Banks have to manage a variety of trading risks and over the past 20 years a number of the tools have been developed to better quantify these risks and their potential impact on the bank. Of these the most widely used is the “value-at-risk” approach, commonly referred to as VaR and sometimes as VAR. For the sake of clarity we will use VAR to denote the actual value at risk and VaR the technique itself in this text. VaR has the following features:

VaR uses historic data to determine past price volatility and relationships between variables.
It depends to a large extent on historic correlations between instruments being maintained.
VAR is usually quoted as an absolute dollar amount.
VaR requires that the time frame over which the VAR estimate is made be specified. This  time frame is often referred to as the holding period.
VAR is defined at a specific confidence level expressed in terms of probabilities.
VAR is also highly model dependent and two banks estimating VAR for an identical portfolio of financial assets will almost always end up with different estimates even though the same holding period and confidence level have been specified.