There's also strange behaviour when the PT WFA finishes:
If you close the WFA report and enter the inputs of the last IS run (then maybe extend the end date, like I do), then run a backtest: The resultant performance report doesn't equate to the inputs.
If you save that PT file and then reopen it, and rerun the backtest, the performance report now equates to the inputs.
No idea what's going on there. But performance reports not relating to inputs is far from desirable behaviour. I've had to go through months of work and rerun these backtests, as most saved PT peformance reports were bogus.
Please update to the latest release because a bug that matches your description was fixed in it. If this doesn't help provide a step-by-step instruction for reproducing this on our end, e.g: perform the 1st optimization -> find the required IS date in the report -> perform the 2nd optimization with the new set of parameters, etc.
Also attach the following:
- your portfolio workspace for the first optimization;
- screenshots of the changes made for the second optimization;
- the portfolio workspace for the second optimization with the new parameters saved;
- export of data for all instruments used in the workspaces in QMD format: https://www.multicharts.com/trading-sof ... rting_Data
- export of testing signals.
Also please let us know if the same can be replicated in the Offline mode.
Can MC provide a way to turn off the final additional IS run please?
It is redundant anyway in PT as users can't view the resultant inputs from this run.
It also can add a lot of time to a WFA; especially an anchored PT WFA.
I would really appreciate a regedit/patch for this.
I’m afraid there’s no patch or registry key for this at the moment, but our developers might be able to come up with some improvements in this respect in MultiCharts 13.