This paper describes, analyzes and evaluates an algorithm for estimating po
rtfolio loss probabilities using Monte Carlo simulation. Obtaining accurate
estimates of such loss probabilities is essential to calculating value-at-
risk, which is a quantile of the loss distribution. The method employs a qu
adratic (''delta-gamma'') approximation to the change in portfolio value to
guide the selection of effective variance reduction techniques; specifical
ly importance sampling and stratified sampling. If the approximation is exa
ct, then the importance sampling is shown to be asymptotically optimal. Num
erical results indicate that an appropriate combination of importance sampl
ing and stratified sampling can result in large variance reductions when es
timating the probability of large portfolio losses.