Extend strategy submission time Q21
-
Hi Quantiacs,
I notice that Q21 is about to run out of submission time (May 1), but now I realize that the time to submit a strategy is very long (takes 3 - 5 days). Hopefully Quantiacs can provide a quick submission strategy solution (I used single backtest, running precheck is also very fast), or it would be really good if Q21 could extend the submission time by a few weeks.
Besides, I would also like to ask if ML and DL algorithms switch from ml_backtest to single backtest, will they overfit when using all the data for training? In ml_backtest, it is allowed to divide train_period and test_period, so how can we divide like that in single backtest? We look forward to your help providing a single backtest code example for ML DL strategy. Thank you. Besides, we hope that Quantiacs will provide single backtest for ML DL to help limit the forward looking situation.
Thanks a lot.
-
@illustrious-felice Hi, sorry for the delay, we extended the deadline.
As far as looking forward is going on, this does not take place during the real-time simulation, as the code cannot access by construction the new data. This is true even if the code is written in a single-pass way.
If the user uses a single-pass implementation, however, looking forward in the in-sample period is possible of course, there it is responsibility of the user to prevent it. Single-pass has access to all the time series.
