If you’re building a time series model and you want any hope of statistical significance, do your analysis on non-overlapping periods.https://twitter.com/macrocephalopod/status/1460302495507525635 …
-
in a regression the traditional fix for this would be something like newey west or hansen-hodrick standard errors. instead of assuming the residuals to be independent they also include the autocovariance terms in the estimate. you may find this useful: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3142575 …
That's the paper I was trying and failing to find earlier, thank you!
