A reality check in calculating VaR. It involves testing how well the VaR estimates would have performed in the past. For example, for calculating a 10-day 99% VaR, back testing would involve looking at how often the loss in ten days surpassed the 10-day 99% VaR that would been calculated for that period of ten days. If this happened on about 1-5% of the days, the methodology will be reasonably reliable for calculating VaR. If that happened on more than 5-7% (and more) of the days, the methodology will be doubtful.