Quantitative Researcher / ML Engineer – Crypto Market Microstructure Signal Validation (Python)

We are seeking an exceptional Quantitative Researcher, Machine Learning Engineer, Applied Mathematician, Statistical Modeler, PhD student, postdoc, or experienced quantitative researcher for a short, well-scoped paid research engagement focused on crypto market microstructure. This is a paid evaluation project intended to identify outstanding quantitative researchers for an ongoing research pipeline involving market microstructure, machine learning, AI-assisted trading research, and statistical signal discovery. The initial task is deliberately focused and should take an experienced researcher only a few hours. PROJECT You'll analyze high-frequency Level 2 order book snapshots and trade tick data (BTC/ETH from a major exchange) to rigorously evaluate a specific hypothesis regarding short-horizon price movement predictability following changes in order book imbalance. This is NOT a dashboard, visualization, ETL, or data-cleaning project. We're looking for someone who can think critically, formulate rigorous statistical tests, distinguish genuine predictive signal from noise, and clearly communicate defensible conclusions. We value rigorous reasoning over complicated models. A simple model that survives proper validation is far more valuable than a sophisticated model that overfits. RESPONSIBILITIES • Ingest and preprocess provided L2 order book and trade data (Parquet/CSV) • Formulate a statistically rigorous hypothesis around order book imbalance and short-horizon returns • Design appropriate validation methodology • Test stationarity, autocorrelation, statistical significance, and multiple-testing effects • Build a small predictive prototype (Logistic Regression, Gradient Boosting, Bayesian approach, or another justified methodology) • Evaluate calibration, overfitting risk, generalization, and out-of-sample stability • Clearly explain assumptions, methodology, limitations, and conclusions DELIVERABLES • Well-documented Jupyter Notebook • Clean, reproducible Python code • Executive summary (1–2 pages) • Statistical justification of conclusions • Recommendation on whether the observed signal appears genuine and worthy of further research The ultimate question is: "Does this appear to be a statistically defensible predictive signal—or simply noise?" REQUIRED SKILLS • Strong probability and statistics • Time-series analysis • Sequential modeling • Hypothesis testing • Multiple-comparison awareness • Machine learning • Regression and classification • Model calibration • Walk-forward validation • Python (pandas, numpy, scikit-learn, statsmodels, PyMC or similar) • Comfortable working with noisy, high-frequency market data NICE TO HAVE • Crypto market microstructure • FX or equities market microstructure • Order book modeling • Queue dynamics • High-frequency trading research • Bayesian modeling • GitHub portfolio • Kaggle • Published research TO APPLY Please include: 1. A brief overview of your quantitative research and machine learning experience. 2. Links to GitHub, Kaggle, research papers, notebooks, or other relevant work. 3. In 2–3 sentences, explain how you distinguish genuine predictive signal from noise in short-horizon financial data. 4. Your estimated turnaround time. 5. To confirm you've read the entire posting, begin your proposal with the word "Centurion." BUDGET Fixed Price: $200 USD This is intentionally structured as a paid evaluation project. Outstanding work is expected to lead to additional paid research projects and longer-term collaboration at higher budgets. We're looking for someone who enjoys solving difficult quantitative problems and values rigorous statistical reasoning over flashy models.