Inferring TE values in financial markets

[Navid Parviny]

Hi,

I'm trying to study the information flow between financial market time series using idtxl package. I use network analysis with JIDTkraskoveCMI. I have a problem with explaining the results. There is some explanation in "An Introduction to Transfer Entropy", where I could understand that the output values show bits of information transferred from one time series to another in a network. However, I have no idea how to explain for example TE(A->B)=0.056. I'm guessing that it means A transfers 0.056 bits of information to B. But still cannot understand what transferring 0.056 bits of information means!

Thanks

[p.wol...@gmail.com]

Hi,

I would suggest to do some introductory reading on the mutual information (as transfer entropy is just a conditional mutual information).

There are some classical text books (https://www.inference.org.uk/itprnn/book.pdf, https://link.springer.com/book/10.1007/978-3-319-43222-9), but also some papers that introduce information-theoretic concepts, e.g., https://arxiv.org/pdf/1304.2333.pdf, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6131830/.

Hope this helps to get you started.
Best,

Patricia