THESIS
2008
xiii, 80 p. : ill. ; 30 cm
Abstract
In this thesis a fast and robust method for calibrating the caplets is introduced. The conventional method for calibrating the parameters generally minimizes the error between the model data and the market data. The brute force method is robust but too slow for practical use. Other searching methods such as the Nelder-Mead method is fast but not robust enough for high dimensional problems. As there are 3 parameters to calibrate in the stochastic LIBOR model, a new method is required. The proposed method is innovative because it solves the parameters numerically, which avoids the time consuming method in brute force and instability in searching. The proposed method is a two stage process, which combines the newly invented method and the existing knowledge in minimization. Under some reas...[
Read more ]
In this thesis a fast and robust method for calibrating the caplets is introduced. The conventional method for calibrating the parameters generally minimizes the error between the model data and the market data. The brute force method is robust but too slow for practical use. Other searching methods such as the Nelder-Mead method is fast but not robust enough for high dimensional problems. As there are 3 parameters to calibrate in the stochastic LIBOR model, a new method is required. The proposed method is innovative because it solves the parameters numerically, which avoids the time consuming method in brute force and instability in searching. The proposed method is a two stage process, which combines the newly invented method and the existing knowledge in minimization. Under some reasonable assumptions, the proposed method can generally work for such kinds of problems and calibrate the caplets instantly.
Post a Comment