Quantitative research · Daniel Kaufman
Kinetic AlphaApplied work on the parts of markets that don't fit a textbook.
A working portfolio: prediction-market portfolio margining, energy-futures risk decomposition, and adjacent microstructure problems — each shipped as a paper, a model, and a live dashboard you can drive.
Projects
Live dashboards and the research behind them.
A portfolio-margin framework for binary prediction markets
A correlation-aware margin engine for prediction-market portfolios: eight asset clusters across crypto, equity, macro, politics, and sports — joined by a generalized asset hierarchy and a multi-factor Monte-Carlo aggregating cluster ES into a single requirement.
- Clusters
- 8
- Coverage
- 99% ES
- Capital release
- ~71%
ICE energy futures contract decomposition
A risk-factor decomposition library for ICE energy contracts — crude, natural gas, power, refined products, NGLs, coal, and emissions — that expands every position (outright, strip, calendar spread, CSO, basis, bullet option) into dated factor legs verified cell-for-cell between a Python engine and the browser dashboard.
- Factors
- 108
- Contracts
- 107
- Python ↔ JS parity
- 32/32
How the work is built
Three rules that keep the dashboards honest.
Verified parity
Every browser engine is checked cell-for-cell against a Python reference. Drift between the calc you see and the calc you publish is a category-of-bug we don't tolerate.
Read the document first
Each project ships the underlying framework as a paper — assumptions, conventions, and the boring choices that drive the interesting outputs. Dashboards are illustrations, not substitutes.
Stress what's load-bearing
Every model exposes the levers that actually move requirements — correlations, thresholds, tenor structure — so you can find where the answer breaks before the market does.
Working on something adjacent?
Margining, clearing, risk decomposition, market structure — happy to compare notes or take a look at the problem.