Quantitative Finance.
The models that price, hedge, and trade — treated like they matter. One practice spanning governed machine learning, systematic strategies from research through execution, and the rates & exotics modelling stack from LMM to smile risk.
A model is a position. Manage it like one.
Machine learning is already pricing, surveilling, and allocating on institutional desks — often with less governance than a spreadsheet. This practice brings the model-risk rigor of a Tier-1 validation function to ML, and teaches your teams to hold it.
The applied curriculum and advisory track: what machine learning genuinely can and cannot do on a trading desk — supervised and unsupervised methods, feature pipelines, and deployment realities — framed for the people who own the P&L, not the papers.
Validation-grade governance extended to machine learning: model inventory, independent validation, drift and performance monitoring, challenger models, and documentation that survives an audit.
Regime shifts, synthetic scenarios, and adversarial inputs — finding where the model breaks before the market demonstrates it for you.
Bias, explainability, and reproducibility reviews with audit trails a regulator can walk — from feature lineage to decision logs.
Twelve disciplines, one production line.
A systematic strategy is only as strong as its weakest stage. The lab covers the full line — from research hypothesis to executed fill — with the same evidence-first discipline the rest of the house runs on.
Hypothesis to signal — data sourcing, feature design, and the research loop that separates signal from story.
Point-in-time data, realistic costs, out-of-sample discipline — backtests built to be distrusted.
Sharpe, drawdown, capacity, decay — attribution that says where the P&L actually came from.
Position sizing and portfolio construction — risk parity, volatility targeting, constraint handling.
Cross-sectional cheapness, carry, and quality composites — priced against fundamentals, not narratives.
Time-series and cross-sectional trend — with the crash-risk controls the literature earned the hard way.
Short-horizon mean reversion — microstructure-aware entries and disciplined liquidity provision.
Earnings, index rebalances, corporate actions — event windows defined, tested, and traded systematically.
Implementation shortfall, schedule vs. opportunistic algos, venue analysis, and honest TCA.
Factor exposures, tail metrics, stress budgets — the risk stack that keeps a strategy alive to compound.
Mean-variance to robust and convex methods — turnover-aware, constraint-aware allocation.
Funding, basis, and perp microstructure — quant discipline applied to 24/7 rails.
Where the term structure meets the smile.
The hardest pricing problems in the building live here — path-dependent payoffs, callable optionality, and volatility surfaces that refuse to sit still. Nine domains, from instrument to model to surface.
Swaps, swaptions, caps and floors — pricing, curve construction, and risk done properly.
Bermudans, range accruals, TARNs — path-dependence priced with the model it deserves.
Call schedules, exercise boundaries, and the hedging behaviour they imply.
Issuance mechanics, embedded optionality, and lifecycle valuation control.
Market-observable forward rates, calibrated to caps and swaptions — the workhorse of exotics desks.
Term-structure dynamics driven from the whole forward curve — the theory the desk models descend from.
Low-dimensional Markovian rates models built for speed — callable books valued before the close, not after.
Smile dynamics and calibration — SABR-style parameterizations, skew risk, and the hedges that respect them.
Inter-rate and cross-asset correlation — estimation, stress, and the structures whose value lives in ρ.
Have a model that has to be right?
Scope it with the practice — we'll tell you honestly where the model risk sits, and what it takes to govern it.