Error calling 'OnBarUpdate' on bar 439: Function does not accept ... NAN values.

[Dr Kerry]

My strategy generates this error message on some futures instruments but not on others:


Error on calling 'OnBarUpdate' method on bar 439: Function does not accept floating point Not-a-Number values.


It works fine, error free, with NQ but if I switch to, for example, ES, CL, or GC it wont run and I get the above error (although the specific bar number may be different).

Any ideas, generally, why this may be happening and how it can be fixed? It seems like there may be missing data or something of this nature but the error never happens with NQ.

Thank you,

Kerry

[NinjaTrader_ChelseaB]

Hello Kerry,

The error is indicating that there is a line of code calling a method, and something supplied has a NaN value.

What is the specific line of code causing the error?

If you are unsure, add prints above each line to see which print is the last to appear before the error appears.

https://ninjatrader.com/support/forum/forum/ninjatrader-7/strategy-development-aa/99209-date-conditions-not-working?p=791121#post791121

[Dr Kerry]

Hello Kerry,

The error is indicating that there is a line of code calling a method, and something supplied has a NaN value.

What is the specific line of code causing the error?

If you are unsure, add prints above each line to see which print is the last to appear before the error appears.
https://ninjatrader.com/support/foru...121#post791121

Hi Chelsea,

Thank you for your suggestion to use print statements to debug. So to be clear, as I wrote above, the strategy script works perfectly fine with NQ but if I switch to another instrument it won't even enable. It immediately shuts down after I click the enable box and the above mentioned error is displayed on the script output screen. I don't think that it will even get to a point where a print statement would execute. Can Visual Studio help in this situation? Otherwise, I may need use brute-force and comment out sections of the script until I isolate the problem. I still also don't understand why it's OK with NQ but switching to another instrument causes it to crash with the NAN error.

[Dr Kerry]

Hello Kerry,

The error is indicating that there is a line of code calling a method, and something supplied has a NaN value.

What is the specific line of code causing the error?

If you are unsure, add prints above each line to see which print is the last to appear before the error appears.
https://ninjatrader.com/support/foru...121#post791121

Chelsea,

I did some additional investigating and found that the error is strongly dependent on the time scale that I choose. The shorter time scales (less than 5 minutes) seem to cause the error to occur more often on all instruments. When using longer time scales (above 5 minutes) I was able to get the strategy script to work on multiple instruments error free. So it seems to me that there is some sort of issue with loading data from the time series especially on the shorter time scales which results in the NAN error. What do you think? It ought to work on all time scales so long as the time series data isn't corrupted (or missing).

Thanks again,

Kerry

[bltdavid]

If you can't post any code to help anyone understand
what you're doing, you need to get creative and add
print statements that will help you see the problem.

Chelsea has supplied the best advice available, given
that you won't (or can't) reveal any code.

Almost guaranteed, you have some kind of logic error,
a bug that gets tickled only under certain conditions.

Many expert C# programmers read these forums,
many would gladly help you, even download and run
your code -- but we're blind as wombats out here.

No one knows what you're trying to do.

Just my 2˘.

[Dr Kerry]

If you can't post any code to help anyone understand
what you're doing, you need to get creative and add
print statements that will help you see the problem.

Chelsea has supplied the best advice available, given
that you won't (or can't) reveal any code.

Almost guaranteed, you have some kind of logic error,
a bug that gets tickled only under certain conditions.

Many expert C# programmers read these forums,
many would gladly help you, even download and run
your code -- but we're blind as wombats out here.

No one knows what you're trying to do.

Just my 2˘.

David,

Thanks for your $0.02, much appreciated. I would have posted the code but the script is rather long and so I wanted to first try and narrow it down to the problematic lines and post those rather than have people waste their valuable time looking over lines and lines of code that were irrelevant to the NAN error that had been generated. That is why I was trying to get some general ideas as to what might be going wrong.

In any case, I was able to resolve the problem, using a few print statements, and explain why it was dependent upon the chosen time scale. It turns out that it was a divide-by-zero issue that occurred when normalizing a probability calculation that I am carrying out in the script.

For example, using a normalization parameter for the calculations such as math.abs(Close[0] - Close[1]):

Then on longer time scales (say 4 minutes or longer), during normal trading hours, and assuming an actively traded instrument, there is a high likelihood that Close[0] is NOT EQUAL to Close[1] and so there will NOT be a divide-by-zero error generated since, in this example, math.abs(Close[0] - Close[1]) is unequal to zero.

However the shorter the time scale (below 4 minutes) the higher the likelihood that Close[0] remains equal to Close[1] which would cause a divide-by-zero, resulting in the NAN error.

On more lightly traded instruments, even on longer time scales, there is a good possibility that Close[0] remains equal to Close[1] which would cause a divide-by-zero resulting in a NAN error to occur with these instruments even in the longer time frames.

So resolving the above noted error came down to checking if the particular normalization parameter, used in each instance, is equal to zero in order to avoid a divide-by-zero and the subsequent NAN error that caused the script to crash when it occurred.

Regards,

Kerry

[bltdavid]

Ah-hah!

The power of the 'Print' strikes again.

Simple advice, huge benefits.

Want to know a professional software engineering secret?

There are thousands of problems solved daily around the
world (considering all languages) using variations of this
technique.

Glad you found your problem.

[Dr Kerry]

Ah-hah!

The power of the 'Print' strikes again.

Simple advice, huge benefits.

Want to know a professional software engineering secret?

There are thousands of problems solved daily around the
world (considering all languages) using variations of this
technique.

Glad you found your problem.

Yes, I know. I use print statements all the time when developing models in Mathematica (a.k.a. the Wolfram Language).

Regards,

Kerry