Selected work
A sample of the systems and research behind the advisory practice. Methods and write-ups available on request.
A desktop application valuing a trading portfolio across futures, stocks, and options in real time — computing VaR, CVaR, and stress scenarios as prices move, not in an overnight batch. Designed, engineered, and operated end-to-end.
Daily analytics across crude, natural gas, and refined products: term-structure dynamics, near-minus-far calendar spreads, seasonal patterns, and refining crack spreads — the lens used for the daily market read.
Pair-spread research with honest verdicts — including studies that killed the idea. The ZB/GC spread failed validation (unstable hedge, no convergence); the ZB/ZN curve pair validated on correlation and mean-reversion diagnostics. Negative results included, always.
Method
Every study starts from a tradeable question and ends with a verdict — including "don't trade this." The machinery is Python-based, data comes from institutional-grade feeds, and results are documented with their assumptions, costs, and limitations stated up front.
That discipline is what advisory clients are buying: not cleverness, but the habit of killing bad ideas early and sizing the survivors honestly.