Description Usage Arguments Value
This function implements multilevel unconditional coverage tests of VaR forecasts for different quantiles. These tests achieve a higher statistical power then multiple unilevel tests.
Theory based on "Evaluating the accuracy of value-at- risk forecasts: New multilevel tests" by Leccadito, Bofelli & Urga (2014).
Code for the multilevel Christoffersen UC test is based on the function "VaRTest" from the "rugarch" package by Alexios Ghalanos.
1 | multilevel_VaRTest(alphas, actual, VaR, confidence, m = 5, B = 2000)
|
alphas |
vector of quantiles |
actual |
vector of actual returns |
VaR |
matrix of VaR forecasts. Order needs to be the same as in alphas. |
confidence |
confidence level for unconditional coverage tests |
m |
number of lags for multilevel Pearson's chi-square test. Leccadito et. al recommend a lag of 5. |
B |
number of Monte-Carlo simulations for Pearson's chi-square test. |
a list with the following items:
Christoffersen conditional coverage test LR statistic.
Christoffersen conditional coverage test critical value.
Christoffersen conditional coverage test p-value.
Christoffersen conditional coverage test decision on H0 given the confidence level.
Christoffersen unconditional coverage test LR statistic.
Christoffersen unconditional coverage test critical value.
Christoffersen unconditional coverage test p-value.
Christoffersen unconditional coverage test decision on H0 given the confidence level.
Pearson's chi-square test statistic
Pearson's chi-square test p-value
Pearson's chi-square test decision on H0 of given the confidence level.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.