Intermixing EFS1 with EFS2 studies

[mbuta]

If this is not the appropriate place for this suggestion, please re-direct.

I have two suggestions for future releases of EFS2:
1. Please consider allowing mixing EFS1 with EFS2 studies.
Some examples:
var XX = new MACDStudy(x, y, hlc3(inv(15)), true, true);
var YY = rsi(x, studyZZ, MACDStudy.HIST);

2. If possible, please consider allowing the declaration of EFS2 studies/data series outside main(), for obvious performance advantages.
Note: If this is possible, first suggestion can be reduced to providing in EFS2 exact replica of EFS1 studies [now we have to declare each component of complex studies, like ADX, Stoch, MACD, which introduces performance degradation].

Thank you.

[ACM]

Re: Intermixing EFS1 with EFS2 studies

Mihai

...now we have to declare each component of complex studies, like ADX, Stoch, MACD, which introduces performance degradation

Actually in this post I believe I provided some evidence to the contrary.
FYI suggestions should be sent to [email protected]
Alex

Originally posted by mbuta
If this is not the appropriate place for this suggestion, please re-direct.

I have two suggestions for future releases of EFS2:
1. Please consider allowing mixing EFS1 with EFS2 studies.
Some examples:
var XX = new MACDStudy(x, y, hlc3(inv(15)), true, true);
var YY = rsi(x, studyZZ, MACDStudy.HIST);

2. If possible, please consider allowing the declaration of EFS2 studies/data series outside main(), for obvious performance advantages.
Note: If this is possible, first suggestion can be reduced to providing in EFS2 exact replica of EFS1 studies [now we have to declare each component of complex studies, like ADX, Stoch, MACD, which introduces performance degradation].

Thank you.

[mbuta]

Alex,

I also provided evidence that there is performance degradation, I also posted the efs for those measurements, as instructed, and nobody replied anymore.

I think is hard to argue that three studies [to take MACD as example, where we have to declare separatelly MACd, SIGNAL and HIST] execute faster than one, like in efs1.

Mihai

[ACM]

Mihai
I believe Dion addressed your point by explaining that you are not comparing equal functionality. You may want to review the thread I linked.
As to the MACD it is interesting that you mention that study as it is exactly the one I used for the comparison results I posted (again you may want to read through my linked post) and it appears clear [to me at least] that the EFS2 version was considerably more efficient.
Regardless as I said in my prior post you may want to send your suggestions to [email protected] as this forum is for requesting studies not enhancements to the application (or any of its components).
Alex

[mbuta]

Well, you guys may know magic that we don't. I have exactly the same structure [just not by themselves, as it is not realistic] and get slower execution.

By the way, your amStudies library has some very nice functions. Thank you for making it available. I plan using some, as soon as I get a resolution on how to use EFS2 with reasonable execution time [because except this performance issue and the fact that I cannot plot/use them correctly in real time, i like EFS2 a lot].

Mihai

[ACM]

Mihai
Thank you for the compliment and you are most welcome
Alex

[mbuta]

Alex,

We thank you!

One more thing about the MACd example: Have you tried your measurements for a higher time interval? Because this is an other diference, I am using EFS2 ONLY for higher intervals.

Mihai

[ACM]

Mihai
Yes I have and as to be expected the performance changes because the functionality will also have changed as Dion has clearly explained to you in the other thread in which this subject has already been discussed extensively.
If you want to compare apples to apples then you need to program the EFS1 equivalent of the study to perform a true syncronization across multiple intervals (which you are not actually doing in your EFS1 scripts).
Anyhow this is not the appropriate forum for this discussion so if you have further issues on the subject I would suggest that you post them in the original thread.
Alex