How to apply tensorflow/Keras to a time series data (stock market) where the target is dependent on multiple assets (stocks) and the predict the entire days stock movement consisting of day long minute level data

TensorFlow
,
1

I have a three - five year minute level financial data for a defined set of stocks. The model needs to predict a days worth of stock movement at minute level (approximately 500 values every day).
Each stock behaves differently , but the target is a weighted sum of all stocks.

Can I get recommendations on where i can read material for the above application to get started for using a GRU and LSTM type models?

thanks

3

Hi @Nazir, Apologies for the delayed response!
To implement this, use an Encoder-Decoder architecture, where the encoder processes multivariate input and the decoder outputs the full day’s vector. Kindly, You can refer to this documents to get started with Time series forecasting, Timeseries classification from scratch. Thanks!