Building ensemble models for daily S&P 500 return prediction in active Kaggle competition. Implementing feature engineering from market indicators including volatility regimes, cross-asset correlations and macroeconomic signals. Experimenting with XGBoost, neural networks and regime-switching models.
┌─[hull-tactical]──────────────┐ │ > Kaggle Competition │ │ > Ensemble Modelling │ │ > S&P 500 Direction │ │ [ENTER] View Details │ └──────────────────────────────┘
Languages
Python
Status
Status: In Progress
Timeline
Started: SEP 2025
Built production-ready pricing engine for interest rate swaps, FX options and structured products. Implemented curve construction (OIS/IRS), SABR/SVI volatility calibration and risk metrics (PV, DV01) under rate shock scenarios. Interactive 3D volatility surface visualization with scenario analysis.
┌─[derivx-fic-analytics]──────┐ │ > Curve Construction │ │ > Vol Surface Calibration │ │ > Risk Analytics │ │ [ENTER] View Details │ └────────────────────────────┘
Languages
Python
Timeline
Updated: OCT 2025
Modified open-source solver to optimize team selection using University of Bristol HPC cluster for 1000s of Monte Carlo simulations. Achieved consistent top 1% performance (3 years running, peak rank 20k/10M players) through hyperparameter tuning of horizon values, transfer dynamics and bench weighting strategies.
┌─[fpl-optimization]────────┐ │ > Monte Carlo Planning │ │ > HPC Optimisation │ │ > FPL Strategy Engine │ │ [ENTER] View Details │ └────────────────────────────┘
Languages
Python, C++
Timeline
Active: SEP 2021 – PRESENT
Statistical Time Series based approach extracting 29 latent factors from 127 macroeconomic indicators achieving 84.6% R² with recession early-warning signals. Implemented factor loadings capturing cross-sectional dependencies with 27-31% improvement over baseline ARIMA forecasts.
┌─[fred-md]──────────────────┐ │ > Dynamic Factor Models │ │ > Recession Signals │ │ > Macro Forecasting │ │ [ENTER] View Details │ └────────────────────────────┘
Languages
R
Timeline
Updated: MAR 2025
LightGBM ensemble achieving up to 8.7% edge relative to Betfair market odds through 68-feature engineering and Bayesian calibration methods. Implemented regularization techniques reducing overfit from 97.9% to 74.9% max confidence with calibration error <0.03.
┌─[horse-racing]────────────┐ │ > LightGBM Ensemble │ │ > Odds Calibration │ │ > Probability Modelling │ │ [ENTER] View Details │ └────────────────────────────┘
Languages
Python
Timeline
Updated: MAR 2025
Architected microservices platform processing 1000+ concurrent requests with sub-3 second latency. Implemented async FastAPI endpoints, Redis caching and PostgreSQL backend. Built comprehensive NLP pipeline using transformers and LangChain for document analysis.
┌─[career-rag-pipeline]──────┐ │ > FastAPI Microservices │ │ > LangChain NLP │ │ > Production Analytics │ │ [ENTER] View Details │ └────────────────────────────┘
Languages
Python
Timeline
Updated: AUG 2025
Project metadata is maintained manually in content/projects/projectsData.ts. Update languages, timelines, and tags when publishing changes.