Input shape for LSTM or RNN models

Lets say I am working on binary classification problem. I have time series data of 50 stocks(their fundamentals) .So now I have a test matrix of 2000x90… I have 90 fundamental features. Now what should be my input size for LSTM or RNN model