Algorithmic trading
Demonstrates end-to-end value extraction. Data farming, domain knowledge, and visualisation identified arbitrage opportunities. A fully automated trading pipeline was implemented. Arbitrage profit was locked-in with hedging. Project concluded net-positive.
Model implementation
Demonstrates implementation of large statistical software, a multivariate regression program for high-dimensional count data. Written in R, C, and CUDA. Maximum likelihood estimates are calculated with a non-linear optimisation program. Numerical algorithms were optimised for GPU computing.