Why is cache not used for Volume Bars in Backtest or Optimization

[aslane]

When running a backtest or optimization with Volume Bars (perhaps any tick-based bars), Ninja seems to re-build the bars on each run which can take a relatively long time. Shouldn't these come out of the cache once built? Especially, once the analyzer is loaded. It would be fine if they were rebuilt once, but subsequent runs should reuse the bars that were built.

[peterslane1]

I agree. A good part of the backtest or optimization time is the time it takes to build the volume bars. It seems inefficient to rebuild these volume bars each run. Could make the backtesting function much quicker by storing the built bars.

[MojoJojo]

While NT isn't caching the bars it seems to be writing some of it to the db. I had this persistent problem where my second data series produced fewer bars than the primary one and therefore was desynchronized. Nothing helped until I started from an earlier date, afterwards the situation with the previous start date was fixed.

Fortunately the initialization time is the shortest part of my optimizations.

[NinjaTrader_Jim]

Hello,

Thanks for opening the thread.

I just wanted to let you know that your message has been received and we will be replying back after we have finished reviewing.

EDIT: I have set up a test where we can observe the bar cache getting written to the NinjaTrader 8\db\cache folder, and how we can observe the difference in speed before the cache is created and after the cache is created. Taking this test, we can see the bar cache is being used.

Demo - https://drive.google.com/file/d/1mo2...w?usp=drivesdk

MojoJojo, if you can provide a test script for this case and some steps that can be taken, I would like to review. Please reach out to me over email with the text "Attn Jim 2604911" if you can provide this.

We look forward to assisting.