DataSet subset for Neuro-Lab Training
Author: ronc
Creation Date: 9/21/2015 9:46 PM
profile picture

ronc

#1
The WL User Guide, Sec. 17.5 says that NeuroLab uses the specified percentages of an entire data set for training. I want to train on specific time periods. NeuroLab allows selection of first/last percentages of a dataset but not start/end dates (or bars). I would like to work around this by creating a new dataset comprising data from the original dataset from some start date to some end date. How can I do this? I am generally working with 1-min bars, and would like to create a new data set from date A to date B.
Thanks,
Ron
profile picture

Eugene

#2
As Neuro-Lab is an extension, why have I found the post in "General" again? Targeting proper category is easy: please take half a minute to read their descriptions. Thanks.

You can not work around it this way because DataSets can not have starting and ending dates. DataSets are like shortcuts in Windows, they are pointing to static data files. If you cut some bars here and there, that would affect all of your DataSets. With EOD data so easily accessible, it might be possible to "dedicate" a different provider's data just to trim its start/end dates using the Data Tool's Truncation feature (Days Before/After). However, Fidelity is the only intraday data feed available (unless you have an IQFeed paid subscription) so like I said, that would trim all of your data - perhaps except the Google intraday feed which is limited to a handful of days for 1-min data. And note that you can not create multiple DataSets with this hackish approach: only one.

Enough theory. Perhaps the only reasonable workaround is to simply export your data to an Excel file (for Excel static provider) or CSV file (to pick it up with ASCII provider). Just set any date range before running the script. This way, your ASCII/Excel DataSets will have start/end dates. Voila.
This website uses cookies to improve your experience. We'll assume you're ok with that, but you can opt-out if you wish (Read more).