James Lewis

MSc Data Science & Statistics @Exeter. Building real-time trading bots, crypto analytics tools, and predictive ML pipelines.

Projects & Research

A collection of applied and technical work, from algorithmic trading systems and on-chain analytics to academic data science research.

🔁 Trading Systems & Blockchain Analytics

Live trading systems and alpha discovery research in Solana markets, combining real-time blockchain data with algorithmic strategies.

Dex-Paid Alpha Signal Bot (Closed Source, Live Trading)

Status Tools Production-grade Solana trading system built in TypeScript, designed for ultra-low latency detection of Dex-Paid tokens and smart alpha-based entries.

  • Actively running live with consistent profitability on low-cap Solana tokens
  • Integrates SolanaTracker, DexScreener, and Helius APIs for deep market microstructure insights
  • Features wallet clustering, liquidity heuristics, and rug-pull detection
  • Built with concurrency optimisations, request throttling, and structured logging
  • Sends real-time trade alerts to Discord and manages positions with configurable exit strategies

(Closed-source, private system actively trading.)

Quantile Regression Forests for Solana Returns (Research Paper)

Tools Peer-reviewed MSc dissertation on interval forecasting of Solana token returns using Quantile Regression Forests (QRF).

  • Developed a robust forecasting framework for 72-hour return intervals
  • Benchmarked against linear quantile regression, and LightGBM baselines
  • Tail risk calibration and coverage efficiency analysis
  • Applied interpretable ML to real-world Solana token data
  • Published on ResearchGate

📊 Academic Data Science Projects

Coursework and research projects applying advanced statistical and ML methods across domains such as ecology, epidemiology, and environmental modelling.

Grade Model Tools Classified conflict-driven fire activity using mixed-effects models with temporal-spatial features and animated visualisations.

Deepwater Horizon: Phytoplankton Disruption

Grade Focus Analysed Gulf phytoplankton anomalies post oil spill via STL decomposition + satellite data.

TB Risk in Brazil (GAMs)

Grade Model Spatial-temporal risk modelling of tuberculosis using Tweedie GAMs across 557 regions.

Grade Compared GLMs for AIDS progression (Poisson, Gaussian, Tweedie) with AIC model selection.

Space & Time Modelling of Global Temperatures

Grade Combined ARIMA, kriging, and dynamic linear models to study SST, AMOC, and California temps.

Reaction Time Modelling & ML Classification

Grade Bayesian + ML workflow: MCMC diagnostics + QDA, RF, SVM classifiers.

Multi-Part Coursework (PCA, Clustering, Regression)

Grade Projects on sea level rise, power demand clustering, and gene prediction via PCA & regression.

👉 For full repos and updates, visit my GitHub or get in touch.