One-step VaR walked forward: at each origin, fit on [0, origin], forecast
the VaR of return origin + 1 from s_{T+1} and the fit's μ and innovations
(or, method: 'fhs', its standardised residuals), and count
a violation when the return fell below −VaR. The windows are
validationWindows' (minTrain 500, step 10 by default; step 1 tests every
day, at one fit per day).
One-step VaR walked forward: at each origin, fit on [0, origin], forecast the VaR of return origin + 1 from s_{T+1} and the fit's μ and innovations (or,
method: 'fhs', its standardised residuals), and count a violation when the return fell below −VaR. The windows arevalidationWindows' (minTrain 500, step 10 by default; step 1 tests every day, at one fit per day).