Neural Networks in Trading - Training DNN using DAILY data, but model calling for MINUTES data

AI & Machine learning
1

Situation : 


  • We noticed that we backtest the DNN model using BAC DAILY data


  • But in sample_NN_model.py. we're calling 1m MINUTE data. code snippet shown below inside the handle_data function :
&nbsp;&nbsp;&nbsp;&nbsp;#&nbsp;Fetch&nbsp;the&nbsp;data<br />
&nbsp;&nbsp;&nbsp;&nbsp;hist&nbsp;=&nbsp;data.history(context.security,&nbsp;[<br />
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&#39;open&#39;,&nbsp;&#39;high&#39;,&nbsp;&#39;low&#39;,&nbsp;&#39;close&#39;,&nbsp;&#39;volume&#39;], minimum_feature_length,&nbsp;&#39;1m&#39;)&nbsp;<br />
```<br />
<br />
<br />
Problem :<br />
As per our understanding, we&#39;re training our DNN in the jupyter notebook to get the best settings &amp; weights, and saved in .json &amp; .h5 format respectively.&nbsp;<br />
<br />
But the notebook uses DAILY data, and sample_NN_model.py calls MINUTE data, and we are about to load the best DNN setting &amp; weights into it.&nbsp;<br />
<br />
Wouldnt that be wrong and makes no sense ? Or please do correct me if we inteprate this wrongly<br />
<br />
Thanks!<br />
<br />
&nbsp;
2

Yes, you are right. It should be the daily data. We have made the necessary changes. Thanks!