Robert Jenkinson Álvarez

Curriculum Vitae

Education, quantitative research and trading experience, publications, and skills.

Quant CV Research CV

Education

MSc Financial Mathematics (Distinction)

London School of Economics and Political Science (LSE)

Sep 2023 – Jul 2024

  • Black-Scholes; Interest Rate/Credit Risk-Hull-White/Vasicek/HJM; Stochastic Processes-Ito; Fixed Income; Portfolio Management.
  • Algathon (Top 2): Built LSTM on BTC/ETH returns with unconventional sentiment data: 1.2 OOS Sharpe, 54% directional accuracy.
  • Analysed CRSP, TRACE/FISD and OptionMetrics across equities, credit and derivatives; yield-curve calibration and option hedging.
  • Finite-difference PDE solvers, variance-reduced Monte Carlo and Euler SDE simulation, analysing stability and convergence.

BSc Mathematics (First-Class Honours)

University of Bath

Sep 2020 – Jul 2023

  • Markov Processes; Martingales; Stochastic Systems; Discrete Probability; Machine Learning; Measure Theory; Statistics, Advanced Real/Complex Analysis; Algebra; ODEs/Control Theory; PDEs/Vector Calculus; Numerical Analysis (97%).

Quantitative research & trading experience

Quantitative Researcher

Sep 2024 – Present

Cross-Asset Hamiltonian Alpha Engine

  • Discovered incremental cross-asset volatility signal beyond model-free implied variance, forecasting 20-day relative realised variance with 0.232 locked-test rank IC, date clustered t=4.40 and positive coefficients in 13/13 asset/month omission tests.
  • Transformed option markets into arbitrage-consistent density, calendar and momentum states, achieving maximal primal/dual feasibility errors of 8.9×10−7/2.1×10−13, 100% projection convergence and 100% sigma-estimator completion.
  • Designed a stochastic Port-Hamiltonian JEPA capturing non-linear surface dynamics beyond damped mechanics, attaining 0.365 out-of-sample relative-state innovation R2 and 0.251 aggregate innovation R2 against the train-fitted baseline at 40 observations.

Chebyshev Global Options Surface Engine

  • Architected a global martingale-density/calendar-flow engine with a surface enforcing density positivity, unit mass/mean, strike convexity and calendar monotonicity with 99.77-100% inside-spread coverage and 0 static-arbitrage violations on 720x61 grids.
  • Outperformed independent SANOS on 100% of 272 material-stability comparisons, reducing gamma-curvature q95 by 87.5%, gamma finite-difference sensitivity by 81.1%, local-variance strike roughness by 82.7% and calendar-curvature RMS by 46.7%.
  • Compiled the certified global surface into analytic Chebyshev price/Greek caches, accelerating price-plus-Greeks evaluation by 11.1-14.4x, decreasing model size by 58-83% and cutting local-variance invalid-domain mass by 98.8-99.4% versus SANOS.

Machine Learning Researcher, NeurIPS 2026

Oct 2025 – Present

  • Developed HamJEPA [1], a phase-space JEPA with symplectic leapfrog/non-isotropic priors, deriving an optimal prior and isotropy bound, beating LeJEPA/SIGReg linear-probe/kNN-20 by 10.6/6.5 (80 ep.) on CIFAR-100 and 7.52/4.82 (45 ep.) on ImageNet-100.
  • Proved state-dimension and finite-bit lower bounds for cumulative learning, quantifying information beyond predictions [2].

Optiver

Derivatives Quant Trading & Market Making Trainee

Oct 2023 – Nov 2023

  • Developed Black-Scholes options market-making algorithms, adapting bid-ask spreads to liquidity and order-book depth, reducing daily PnL variance by 30% and 95% VaR by 25% via delta-hedging and gamma rebalancing (IOC orders).
  • Improved order placement latency by 40% with a priority queue architecture, using real-time Greeks caching for high-frequency updates and pruning low-liquidity strike/expiry pairs in a 3-person team on Optibook’s live simulated exchange.

Selected publications and manuscripts

  1. Beyond Isotropy in JEPAs: Hamiltonian Embeddings and Symplectic Prediction

    NeurIPS 2026: main-track accept – Poster.

  2. Prediction is not a Sufficient Memory State

    ICLR 2027 – Under Review.

  3. Variational rigidity for alternating symplectic billiards

    Submitted to Annals of Mathematics, 2026.

    Resolved an open billiard problem, proving integrability forces an inner-product ellipse without reversibility or finite symmetry.

  4. Rational-slope rigidity with identity holonomy through finite-amplitude resonant cells

    Submitted to ETDS, 2026.

    Established one-dimensional rigidity under identity holonomy via finite-amplitude classification of resonant variational cells.

  5. Minimizing foliations: counterexamples and common-volume rigidity

    Submitted to Annals of Mathematics, 2026.

    Answered the higher-dimensional question negatively and derived an exact rigidity criterion under common invariant volume.

  6. Arbitrage-free option price surfaces via Chebyshev tensor bases and a Hamiltonian fog post-fit

    arXiv:2512.01967, 2025.

    Built arbitrage-free surfaces via CVXPY/OSQP convex QP, achieving >98% inside-spread, <1% no-arb violations with 3D fog noise.

Skills & interests

Technical Skills: Python (NumPy, Pandas, PyTorch, TensorFlow, SciPy, QuantLib, OSQP), MATLAB, R, SQL, LaTeX.

Communication: Tutored metric spaces & stochastic processes to 5 university students, improving pass rates by 10%.

Languages / Interests: Native Spanish & English (CPE C2). Piano (ABRSM Grade 8), Giveaway chess (2220 Elo, top 100 worldwide).