Description

Responsibilities:

Own the system framework, responsible for the end-to-end signals system that turns raw signals into validated, production-ready predictions, and for the rigor of every stage in between.

• Make promote/hold/deprecate decisions on all candidate signals
• Interrogate promotion evidence and statistical rigor
• Validate holdout and placebo testing before production release
• Distinguish genuine inverse signals from artefacts

Backtesting and research design

• Own the walk-forward backtesting framework and challenge its design
• Design and test compound research hypotheses
• Own and evolve the signal promotion threshold policy
• Guard against overfitting, leakage, survivorship, and multiple-comparisons abuse

Prediction quality and calibration

• Assess live forecast accuracy and calibration
• Own the accuracy-vs-coverage trade-off
• Set and tune the abstention policy
• Benchmark performance against naive baselines

Monitoring and decay

• Monitor signal decay and govern deprecations
• Maintain integrity of the outcome-resolution pipeline

Client-facing methodology

• Write and defend client-facing methodology documentation
• Communicate statistical concepts to commercial/editorial stakeholders
• Serve as technical authority in pre-sales and onboarding

Requirements:

• 5+ years in a quantitative research, quantitative analyst, systematic strategy or financial data science role, with demonstrable ownership of signal research — not solely model implementation.
• Statistical rigour: Working command of hypothesis testing, multiple-comparisons correction (Benjamini–Hochberg or equivalent), rank correlation, ROC/AUC, calibration and proper scoring rules.
• Must be able to explain what a q-value guarantees that a p-value does not.
• Practical experience with walk-forward and purged cross-validation, look-ahead bias prevention, holdout design and regime-dependent performance.
• Financial markets literacy: Comfortable with equity index and ETF return data, forward-return construction, trading horizons, volatility regimes and macro context (VIX, yield curve, FRED series).
• Python: Fluent in Python 3.11+ with pandas, NumPy, scipy, statsmodels and scikit-learn — sufficient to reproduce, modify and extend the discovery and evaluation code, not merely to consume its output.
• SQL: Able to write non-trivial analytical SQL directly against PostgreSQL to interrogate signals, predictions and resolution coverage without waiting on an engineer.
• Intellectual honesty: A track record of killing their own results. This role exists to prevent AP publishing a false edge; scepticism must be a reflex, and must survive commercial pressure.
• Communication: Able to produce written methodology that stands up to a buy-side reader, and to explain it verbally to an executive audience.
• Experience with NLP-derived or alternative-data signals (news, sentiment, filings, satellite, transactional) and their particular failure modes.
• Familiarity with gradient-boosted ensembles (LightGBM, CatBoost, XGBoost), stacking with out-of-fold predictions, and isotonic or Platt calibration.
• Exposure to conformal prediction, abstention/selective-prediction frameworks, or cost-sensitive decision thresholds.
• Prior work in a commercial data-product context where methodology was client-visible and contractually relevant.
• Working knowledge of GCP (BigQuery, Vertex AI, Cloud Run) or an equivalent cloud analytics environment.
• Graduate degree in statistics, financial engineering, econometrics, physics, mathematics or a comparable quantitative discipline

Key Skills

Education

Any Graduate