how to build a prediction model on variable length multivariate time series data

Question

I have a time series price data which have a variable length. It is stated in the data description that total number of features is 269. Data is unstructured, and for each d