Backtest in progress VS new compilations

[misova]

I would like to ask,

PROBLEM
when there is a long backtest running (for example it runs 3 hours already and will need additional 2h to finish). During that long time, I change sources more times and run new separate smaller backtests.

QUESTION:
Can be the first long-running backtest somehow influenced by the new changes, that were implemented later?

[NinjaTrader_PaulH]

Hello misova,

Thanks for your post.

The issue is unlikely as each instance of strategy analyzer would run on its own and would copy the data.

[misova]

I found related problem:

Imagine scenario:

1. Long optimization is running - 200 combinations, takes 8 hours to finish.
2. During that optimization above - a change something in codebase, remove some properties and compile.
3. As a result of the changes, that optimization is influenced - new iterations in optimization (after changed was done) produce inconsistent results.

Is it possible, that new iterations of long-running optimization (which starts many individual backtest, that are based on new compiled *.dll) can be influenced with new changes and compilations during the long-running optimization?

I assume, that when one backtest is started, it uses the current version of compiled DLL.
When running optimization, there are many individual backtests, so when compiled DLL changes during the optimization, new started backtest will use the new DLLs and so the whole optimisation is inconsistent = different DLLs are used during lifetime of the optimisation.

[NinjaTrader_PatrickH]

Hello misova,

Thank you for your response.

Each time the parameters increment of the optimization goes to the next instrument in the list, the Strategy Analyzer will initialize the strategy once again and any changes we made in code would be seen.

If you need to change the code it would be best to create a copy of the running strategy and then make changes.