Inside the neuro lab->select trainning data, there is no starting /ending date . In the data range panel, you can specify the date, but can this panel control neural training? I doubt it. No matter the start/end date I chose, the training bars in neural lab is the same. So, how can I do the training from such as 2000 to 2001?
Size:
Color:
No doubt: it does control the training period. However, when you read the Wealth-Lab User Guide > Neuro-Lab® >
Select Training Data, pay attention to a note marked
Important! which may apply to your case.
You could also reach this by using the
Data Tool which lets you physically truncate the data
before a date in the selected DataSet, for example.
Size:
Color:
How to install data tool?
I looked at the extension manager, I didn't find it.
Size:
Color:
Again, everything is covered in the User Guide which I strongly recommend. The EM lists the already
installed extensions.
To install Extensions please visit the Extensions section of our website. On how to install/uninstall/update Wealth-Lab extensions, review
this illustrated tutorial and/or check out the Wealth-Lab User Guide >
Extension Manager.
Note that the Data Tool approach will be cumbersome. First you'd delete the data to train on the old data and then update it again to apply the network. Instead you may consider configuring the Training Period (First N %) accordingly and load "Fixed N bars" on a synched DataSet.
Size:
Color:
I found it
Size:
Color:
I can't truncate the etf symbol such as spy using before or after certain day, I can truncate spy by days. Is this normal?
Size:
Color:
my goal is training only on uptrend, such as 2003 to 2007. I can't figure out how to select the data set.
Size:
Color:
QUOTE:
I can't truncate the etf symbol such as spy using before or after certain day, I can truncate spy by days. Is this normal?
Not enough detail to duplicate. Everything works fine for me.
QUOTE:
my goal is training only on uptrend, such as 2003 to 2007. I can't figure out how to select the data set
Just pick a date range. It should work properly for a single symbol (switch First/Last in the Training Period to verify) and also for portfolios after creating a DataSet synched by Starting Date using the Data Tool by truncating the bars before some date e.g. 2003. Of course this assumes that all of its symbols have started trading before 2003 so the resulting bar count would be equal.
Size:
Color:
I have just found SPY is only starting from 2003, I want to go back to original date (1993), I reloaded all the symbols from data manager, I removed data tool, But spy still starts from 2003, instead of 1993
Size:
Color:
it seems to be truncated by data tool, but how to recover it? I created a new potofolio of spy, still the same. can't go back
Size:
Color:
This is by design. And I did warn you that it's going to be cumbersome and not required for just one symbol (see reply #8).
For one symbol (SPY), open up a chart, right-click, "Reload chart history".
For a DataSet of multiple symbols, reinstall the Data Tool and click "Remove all data", then update the DataSet as usual.
Size:
Color:
Hi!
I have a similar question about specifying start and end date for my NN training data.
There is a topic under the Wealth-Lab Pro User Guide about Filtering the Training Data and I think it can be used for this purpose (Am I correct?).
So, could I use this approach to specify the start and end date for training data?
If I would like to include all the data from 3/30/2005 to the most recent date (for example 1/19/2018), how should I modify this code block to achieve this?
I have written the following code:
CODE:
Please log in to see this code.
Will this block of code filter my NN training data for only from 3/30/2005 to 1/19/2018? I will put this block of code in both input and output script. But, will it work?
Many thanks in advance.
Alex
Size:
Color:
Alex, I wonder if the code you came up with did the job for you?
Size:
Color:
Hi! Eugene,
I have used the code to train my NN for specifying start and end date for the training data. My NN does not produce any error and I have no other way to verify if it indeed trains the right period of data, so I am assuming it works for now.
Alex
Size:
Color: