How to standardize/normalize values MULTIPLE SEQUENCES in neural networks?

I have an cell array of {1x491} with [2x33] values. Technically, I have 33 time series of length 491 with padded NaNs.

I need to standardize/normalize my inputs and outputs to find the significant lags of the inputs and the feedback.

I have tried this, however, I lose my resolution and everything is converted to -1 or 1. I don't know if I should do the zscore of X and T instead, and how to do it?

x  = cell2mat(X);
t  = cell2mat(T);

%x = zscore(x, 1);
%t = zscore(t, 1);