SunGard creates real-time credit risk analytics

SunGard Trading and Risk Solutions, a unit of US technology provider SunGard, has built a new credit tool that can offer traders access to virtually real-time risk exposures.

The new product, Credient MC2, effectively approximates current Monte Carlo simulation, but can provide results for a portfolio of 1,000-plus deals within a second. Traditional Monte Carlo would take about an hour.

Nick Lea, head of Credient analytics, told RiskNews that the patent-pending product was designed for use at large investment banks that need an immediate portfolio credit model for trading decisions.

Credient MC2 is most suited to netted credit porfolios, but also works on un-netted portfolios. Lea claimed that in the case of linear, netted credit porfolios, accuracy is within 1% of full Monte Carlo simulation. For non-linear netted portfolios, the margin of error rises to between 3% to 4%; while non-linear, non-netted portfolios can produce a margin of error of about 5%.

Credient MC2 works by fitting a hyper-dimensional surface to the portfolio's pay-off function, said Lea. This then provides a tractable analytics solution. In effect, Lea and his team have developed complex mathematical algorithms that can transform highly complex trading portfolios into a simplified version, without losing too much accuracy.

"Credient MC2 also calculates credit exposure at all the salient points in the life of a portfolio, but is able to do so in real-time because of the proprietary algorithms used," said Lea. "Unlike other alternative approaches to Monte Carlo simulation, no compression techniques are used, thus preserving the effects of portfolio diversification and opposing sensitivities to market variable."

But like Monte Carlo, Crediient MC2 is also vulnerable to certain 'pathological portfolios' that can "knock the system", said Lea. This includes butterfly straddles with very small peaks.