The course starts by providing to the student knowledge and comprehension of the different types of information that were generated in the shape of time series, so that it could study in depth the properties of the same ones. With it it happens(passes) to be analyzed: a) the evolution that the local averages of the above mentioned information usually show, so much in the shape of systematical growth (that gathers the trend evolution from the analyzed dynamic phenomenon), as in cyclical form of annual regularity (seasonal nature); and b) the existing temporary dependence on the deviations that the real(royal) information shows on the mentioned footpath of tendency and seasonal nature: (stationary) short term oscillations. On the base of the previous thing the models interfere univariantes to explain the generation of individual series, using initially the hypothesis of which suspense does not exist on the future of the tendency and of the seasonal nature (you structure determinists), to contemplate later schemes of unitary roots in which such components incorporate in every moment of the time shocks random (unpredictable), that are perpetuated towards the future, introducing this way the models ARIMA.
The course finishes with an introduction the models GARCH and the stochastic volatility models, that allow to represent the suspense of series of financial yields. The above mentioned models are implemented to obtain estimations of the volatility of real series. Also management of the risk illustrates its importance in some financial models like, such as models of the valuation of financial assets or risk management.
Each topic exercises must be conducted using the R software.
Topic 1. Introduction. Classic approach of analysis of series of time: descriptive study of temporary series.
Topic 2. Box-Jenkins methodology.
Topic 3. Regression models for stationary and non-stationary time series.
Topic 4. Linear stationary models for time series.
Topic 5. Linear non-stationary models.
Topic 6. ARIMA model.
Topic 7. Conditional Heteroscedastic models: ARCH y GARCH.