Chapter 17: Time Series Flashcards
1
Q
Stationarity WICS TW
A
Weak Integrated Covariance Strict - Identical joint distributions Trend White Noise
2
Q
Testing fit of time series models VALR TD
A
Visual assessment AIC, BIC Ljung-Box test Residual assessment Turning point test Durbin-Watson statistic
3
Q
AR and MA process PEL E
A
Previous terms
Error
Linear combination
Extra q white noise terms added
4
Q
Modelling data features SAS
A
- Seasonality – indicator variables
- Altered rates of change – drift or mean reversion and Chow test
- Step changes - Poisson
5
Q
Properties of GARCH model PLUM LT
A
- Past volatility and values influence volatility
- Longer periods of high volatility catered for compared to ARCH
- Unstandardised and standardised residuals created
- MLE used to fit GARCH model
- Less parameters than ARCH model
- Test standardised residuals for white noise
6
Q
Scaling up statistics VETS I
A
- Volatility times by square of T
- Estimate annual volatility form monthly view
- Timescale times by T
- Stochastic modelling considered as alternative
- Insufficient data
7
Q
Strict and weak stationarity I CCN
A
Identical statistical properties at any time in the process
Constant mean
Covariance depends only on time difference k:
N-order weak stationarity
8
Q
White noise FUF
A
Fluctuates around zero
Uncorrelated with past observations
Finite variance – strict white noise
9
Q
Trend stationarity TOS
A
Time dependent function
Observations oscillate around a trend
Steadily changing
10
Q
Merits of the ARCH model SLC SAPI
A
- Simple
- Leptokurtic
- Clustering of volatility
- Short periods of volatility considered
- Application is limited
- Positive and negative values are possible
- Intuitive meaning lacks – purely statistical model
