How can I prepare my data?


So I have a large ~1 GB CSV file with forex tick price data for a whole year, formatted like below:

My question is, how can I wrangle with this data in the following ways:

  • Cut out 10000 randomized 30 minutes sections of the data.
  • For every 30 minutes section, make simulated Long AND Short trades at the 6 minute mark, with a certain predetermined stop loss and take profit. Then label each piece of 30 minute section into one of 4 categories: Should go long, Should go short, Both are profitable (unlikely, but let’s list it out anyway), and Either is okay.
  • Generate timeseries data of the first 5 minute of every 30 minute section in a format that an LSTM neural network in DLS can process, with correct labeling and everything.


Hello you can’t do this directly with DLS for this you have to write a python script if you want I can help you with that.
Meanwhile you can refer to this article for creating Time series data for LSTM



Thanks for the reply. And yes I’ll greatly appreciate your help!
So the Excel file in the article is how my data should look? I thought my data should have looked like the arrays in the IMDB data set.
EDIT: Also does the prepared data needs to have a uniform data interval? For example, do we need to explicitly list out everything that happens during a millisecond even if nothing actually happens?