A quantization approach to the counterparty credit exposure estimation

During recent years the counterparty risk field has received a growing attention because of the Basel Accord, which asks banks to fulfill finer conditions concerning counterparty credit exposures arising from banks’ derivatives, securities financing transactions, default and downgrade risks characterizing the Over The Counter derivatives market, etc. Consequently, the development of effective and more accurate measures of risk have been pushed, particularly focusing on the estimate of the future fair value of derivatives with respect to prescribed time horizon and fixed grid of time buckets. Common methods, used to treat the latter scenario, are mainly based on ad hoc implementations of the Monte Carlo approach, characterized by a high computational cost, being strongly dependent on the number of considered assets. This is why many financial players moved to more effective and time saving technologies, e.g., based on grid computing and Graphics Processing Units (GPU) capabilities. In the present paper we exploit an alternative approach based on different algorithmic strategies by showing how to implement the quantization technique to derive accurate estimate for both pricing and volatility values. Our approach turns out to produce sharp results for the counterparty risk evaluation, with great computational benefits if compared to the Monte Carlo approach.