where p = q + r + 1 is the dimension of ˇ and l is a function of a finitenumber of lagged observable values. The autoregressive termsˇiVaRt−i(ˇ) i = 1,. . ., q ensure that the quantile changes “smoothly”over time. The role of l(rt − 1) is to link VaRt(ˇ) to observable vari-ables that belong to the information set. A natural choice for xt − 1is lagged returns. The advantage of this method is that it makes nospecific distributional assumption on the return of the asset. Theysuggested that the first order is sufficient for practical use :