Financial Engineering

Our research in financial engineering focusses on derivative pricing; and applying machine learning techniques  time series prediction and the hedging of foreign exchange risk.

Managing foreign exchange risk

ARC Linkage Grant: LP110200413

Partner/Collaborating Organisation: Westpac Banking Corporation

Financial institutions are required to absorb risks associated with foreign exchange (FX) transactions. This work involves applying recent parallel computing and machine learning techniques to better understand and manage such exposure. An environment is being developed to facilitate the testing of risk management strategies and provide an interface to scalable computational resources. The environment will be used to develop improved techniques for: customer flow and exchange rate prediction, hedging strategies, and FX market models. This research will potentially enable national banks to better quantify and manage risk, allowing them to be more competitive in global FX markets.

[1] Farzad Noorian, Barry Flower, and Philip H.W. Leong. Stochastic receding horizon control for short-term risk management in foreign exchange. Journal of Risk, 2015. In press.

[2] Anthony Mihirana de Silva, Richard I.A. Davis, Syed A. Pasha, and Philip H.W. Leong. Forecasting financial time-series with grammar guided feature generation. Computational Intelligence, 2016. (doi:10.1111/coin.12083)

[3] Farzad Noorian, Anthony M. de Silva, and Philip H.W. Leong. gramEvol: Grammatical Evolution in R. Journal of Statistical Software, 2015. In press.

Cluster and FPGA-based Acceleration of Derivative Pricing

We have been collaborating for many years with Cluster Technology to develop The Clustertech Parallel Environment (CPE) [1], a C++ software platform which facilitates the development, deployment and execution of parallel applications. CPE provides optimized domain-specific libraries for Monte Carlo (MC) simulation and for the solution of partial differential equations using the finite difference (FD) method, which greatly reduces the development time of parallel codes. It greatly simplifies the development of parallel derivative pricing products and is production usage in several banks. 

We have also developed high speed FPGA-based derivative pricing hardware [2] which can be used for portfolio analysis and algorithmic trading. Key to this are efficient Gassian random number generators [3].

[1] M.P. Leong, C.C. Cheung, C.W. Cheung, P.P.M. Wan, K.H. Leung, W.M.M. Yeung, W.S. Yuen, K.S.K. Chow, K.S. Leung, and P.H.W. Leong. A parallel library for financial engineering applications. IEEE Computer, 38(10):70–77, October 2005.

[2] G. L. Zhang, P. H. W. Leong, C. H. Ho, K. H. Tsoi, C. C. C. Cheung, D-U. Lee, R. C. C. Cheung, and W. Luk. Reconfigurable acceleration for monte carlo based financial simulation. In Proc. International Conference on Field Programmable Technology (FPT), pages 215–222, 2005.

[3] David B. Thomas, Wayne Luk, Philip H.W. Leong, and John D. Villasenor. Gaussian random number generators. ACM Computing Surveys, 39(4):11:1–11:38, 2007.