Coding Projects
Investment tools, analytical platforms, and interactive research — designed, coded, and deployed from scratch using Python and TypeScript.
Personal portfolio research workspace tying eleven tools into one keyboard-driven app — morning-brief dashboard, screening, valuation heatmap, portfolio visualizer, construction, correlation, base rates, triage, research library, thesis tracking, and a long/short book. Runs on a 3,500-name global universe with the live book benchmarked against MSCI World Value.
Portfolio correlation analysis with Ward-linkage cluster reordering. Computes pairwise correlations across the entire holdings list, distils them into effective independent bets and a diversification score, and flags surprise diversifier pairs.
This website. 18+ research articles blending quantitative and qualitative thinking, with 15+ custom interactive D3 widgets including Monte Carlo simulators, probability heatmaps, network graphs, Bayesian updaters, and base rate quizzes. Built end-to-end as a static Next.js site.
Two sleeves on one book. Net style exposure across value, quality, momentum, and size percentiles shows the systematic bets the book is making, while regression-based macro exposures (market beta, commodities, rates, FX) reveal what it's implicitly long and short of.
Stochastic financial statement builder running 10,000 Monte Carlo simulations per company. Fits growth and margin distributions from 5–15 years of historical data, generates P5–P95 fair value ranges, and decomposes which drivers contribute most to valuation variance.
Risk scenario modelling for mapping geopolitical events to portfolio exposures. Probability-weighted impact analysis with tail distribution visualisation and Monte Carlo simulation across multiple risk scenarios.
Which sports produce the most dramatic, unpredictable matches? Interactive analytics platform ranking sports by a custom jeopardy index derived from lead changes, scoring entropy, and comeback frequency.
Screen a 3,500-name global equity universe across valuation, profitability, and quality. Composite cheapness and quality scoring with percentile gates, sector and region filters, and a per-name drawer linking straight into research, base rates, and correlation.
Sector × region heatmap of valuation metrics across a 2,400-name global universe. Switch between P/E, CAPE, FCF yield, and momentum views, then layer your portfolio over the market to spot exposure gaps and cheap pockets the book is missing.
Drill into any correlation cluster to see why names move together — common macro drivers, factor loadings, and within-cluster pair detail, with a narrated summary of what the cluster is really a bet on.
Multi-factor correlation analysis platform. Pairwise and rolling correlations between equity factors and individual stocks, with dendrogram clustering, time-series decomposition, and regime detection.
Build and modify portfolios with real-time impact analysis. Side-by-side current vs. proposed weights with delta tracking, live weighted-fundamentals impact (P/E, CAPE, ROE), sector deltas, and a plain-English change log.
One-page tearsheet of the live book — sector × region composition, weighted fundamentals vs. the MSCI World Value benchmark, concentration stats (top-10 weight, HHI), and thesis-breach triggers, with table and constellation views.
Institutional-grade emerging markets equity research suite. Eight integrated modules — stock screening, valuation heatmap, correlation matrix, portfolio construction, country risk dashboard, base rate analysis, and geopolitical risk monitoring — built end-to-end and deployed live for daily use by an EM value strategy.
24 emerging-market countries scored across 5 risk pillars — macro, political, market, external, structural — with customisable pillar weights. Sortable cards with composite scores and risk-band classification.
Historical peer comparison for the “outside view” — drops a company’s ROE, margins, and growth into its 50-name sector peer distribution over 10 years, with percentile ranking, trailing trends vs. the peer median, and an explicit survivorship-bias warning.
End-to-end investment research pipeline. Screens ~2,200 large-cap equities globally, triages them down to a shortlist using a 9-section quantitative brief, then runs full deep-dive analysis on each — value trap checklist (hidden liabilities, earnings distortions, balance sheet risk, etc.), Monte Carlo valuation, DCF, and a final publication-quality investment memo. The attached screenshot shows a real deep-dive output on Diageo.
Educational stock market simulator with virtual trading, real market data from EODHD, portfolio tracking, and a competitive leaderboard system.
3D city visualisation where buildings represent countries — height mapped to CAPE yield, width to market cap. Interactive WebGL fly-through experience with real-time valuation data.
Python analysis suite examining the accuracy of historical valuation models with the benefit of hindsight. Back-tests valuation assumptions against actual outcomes across thousands of stocks.
Frontend
React, Next.js, TypeScript, Tailwind CSS
Visualisation
D3.js, Recharts, Three.js
Backend
FastAPI, Flask, Python, Uvicorn
Data & Infrastructure
SQLite, PostgreSQL, Pandas, Vercel, Tauri
Every project is self-directed and built end-to-end — from data ingestion and backend API design to interactive frontend visualisation. The tools encode specific ways of thinking about investment problems; collectively they represent a process infrastructure that would be difficult to replicate.