Multi-Instrument Strategy Analyzer INCREDIBLY slow when displaying analysis

[Skifree]

I'm trying to run my strategy for 6 months, 5 min bar and 15 sec bar. Per instrument, it takes about five minutes which is fine, but the minute i run optimizations or multiple, it completely tanks.

In the screenshot below, I ran these strategies 6 hours ago - i'm not sure when they completed but it does it in a reasonable amount of time. But displaying the results is near impossible. It's now been stuck "Generating Report" for 15 minutes. The XML files are all generated in logs, but it can't seem to produce anything in the UI which makes this very tedious.

How can I improve this feature performance, because TBH, it's very subpar and worthless if I can't analyze the results in a reasonable amount of time.

again, to be clear, this specific issue is not per strategy but loading the data in multi

[Skifree]

This is also really frustrating because my printouts of the strategy finish lightyears before the analysis page actually loads. I'm currently waiting for an MGC test to load, for only 134 trades.

My code is clearing out all lists and dict on exit, VS shows no custom methods consuming irregular amounts of CPU. My NT8 folder has about 15GBbut its only market data for backtests, otherwise everything seems efficient.

At this point it seems impractical to use the analyzer. It used to be perfectly fine in prior versions?

[Skifree]

here is the problem now on two different computers. Screenshot below is from my VPS, a much leaner version. Even after the strategy loads results back into the UI, it absolutely crawls just trying to change sections and save. It continually goes into 'the app is not responding do you want to wait or close" it is only consuming 359MB of ram while stuck and 36% CPU on a computer with otherwise nothing running

[NinjaTrader_Jesse]

Hello Skifree,

From the description of the vps that sounds like it may be a pc performance issue, if displaying the results is taking that much cpu or a lot of time that may not be capable of running larger tests on that machine.

When doing an optimization the time that takes heavily relates to your strategy and its parameters. Depending on how many parameters you are optimizing at once that can exponentially increase the time the test takes. Optimizations are generally for targeted tests where you have a specific range of values you wanted to check for certain parameters. If you optimize large differences in parameters or many parameters at once that will create many permutations that need to be tested and filtered causing a longer test time.

This can also relate to what your strategy is doing in its logic, it doesn't sound like you are doing anything memory intense that would cause you to run out of ram, 300mb of ram is not much at all. If your strategies logic is not optimized that may also affect the test time. You can test the SampleMACrossOver strategy in the same use case to get an idea if the problem relates to the strategies logic, that is a very simple strategy and should test fairly quickly.

[Skifree]

It's not a pc performance issue. It's an Alienware gaming pc with top gpu, cpu and lots of ram.

What does strategy analyzer do after running that would take so long? My strategy has printouts that display stats at the end of the strategy and it does cleanup of objects - but why would it take so long to render results for relatively few trades?

Comparing it to the sample strategy isn't really effective because that's an over simplified script whereas mine is more complex but still should be within NT8s capabilities.

[NinjaTrader_Jesse]

Hello Skifree,

Unfortunately I wouldn't have any details on how the analyzer works internally to process the results but that can take some time depending on how many trades there were, if multiple instruments were used and in an optimization how many permutations have to be analyzed and filtered.

Comparing to a sample strategy sets a baseline, that strategy places a good number of trades so there can be quite a bit of results and has 2 parameters that can be optimized. If that strategy is able to optimize fast and also display results fast then the slowness would be a result of something specific to your strategy. That could include the setup, you mentioned multiple instruments which will add to loading, testing and result calculation time. If you have a large number of parameters being optimized at once that could be the difference between hundreds of tests or thousands, the higher amount of permutations will cause a much larger set of results to sort which could add to the time it takes to display the results when complete.

To find out what is happening with your strategy you can try to simplify it or exclude parts like the secondary instruments to see if there is a point when it has faster test times, once it is known what is causing the longer display times we could more accurately say if that is expected or not.