Trained a time series neural network (NAR) with multiple inputs, now network requires multiple inputs to work/run. How do I train a time series neural network on multiple datasets, while keeping only 1 time series input to network

I am trying to use a time series neural network (NAR) to predict stock prices.

I would like to train the network on different stocks, say 5 different stocks. And I have say 100 time steps. So I created a matrix with 100 rows and 5 columns. I then trained a NAR neural network on the dataset.

Now I want to see what the network will predict if I give it some data on a new stock. But, it seems to me like my network is expecting 5 inputs. That's not what I wanted. I just wanted to give the network 5 different datasets to train on, not use them all together to predict the next 5 prices.

Am I understanding my network correctly? Does it truly require 5 inputs? Or is there an option to change it back to one input, now that it has been trained.

If not, is there a way to train a network on mulitple inpedendent data sets?

Thanks!

- Dan