Selected client work and independent research. Details available upon request.
Methodology for live trading systems is proprietary and not publicly disclosed.
Manual signal generation across data prep, model runs, and line-shopping across 8 books was consuming 4+ hours per week. Not scalable, not sustainable, and not repeatable.
Built a quantitative model combining multiple independent algorithms—each with different logic, different data sources, and different blind spots—so model failures don't compound, they cancel. Validated manually over a live test window before rebuilding as automated infrastructure. No capital was put at scale until edge was statistically confirmed.
After proving the edge, rebuilt from scratch as production infrastructure:
What took 4 hours/week now runs in minutes.
Architecture built to scale to NCAA Football, NCAA Basketball, and NBA. Future product: direct signal access for qualified clients.
The underlying model methodology is proprietary and not disclosed. Results are from a live test window, not backtested simulations.
Cloud-based pipeline capturing live prediction market microstructure from Polymarket. 650K+ rows, 13K+ trades collected. Early analysis surfacing potential alpha in tail-size trades and asymmetric position dynamics.
LSTM-based deep learning model for t+1 Bitcoin price prediction using Fear and Greed Index as a feature, with hyperparameter tuning via Optuna. Full training, validation, and evaluation pipeline.
End-to-end neural network trading system with data preprocessing, model training, backtesting engine, and performance evaluation. Built for extensibility across multiple market regimes.
120-page quantitative research report on multi-season NFL ticket market dynamics. 1,000+ visualizations, probabilistic modeling across primary and secondary markets. Informed $1.1M+ in revenue-generating trading decisions.
Built implied volatility models across 15 commodity, currency, and index futures markets. Used for daily P&L and derivatives book risk management at an institutional hedge fund.
Developed financial models for loan performance, counterparty exposure, and collateral adequacy to support institutional crypto lending operations. Contributed to scaling the portfolio from $60M to $180M.