How I work

Markets first, code second

Discretionary trading taught me what a chart costs when you're wrong. I build models to answer questions the market already asked me — not to go hunting for patterns in noise. The intuition comes first; the Python is how I check whether the intuition survives.

Regimes, not averages

A single number over a decade hides the two months that mattered. Most of my research is about detecting when the rules of the game have quietly changed — hidden-Markov states, correlation structure, tail behaviour — and sizing accordingly.

Report it straight

Every figure on this site is labelled: live or backtest. A strategy at 0.3 Sharpe is a strategy at 0.3 Sharpe, and I'll say so before you ask. I'd rather be trusted than impressive.