TS

Time series

Stationarity, ARMA, cointegration and volatility models — plus the out-of-sample discipline that stops a backtest lying to you.

1
Foundations
Stationarity, ergodicity, ACF and PACF, and the Wold decomposition.
- 1.1Foundations: stationarity, autocorrelation and the Wold decomposition12 min
2
The ARMA family
AR, MA, ARMA and ARIMA: identification, estimation and forecasting.
- 2.1The ARMA family: identification, estimation and forecasting12 min
3
Unit roots and cointegration
ADF, spurious regression and the basis of pairs trading.
- 3.1Unit roots, spurious regression and the basis of pairs trading13 min
4
Volatility modelling
ARCH, GARCH, asymmetry and realised volatility.
- 4.1Volatility models: ARCH, GARCH and realised measures13 min
5
State space and filtering
The Kalman filter and dynamic linear models.
- 5.1State space and the Kalman filter12 min
6
Financial stylised facts
Fat tails, clustering, the leverage effect and variance ratios.
- 6.1Stylised facts: what financial returns actually look like12 min
7
Forecast evaluation
Walk-forward design, purged cross-validation and backtest overfitting.
- 7.1Forecast evaluation: walk-forward design and backtest overfitting13 min