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 …
-
2/2 For vol estimation, the Hodges and Tompkins correction can be used to correct this bias (the bias in the variance of vol). Do you know if there's a similar method for general regressions?
Yeah you do throw away some data. The best approach is to stick with the smallest interval you have (one month in this case) and do all analysis there, which maximises power without requiring a complicated correction for overlapping periods.
-
-
thank you ser
I think the analytical overlap corrections give a false sense of security. after implementing them i thought i was gmi but then i looked into all the assumptions and wasn't so sure.
i draw more comfort from empirical approaches nowadaysThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
For the original post, the most important point isnt about statistical adjustments but... "If you can't find bleeding obvious evidence for something you think takes about 10 years to play out - how much you gonna bet on it?!"
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.