Time Series Analysis: Why are Unit Roots Important?

[@samxware]

damn, besides being a quant expert, bro is handsome as f

[@irmdev595]

can i suggest fixed income python topics as well?

[@barricuda5]

Would love to see more TS topics

[@TheRoooot]

awesome channel Does this test mean that the hypothesis is false and there is no need for other tests?

[@QuantGuild]

Glad you like the channel! When you get to the stage of testing (mainly backtesting for strategies) you are looking to corroborate your hypothesis, if your test(s) don't this is where you have to ask why and go back to the start scientific method style to see where the idea breaks down. For example, if I believe news sentiment yesterday can predict returns tomorrow for a given equity and I backtest a sentiment model based on this hypothesis and the results are not profitable then I'm not going to keep optimizing the backtest until it fits my hypothesis - instead, I'll decompose the tests to see why we aren't producing a profit (is it already priced in? can we see it in contemporaneous price impact? etc.) Hope this helps!

[@TheRoooot]

@@QuantGuild thank you

[@sander2024]

If you can, use VEC or VECM model instead of differencing, you lose less information