Library: | VaR |
See also: | VaRest VaRopt VaRpred VaRdiagtable |
Quantlet: | VaRver | |
Description: | verifies probability forecasts |
Usage: | stat = VaRver(tab) | |
Input: | ||
tab | n x 3 matrix, frequency table of forecasts and its success. | |
Output: | ||
stat | list containing various statistics suitable for the verification of probability forecasts: "stat.miscal": which is measure of miscalibration "stat.miscreg": the regression based on miscalibration "stat.varf": variance of forecasts "stat.resol": measure of resolution "stat.diffm": difference in conditional means "stat.sigfx1", "stat.sigfx0": conditional variances of forecasts "stat.me": mean error (ME) "stat.biasreg": regression based measure of bias "stat.mse": mean square error (MSE) Brier score "stat.ss": skill score "stat.potskil": measure of potential skills of a perfectly calibrated procedure |
library("VaR") x=read("kupfer") x=x[1:1001] y=diff(log(x)) ; returns sig=VaRest(y)[,2]/qfn(0.99) y=y[251:1000] table=VaRdiagtable((-0.01~0.01),sig,y,10) stat=VaRver(table) setenv("outputformat","% 8.5f") stat.mse stat.me
MSE and ME of the forecasts: Contents of mse [1,] 0.23450 Contents of me [1,] -0.06667