Why this site exists
LinkedIn rewards short, polished bullet points. Most of the work that goes into a model — the reason a particular method was chosen, the assumption that almost broke it in production, the analytical detour that turned out to be the actual insight — doesn't survive that compression.
This site is the long-form record. Six engagements are written up under Experience; each one includes the methods that worked, the methods that quietly didn't, and the reasoning behind the choice.
The thesis
Most quantitative tooling is built for the convenient case: stationarity, finite variance, independent shocks. The systems I build live in the inconvenient case — heavy tails, regime shifts, structural breaks, rare events that disproportionately drive outcomes.
That's a methodological choice, not a marketing one. It's why the philosophy section is on the homepage and not buried in a footer. Six principles, six honest constraints; you can read them here.
What to expect
I'll post here irregularly — short notes on:
- Specific methods (fractal estimators, EVT tail diagnostics, regime classifiers) and where they break
- The recurring places where standard quantitative tooling silently fails on real data
- Occasional retrospectives on engagements once they've shipped and embargoes lift
- Things I'm reading that I think are good
No SEO, no posting cadence, no list-building. If a piece of writing is useful, it goes up.
If something here resonates — for collaboration, hiring, or argument — the comments are open below, or my contact is on the homepage.
Comments
No account required. Comments are moderated — they appear once approved.