FinAcademy › Risk Track I › Time series

Module R5 · QTA 10–11

Time series

Stationarity, white noise, AR, MA and ARMA models, reading ACF and PACF, forecasting and choosing lag length, then trends, unit roots and the Dickey-Fuller test.

3 lessons · 8 interactive figures · 21 exercises

1. 1Stationarity, white noise and AR(1)
   A series you can model is one whose mean, variance and autocovariances do not drift. AR(1) is the smallest model with memory.7 exercises
2. 2MA, ARMA, forecasting and model selection
   MA models remember shocks, AR models remember levels, ARMA remembers both. Information criteria stop you from remembering too much.7 exercises
3. 3Trends, unit roots and spurious regression
   Two random walks regressed on each other give a beautiful R². That is the trap. Dickey-Fuller is the way out.7 exercises