Hello! Would some body be able to explain why on fitted arima models the var_coef matrix is not always diagonally symmetrical? The asymmetry seems to occur when order 'p' >= 2. If this is unexpected, I can open github issue and provide reproducible examples.

Which version are you using?

statsforecast==1.7.3

Can you please open an issue? It had a problem but it was fixed in 1.7.0

yes, will do. I'll open when I get a moment. Thanks!

Thank you!

issue create, https://github.com/Nixtla/statsforecast/issues/847