Browsing by Author "MacDonald, Iain L"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
- ItemOpen AccessHawkes processes and some financial applications(2014) Lapham, Brendon M; MacDonald, Iain LThe self-exciting point process, which is now more commonly known as the Hawkes process, is a model for a point process on the real line introduced by Hawkes (1971). The distinguishing feature of such processes is that they allow all past `events' to affect the intensity function at the current time. Over the years such processes have been applied in seismology and neurophysiology in particular, and in more recent years there have been significant financial applications. In almost all of these applications, the route used to find the maximum likelihood estimates (MLEs) is direct numerical maximisation (DNM) of the likelihood. An EM algorithm, which makes use of the Poisson cluster process interpretation of the Hawkes process, is an alternative route to the MLEs. This particular EM algorithm has received attention in the literature and has been claimed to have advantages over DNM of the likelihood. We carry out a simulation study for a simple Hawkes process to clarify statements made in the literature about these advantages. For the simple Hawkes process models that we consider, DNM of the likelihood is the preferable route to finding the MLEs. We then use DNM of the likelihood to _t marked Hawkes process models to South African asset data. These applications to South African data include the modelling of extreme asset returns and the forecasting of conditional value-at-risk (VaR) and expected shortfall (ES). The models investigated include mostly models found in the literature, but also include some variations introduced here. In a backtesting exercise, we compare the conditional VaR and ES forecasts found by using the marked Hawkes process models with those found via some nonstandard stochastic volatility (SV) models. We find that the marked Hawkes process models give mostly competitive forecasts of conditional VaR and ES when compared with the nonstandard SV models.
- ItemOpen AccessOn improving the forecast accuracy of the hidden Markov model(2016) Rooney, Thomas J A; MacDonald, Iain LThe forecast accuracy of a hidden Markov model (HMM) may be low due first, to the measure of forecast accuracy being ignored in the parameterestimation method and, second, to overfitting caused by the large number of parameters that must be estimated. A general approach to forecasting is described which aims to resolve these two problems and so improve the forecast accuracy of the HMM. First, the application of extremum estimators to the HMM is proposed. Extremum estimators aim to improve the forecast accuracy of the HMM by minimising an estimate of the forecast error on the observed data. The forecast accuracy is measured by a score function and the use of some general classes of score functions is proposed. This approach contrasts with the standard use of a minus log-likelihood score function. Second, penalised estimation for the HMM is described. The aim of penalised estimation is to reduce overfitting and so increase the forecast accuracy of the HMM. Penalties on both the state-dependent distribution parameters and transition probability matrix are proposed. In addition, a number of cross-validation approaches for tuning the penalty function are investigated. Empirical assessment of the proposed approach on both simulated and real data demonstrated that, in terms of forecast accuracy, penalised HMMs fitted using extremum estimators generally outperformed unpenalised HMMs fitted using maximum likelihood.
- ItemOpen AccessTime series models for discrete data(1992) MacDonald, Iain L