Value-at-risk can be broadly defined as follows: if the density function f(w) of the future value of portfolio W is known, VaR is the worst possible realization of the W* portfolio value such that, at a given trust level c the likelihood of attaining a higher value is c.

Owing to some of its features (such as the fact that it measures the risk of an organization´s entire operational activity), VaR is readily applied in risk reporting, in decisions concerning the allocation of limited capital resources as well as in the assessment of the operation of profit centers.

While the above definition provides a broad platform for the concept, the specific implementation methods differ significantly. However, two basic approaches can be distinguished amongst them. The first one is based on the local measures of risk. This category contains, among others, the portfolio-, asset- and delta-normal methods. These methods require that certain simplifying assumptions be made as to, on the one hand, the estimation of changes in a portfolio´s market value following the changes in market prices, and on the other, the probability distribution of fluctuations of market variables. The necessity to make these assumptions attracts some criticism of the approach. Besides, they have a significant bearing on the concluding phase. The other approach involves the so-called full valuation models. Under this heading we can list the historical simulation model, the Monte Carlo and scenario simulations. Full valuation models require a revaluation of a portfolio for every series of prices. While this approach is more accurate, it also demands vast computation capabilities. In addition, the Monte Carlo method, in which a specific stochastic process is assumed, is sensitive to model risk. The risk can be eliminated by applying a draw from among historical data (the bootstrap method ). The methods listed above are insufficient under the conditions of sudden and significant changes to the market situation. In this case, specially defined simulations are called for.