Hi, I'm Sam.

Developed statistical models to estimate return to work probability for different KPI thresholds using longitudinal case and operational data, serving clients across a broad range of industries and disciplines. Rigorously validated the model out-of-sample and built a triage prioritisation engine based on the insights from the model.

Professional work. Technical description excludes confidential data and implementation details.

Systematic research pipeline to test whether information in the US treasury curve and commodity markets offer predictive utility about forward returns across equities, fixed income, commodities, and FX.

Constructs rolling yield curve, momentum, and mean-reversion signals and screens candidate predictors across multiple time horizons using statistical analysis and converts selected forecasts into bounded trading positions. A final hold-out backtest is then run.

Inspired by previous work led by Dr Brent Davis, I am researching the use of Discrete Time Markov Chains and Hidden Markov Models in attempting to decipher scripts and whether or not this could be applied to ancient Minoan Linear A.

This project is ongoing to mixed results as one might expect, but it is always exciting to be greeted by an opportunity to work at the intersection of my professional and extracurricular interests.

Built a modular derivatives pricing engine for barrier, multi-asset, path-dependent, and early-exercise derivative securities. The project compares finite differencing PDE methods, Monte Carlo simulation, variance reduction techniques, and other numerical mehtods according to the structures and dimensionalities of each modelled product.

The engine prices Asian, barrier, American, and basket derivatives, computes numerical option greeks, and produces confidence intervals.

Built and continue to develop a data-driven system for pricing thoroughbred horse races in Australia that collects and standardises races, runner and market data, and produces probability estimates. I am then able to discretionarily compare these modelled probabilities against available market odds to identify potential value in the form of mispricings.

NB: my interest in horse racing is primarily from a modelling and market-pricing perspective rather than as an avid bettor. I do not endorse the significant animal welfare and broader ethical issues associated with the industry, and any use of betting markets should be approached responsibly.

My girlfriend loves to read, as well as to buy books she may one day read but can't quite get around to yet. With many hundreds of books to keep track of she was starting to get a bit tired of having to, let's say, O(n) search (she ruled out my alphabetical bookshelf system long ago) her entire bookshelf to find if she already owns a title before going shopping. Behold, Bookcase, a simple website I built while she was away that allows her to file every book on her shelf, every book she would like to put on her shelf, and every book she has read but has not quite made it to the shelf.

It gives you real-time reading progress statistics for series, top genre information, periodic reading stats such as "You've read 10 books this year", ratings, wish lists, etc., and is an ongoing project I continue to build with her guidance on what she would like it to do.

The next stage I am currently working on for this project is implementing a recommendation engine that, based on a blend of long-term and short-term historical reading habits, identifies what books she has not read that she might like.