QLIKE (Patton 2011): mean(r/f − ln(r/f) − 1) over variances. ≥ 0, 0 at r = f, and harder on UNDER-prediction — the dangerous direction for a risk read. A zero realised variance is floored so one flat window cannot blow up the mean.
QLIKE (Patton 2011): mean(r/f − ln(r/f) − 1) over variances. ≥ 0, 0 at r = f, and harder on UNDER-prediction — the dangerous direction for a risk read. A zero realised variance is floored so one flat window cannot blow up the mean.