Aarhat Multidisciplinary International Education Research Journal (AMIERJ) Page 1 SJIF Impact Factor 6.236 Peer Reviewed Refereed Journal AMIERJ Volume–VIII, Issues–II ISSN–2278-5655 March - April 2019 VARIANCE – COVARIANCE (DELTA NORMAL) APPROACH OF VAR MODELS:AN EXAMPLE FROMBOMBAY STOCK EXCHANGE Prof. Raghavendra S Bendigeri Assistant Professor, Finance& Marketing, Oriental Institute of Management, VashiNavi Mumbai Abstract Numerous investors are inclined to understand, the quantum of wealth or capital they can lose in a specific time period, which could be one day or 5 days or 10 days.In this research paper, out of numerous approaches, variance – covariance approach of VaR is discussed.This method helps in prediction of maximum loss that can occur for a specific time period and given probability. Here in order to calculate VaR, portfolios are created, which is followed by identification of returns distribution. Finally VaR of portfolios is calculated. Daily loss is calculated using data for the period of 01 st January 2018 to 31 st December 2018as historical data consisting of 246 days. Companies were selected from Bombay Stock Exchange (BSE). VaR has been computed for both 95% and 99% confidence intervals for holding period of 1 day and 10 days. Keywords : Risk & Return, VaR, Maximum Loss, Variance – Covariance approach, Correlation. 1. Introduction The key yardsticks or benchmark for any investment happens to be Risk and Return. The primary assumption w.r.t investors are that they want to maximize their utility and reduce risk. In other words they are risk averse. From that perspective, risk management has emerged as a vital facet for evaluation and choice of investments. Numerous organizations and regulatory authorities too attribute tremendous significance to analysis and measurement of risk in the aftermath of 2008 financial crisis. A number of methods exist to compute market risk, which primarily consists of exchange rates, interest rates etc. Value at Risk models (VaR) have been implemented since 1994. The primary disadvantage of VaR models is that they estimate or predict only those risks that can be quantified. Qualitative risks such as regulatory risk, operational risk or political risks cannot be measured using this model.VaR model estimates the highest loss for a portfolio or a stock w.r.t a given holding period and a level of confidence. This gives investors an opportunity to know, how much they stand to lose in a particular time frame. Different results are thrown up by different models.VaR models have both advantages as well as disadvantages. The classification of VaR models can be parametric or non – parametric. Variance – Covariance model (also known as delta normal) is the most widely used VaR model in finance.