Research
Synthesises perspectives across academic research, practitioner reports, and regulatory guidance to evaluate how ML models behave under distribution shift and what constitutes robustness in financial forecasting. Proposes an integrated framework for evaluating robustness in financial time-series forecasting.
Implements an automated system for detecting and analysing currency exchange arbitrage opportunities. Exchange rates are modelled as a directed weighted graph; the Bellman-Ford algorithm is applied in −log(rate) space for negative cycle detection, with a profit threshold to suppress floating-point false positives. Includes a live exchange rate pipeline via exchangeratesapi.io.