In his PhD thesis, CWI PhD student Alvaro Leitao Rodriguez proposes new methods to tackle complex problems in the financial sector. With these methods, Leitao Rodriguez successfully simulates the movements of the interest rates in the Foreign eXchange (FX) markets and evaluates corresponding risks. Leitao Rodriguez will defend his thesis on June 27th, at the Delft University of Technology.
To shed light on complex problems in the financial sector, Leitao Rodriguez uses so-called Monte Carlo-based hybrid methods. Monte Carlo methods are based on the simulation of possible future scenarios, and the aggregation of their values for a possible solution. They are widely used in many mathematical problems. However, ‘traditional’ Monte Carlo methods are often not sufficient to solve modern and complex problems in the financial sector. In that case, hybrid methods, in which Monte Carlo methods are combined with other methodologies, can provide the answer. For hybrid methods, one can rely on mathematical approaches, computational approaches, or a combination of both.
The hybrid methods Leitao Rodriguez proposes, provide a way to use approximations to recover expressions for statistical distributions. Leitao Rodriguez thus managed an efficient and 'exact' simulation of complex financial models. For instance, he could simulate the movements of the interest rates in the Foreign eXchange (FX) markets and the risks associated to these fluctuations. Also, his technique could be employed for simulating the possibility of bankruptcy in a financial institution.
The complex problems of the financial world can be considered extreme cases, because these are very high-dimensional problems. To be able to use the hybrid Monte Carlo techniques, Leitao Rodriguez therefore had to employ some cutting-edge scientific computing advances, such as high-performance computing on GPUs. This allowed for fast and robust simulations in the context of multi-dimensional models, efficiently handling complex financial problems in 10D, 20D or even 50D.
Defense: 27/06/2017 - 13:00 - 14:00 hrs
Location: Senaatszaal of the Auditorium, Mekelweg 5, Delft
Promotor: Prof.dr. Kees Oosterlee
Thesis: ‘Hybrid Monte Carlo methods in computational finance’
