The esback can be used to backtest forecasts of the expected shortfall risk measure.
You can install the released version from CRAN via:
install.packages("esback")
The latest version of the package is under development at GitHub. You can install the development version using these commands:
install.packages("devtools")
devtools::install_github("BayerSe/esback", ref = "master")
This package implements the following backtests:
- Expected Shortfall Regression Backtest (Bayer & Dimitriadis, 2020)
- Exceedance Residuals Backtest (McNeil & Frey, 2000)
- Conditional Calibration Backtest (Nolde & Ziegel, 2017)
# Load the esback package
library(esback)
# Load the data
data(risk_forecasts)
# Plot the returns and expected shortfall forecasts
plot(risk_forecasts$r, xlab = "Observation Number", ylab = "Return and ES forecasts")
lines(risk_forecasts$e, col = "red", lwd = 2)
# Backtest the forecast using the ESR test
esr_backtest(r = risk_forecasts$r, e = risk_forecasts$e, alpha = 0.025, version = 1)