We describe a novel method of heavy tails estimation based on transformed score (t-score). Based on a new score moment method we derive the t-Hill estimator, which estimates the extreme value index of a distribution function with regularly varying tail. t-Hill estimator is distribution sensitive, thus it differs in e.g. Pareto and log-gamma case. Here, we study both forms of the estimator, i.e. t-Hill and t-lgHill. For both estimators we prove weak consistency in moving average settings as well as the asymptotic normality of t-lgHill estimator in iid setting. In cases of contamination with heavier tails than the tail of original sample, t-Hill outperforms several robust tail estimators, especially in small samples. A simulation study emphasizes the fact that the level of contamination is playing a crucial role. The larger the contamination, the better are the t-score moment estimates. The reason for this is the bounded t-score of heavy-tailed distributions (and, consequently, bounded influence functions of the estimators). We illustrate the developed methodology on a small sample data set of stake measurements from Guanaco glacier in Chile.

We study ARCH/GARCH effects under possible deviation from normality. Since skewness is the principal cause for deviations from normality in many practical applications, e.g. finance, we study in particular skewness. We propose robust tests for normality both for NoVaS and modified NoVaS transformed and original data. Such an approach is not applicable for EGARCH, but applicable for GARCH-GJR models. A novel test procedure is proposed for the skewness in autoregressive conditional volatility models. The power of the tests is investigated with various underlying models. Applications with financial data show the applicability and the capabilities of the proposed testing procedure.

Different questions related with analysis of extreme values and outliers arise frequently in practice. To exclude extremal observations and outliers is not a good decision, because they contain important information about the observed distribution. The difficulties with their usage are usually related with the estimation of the tail index in case it exists. There are many measures for the center of the distribution, e.g. mean, mode, median. There are many measures for the variance, asymmetry and kurtosis, but there is no easy characteristic for heavy-tailedness of the observed distribution. Here we propose such a measure, give some examples and explore some of its properties. This allows us to introduce classification of the distributions, with respect to their heavy-tailedness. The idea is to help and navigate practitioners for accurate and easier work in the field of probability distributions. Using the properties of the defined characteristics some distribution sensitive extremal index estimators are proposed and their properties are partially investigated.