended Scorecard crashes Wealth-Lab
Author: wl64bit
Creation Date: 3/4/2012 8:32 AM
profile picture

wl64bit

#1
Is there any way to let the Wealth Lab check if there is enough bars to perform calculations on the particular symbol first, and if it is not just skip the symbol instead of reporting this problem, or if there is not enough bars, this error will be always reported without skipping the symbol?
profile picture

Eugene

#2
Would be easier with an example which includes code and more details about your data.
profile picture

wl64bit

#3
Wealth Lab crashes when Optimizer encounter any error in the dataset while performing optimization. I think it has to do with the reporting of an error in the report log in the Optimizer.
profile picture

Eugene

#4
If you want us to help you, it's always a good idea to provide as many details of the error as possible:

* Everyone's definition of a "crash" is different. What's yours? How about a screenshot?
* Have we missed the Strategy code? Or maybe it affects some particular publicly uploaded Strategy?
* Symbols
* Data provider(s)
* Data loading range
* Enabled performance visualizers

That would be something to start with.
profile picture

wl64bit

#5
Take any daily dataset.
Ad a symbol ABCDEFG. This symbol will not have any bars to work on in the Dataset.
Take any strategy and add:
DataSeries SMATII= SMA.Series(TII.Series(Close,2,3),10);
While clicking on this symbol the strategy will generate an error at the bottom of the screen, while other symbols, which do have bars, will not.
If you start to optimize this strategy on this Dataset, the Wealth Lab will crash "wealth Lab has stopped working. Windows will close the program…"
If you delete this ABCDEFG symbol from the Dataset, Wealth Lab will not crash.
This way you will be able to reproduce the problem.
profile picture

Eugene

#6
Using WL 6.3, nothing crashes for me with either one of the three installed Optimizers (Yahoo! data).
profile picture

wl64bit

#7
Does the DataSeries SMATII= SMA.Series(TII.Series(Close,2,3),10); produce an error at the bottom of the screen when you click on the ABCDEFG symbol?
In my case, both Wealth Lab Pro 6.3 and Developer 6.3 crash on the BBFree and Quote Media datasets. I did not try Yahoo dataset.
profile picture

Eugene

#8
QUOTE:
Does the DataSeries SMATII= SMA.Series(TII.Series(Close,2,3),10); produce an error at the bottom of the screen when you click on the ABCDEFG symbol?

Sure it does - an IOOR error (index out of range) caused by accessing data on a bar that doesn't exist. (By design)
QUOTE:
In my case, both Wealth Lab Pro 6.3 and Developer 6.3 crash on the BBFree and Quote Media datasets. I did not try Yahoo dataset.

Sorry to rain on your parade but it's OK here with BBFree data as well. Point is, that single line is not enough to cause the crash. Provide us with a complete code example.
profile picture

wl64bit

#9
Could you please try extended scorecard?
profile picture

Eugene

#10
Thank you, that did the trick. I get a crash now with Extended Scorecard (only), even without smoothing the TII.Series. Unfortunately, the cause of that Argument.OutOfRangeException isn't evident to me. I'll file the bug report; entering a ticket is up to you.
profile picture

wl64bit

#11
I would use MS 123 scorecard but it is missing the most important parameters such as Profit Factor, Recover Factor, and Ulcer Index.
Is there any way to add those parameters to the MS 123 scorecard?
profile picture

Eugene

#12
You're free to program them as a DIY project as the solution's code is open source, but we're not going to duplicate the built-in parameters in MS123 Scorecard. We'll find the culprit of the Extended Scorecard crash.
profile picture

wl64bit

#13
I would not say it is a duplication, because if you get results with the MS 123 scorecard you can only see the parameters included in this particular scorecard and there is no way to switch to other scorecards after the optimization is complete.

Without these basic and important parameters that I mentioned, the MS 123 scorecard is incomplete and useless.
profile picture

Eugene

#14
The inability to switch to other scorecards is by design.

"Useless to me", you forgot to add, because other users who requested those advanced performance metrics would disagree with this opinion. However, I'll stress it once more: we never aimed at duplicating the contents of the built-in scorecards in MS123 Scorecard. Thus, it's not an "Extended Scorecard + ..." but an independent collection of unique performance metrics.
profile picture

Eugene

#15
I'd like to tell that the problem is in Community Indicators's TII.Series, offering a chance of a quick fix... Alas.

Conclusion: the problem is clearly caused by Extended Scorecard on symbols with zero bars. The issue will be reported to Fidelity.

Workaround: when working with Extended Scorecard, double check that all symbols actually have data.
profile picture

wl64bit

#16
Can you decide on quality of a trading system just by looking on the fancy parameters of an equity curve such as Sortino Ratio, Lake Ratio without having the Profit Factor?

The idea is to have the biggest picture possible but not just spots of information.
If it is doable, why not to add these crucial parameters to the comprehensive MS123 scorecard?
profile picture

Eugene

#17
OK that's a solid point. Although stuffing the optimization tab with duplicating elements isn't going to make everyone excited, the list of Extended Scorecard's remnants is finite. ;) The last ones: Profit Factor, Recovery Factor, and Ulcer Index are to be included on MS123 Scorecard too (next release).
profile picture

wl64bit

#18
Thank you so much!

If someone is concerned about too many columns in the scorecard, it is easily manageable by copying the scorecard and pasting it into Excel. From there you can readily delete the unnecessary columns by one click.
profile picture

Eugene

#19
Version 2012.03.1 has been released.

* New: Profit Factor, Recovery Factor, and Ulcer Index added to MS123 Scorecard (requested by wl64bit)
* Fix: Portfolio Inspector: unhandled exception when testing on non-overlapping data ranges (reported by alex arzh)
profile picture

wl64bit

#20
It looks like after this update the Optimizator stops optimization if it encounters an error "DataSeries do not have equal number of values for mathematical operation". When it reaches the symbol that does not have any bars in the history it reports the error in the Errors tab, skips the currently running optimization and start optimization with the second set of parameters.
This website uses cookies to improve your experience. We'll assume you're ok with that, but you can opt-out if you wish (Read more).