PhD in Management Science
Shane joined the Management Science Group in October 2010 to begin his PhD. Prior to this, he obtained a BEng/MEng degree in Electrical and Electronic Engineering, from Imperial College London in 2008, during which he learned operational research and digital processing techniques and used them to complete his thesis in financial time series tracking. Since graduating from Imperial he has been conducting research with Nicos Christofides, Professor Emeritus of Quantitative Finance at Imperial College, during which he has gained additional experience in financial modelling, approximate dynamic programming and conjugate gradient search techniques.
The primary objective of Shane's research is to design a neural network (NN) structure using combinatorial optimization algorithms (both exact and heuristic) to evaluate the potential of using the technique in forecasting trends for financial time-series. This has an impact on the computation of risk exposures to market risk over short horizons, as well as being of obvious advantage to portfolio selection and other investment activities.
Mathematical modelling and trend forecasting for stochastic financial time series has long been a popular research approach. Most methodologies have involved statistical models such as regression analysis. These linear models are unable to deal with the complex and generally nonlinear interactions that determine the asset price movements and their track record in forecasting is very poor. In Shane's approach he deals with both nonlinearities and economic regime changes by designing an adaptive approximate dynamic programming system that changes its structure based on endogenous and exogenous information that flows through the neural network.
Shane's key areas of interest are: dynamic programming; conjugate gradient search;
pricing financial products; time series tracking and trend forecasting.