20.500.12556/RUL-149756 Kopule in upravljanje s tveganji magistrsko delo Copulas and risk management Obstaja več načinov, kako lahko merimo stopnjo tveganja portfelja. Eden od teh načinov je s kopulami. V tej nalogi bomo spoznali kopule in pokazali razliko med izbiro različnih kopul pri upravljanju s tveganji. S testi za prileganje porazdelitev podatkom bomo videli, da se najbolje izkaže uporaba Studentove kopule. Tudi Gaussova kopula nakazuje sprejemljive rezultate, medtem ko nam da Gumbelova kopula nekoliko slabše rezultate. Pri analizi skupnega gibanja cen navzdol oziroma navzgor vidimo, da glede na Studentovo kopulo ostali dve podcenjujeta verjetnost, da bo prišlo do ekstremnih dogodkov. Pri analizi razpršitve portfelja pa se izkaže, da nam Gaussova kopula da bolj optimistične napovedi, Gumbelova pa bolj pesimistične v primerjavi s Studentovo kopulo. There are several ways to measure the portfolio risk. One of these ways is using copulas. In this article, we will learn about copulas and show the difference between choosing different copulas in risk management. With goodness-of-fit tests for distributions we will see that the use of Student's copula performs best. Gaussian copula also indicates acceptable results while Gumbel copula gives us worse results. When analyzing the joint extreme downward and upward movements, we see that according to Student's copula, the other two underestimate the probability that extreme events will occur. When analyzing portfolio diversification benefits, it turns out that the Gaussian copula gives more optimistic predictions, while Gumbel's gives more pessimistic predictions compared to the Student's copula. kopula tveganje odvisnost slučajnih spremenljivk upravljanje s tveganji Gaussova kopula Studentova kopula Gumbelova kopula copula risk dependence of random variables risk management Gaussian copula Student copula Gumbel copula true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2023-09-09 08:15:02 2023-09-09 08:15:05 2024-05-29 10:43:08 0000-00-00 00:00:00 2023 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 519.2 135603 163759363 7787.pdf 7787.pdf 1 957B9B99DCA9ED423CED1DCAF313C1FA b0242dd7e64ce8d1091bb43aa10c419b3df5e7fd21459f0463179f22f1f5622b 3757c700-4ed8-11ee-b6f1-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=174256 Fakulteta za matematiko in fiziko 0 0 0