Non-homogeneous regression models are widely used to statistically post-process numerical ensemble weather prediction models. Such regression models are capable of forecasting full probability distributions and correct for ensemble errors in the mean and variance. To estimate the corresponding...

Non-homogeneous post-processing is often used to improve the predictive performance of probabilistic ensemble forecasts. A common quantity to develop, test, and demonstrate new methods is the near-surface air temperature frequently assumed to follow a Gaussian response distribution. However,...

We define risk spillover as the dependence of a given asset variance on the past covariances and variances of other assets. Building on this idea, we propose the use of a highly flexible and tractable model to forecast the volatility of an international equity portfolio. According to the risk...

Since its introduction in 2003, volatility indices such as the VIX based on the model-free implied volatility (MFIV) have become the industry standard for assessing equity market volatility. MFIV suffers from estimation bias which typically underestimates volatility during extreme market...

Accurate prediction of future claims is a fundamentally important problem in insurance. The Bayesian approach is natural in this context, as it provides a complete predictive distribution for future claims. The classical credibility theory provides a simple approximation to the mean of that...

The bootstrap is, at heart, a way to obtain an approximate sampling distribution for a statistic (and hence, if required, produce a confidence interval). Where that statistic is a suitable estimator for a population parameter of interest, the bootstrap enables inferences about that parameter. In...

The pricing of longevity-linked securities depends not only on the stochastic uncertainty of the underlying risk factors, but also the attitude of investors towards those factors. In this research, we investigate how to estimate the market risk premium of longevity risk using investable...

We define the nagging predictor, which, instead of using bootstrapping to produce a series of i.i.d. predictors, exploits the randomness of neural network calibrations to provide a more stable and accurate predictor than is available from a single neural network run. Convergence results for the...

In this paper we show how to quantify the uncertainty in the difference between the best estimate for the ultimate claim viewed at the beginning and at the end of one year. A second aspect in this paper is how bootstrapping techniques can be used to simulate these uncertainty for several...

We show that by modeling the time series of mortality rate changes rather than mortality rate levels we can better model human mortality. Leveraging on this, we propose a model that expresses log mortality rate changes as an age group dependent linear transformation of a mortality index. The...