Dr. Egloff defined the bank as a risk restructuring machine that acquires risk from the market in order to restructure the risk and resell it to the investors. In this process restructuring may cause a gap or mismatch, and hence create residual risk, the speaker explained. Therefore bank management has to gain a thorough understanding of how to price risk and seek methods to quantify future unexpected losses from residual risk.

This is where high performance computing comes in to cope with the challenge of pricing complex financial products. This attempt of pricing is related to multiple assets with correlations between them, which may have a partial but yet important influence on the bank's balance sheet.

In addition, bank managers have a responsibility to analyse and quantify rare event risks such as credit and operational risk. Fortunately, events such as bankruptcy of an obligor, fraud, or an IT platform failure appear

infrequently. On the contrary, effects such as financial losses and reputation damage may be very substantial if they occur, noted Dr. Egloff. This occurrence naturally increases in a distressed market environment or an economic down turn.

Banks tend to use massive Monte Carlo simulations to calculate their risks. In the trading business product pricing can be sped up in order to shorten the response time and to increase the throughput. As far as risk management is concerned, the speaker explained that the computer allows to estimate losses from credit and operational risk and to calculate the economic capital for the market, credit and operational risk.

HPC is also used for the capital management and bank steering to optimize the capital structure and the risk return profile. Simulations provide insight in the analysis of the effect caused by tailor made off balance sheet transactions on economic and regulatory capital requirements. Last but not least, advanced computational power pays great services in improving the return on equity. As such the capital can be allocated across business lines and the bank is able to comply with regulatory capital requirements.

Dr. Egloff further elaborated on the game with the Monte Carlo simulations. This largely consists in rolling the dice to simulate possible future outcomes of losses and profits as well as to simulate the various factors in a distressed market environment.

In this type of exercises, however, supercomputing does nor can not provide a magical formula. Dr. Egloff cited Goethe's Gretchen to express his reservations about HPC: "Now tell me, how about you and religion (= HPC)? You seem to be a good and kind man but still, I believe you have not much up with religion."

Or to put it in ecomomical terms: Can all these sophisticated products and models, which require massive