It confuted one-day VaR for the petroleum products and their nearby futures at 99% and 95% confidence levels using parametric, semi-parametric and non-parametric approaches. For each VaR calculation it computed the corresponding expected Shortfall (ES). For backtesting VaR it used Christoffersen's conditional coverage test and for backtesting the ES it used a quantratic score function based on a loss function. In order to decide on the best method for VaR computation, it followed two approaches and compared them to check whether they yield similar results.