Goodness-of-Fit Tests for Truncated Distributions - The Empirical Study
Krzysztof Echaust , Lach Agnieszka
AbstractSince an assumption of normality was rejected on financial markets, many heavy-tailed distributions were pro-posed in literature. Many researchers showed, that semi-heavy tailed distributions are the most suitable to de-scribe stocks or indices behaviour. However, modelling distributions of returns is of less importance around median than it is in the tails, where extreme events appear. Studies of tail thickness confirmed, that they are embraced between Gaussian and alpha-stable models. Our research also concentrates on the tails. We tried to approximate them with several distributions, covering the whole range of possibilities, from thin to fat. Addi-tional value comes from applied goodness of fit tests. When data is truncated, which is our case, testing goodness of fit using standard statistical tests is inappropriate. Instead, we suggest employing modified standard tests like AndersonDarling, Kolmogorov-Smirnov, Kuiper, Cramér-von Mises ones and tests designed specifically to measure the fit in tails. We conducted research based on 19 assets like indexes, currencies, stocks and future contracts, putting special emphasis on the Polish stock market.
|Publication size in sheets||0.5|
|Book||Pražák Pavel (eds.): 35th International Conference Mathematical Methods in Economics MME 2017 : Conference Proceedings, 2017, Gaudeamus, University of Hradec Kralove, ISBN 978-80-7435-678-0, 896 p.|
|Keywords in English||Goodness of fit tests, truncated distributions, fat tails|
|Score||= 15.0, 11-03-2020, ChapterFromConference|
|Publication indicators||= 0|
* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.