How historical simulation made me lazy
categories: RMA, FINGER Christopher, Research Monthly
In this month's note, we discuss a somewhat disturbing trend: the movement toward utilizing historical simulation as the sole statistical measure of risk. Though the method does have its benefits, we argue that few of these relate to producing good forecasts. Further, one of the primary benefits cited, that historical simulation somehow requires fewer assumptions than its alternatives, is misguided. Thus, our note is an appeal that just because statistical forecasting is difficult, we should not give up; there is much to be gained by revitalizing the debates on risk forecasting we used to have regularly.