David Maguire

A public research notebook documenting the mathematics, models, and experiments behind my work in quantitative finance, machine learning, and AI. The aim is simple: to demonstrate understanding rather than claim it — a record of research capability I am building as I develop a PhD proposal. The two core libraries and a full decision-systems tier — 74 entries, every number verified against real market data — are complete and build the methodological foundation of the proposal; a market-state specialist tier — the methods the updated proposal requires — is now complete alongside them.

The libraries

About

Background in banking analytics, my quant-finance interests, and the AI research direction I’m heading toward.

∑ Equation Library

Complete — 37 entries. The mathematical foundations of quant finance and machine learning, each taken from equation to plain English to Python, and tested on real Nasdaq data.

Machine Learning Library

37 core models, plus a complete specialist tier. From linear regression to transformers to a full reinforcement-learning and decision-systems tier — and now the market-state specialist methods the proposal requires: assumptions, mechanics, failure modes, and an honest market test for every one.

Capstone — the findings

What 74 entries add up to: one market, every model, the verdict on what is predictable and what isn’t, and the risk-governed decision system built on it — with a reading path through the whole site.

Quant Research Lab

Reproducible strategy experiments with explicit hypotheses, methods, and honestly reported results — including the ones that don’t work.

AI Trading Systems

Complete — 11 entries. How LLMs and AI agents could support market decision-making, grounded in real Anthropic API code — the agent loop, the News Agent, and the controls that make a system credible.

PhD Research

Full proposal published. Adaptive Market Intelligence — a probabilistic, multi-agent architecture for real-time market-state inference, strategy suitability and risk-governed decision-making, with an honest tracker of the evidence already built and the specialist methods still to add.

Code Repository

GitHub repositories and reproducible notebooks behind every experiment on this site.

How I decide what to publish

Nothing goes on this site until I can answer six questions about it without notes: what it means in plain English, why it matters in quant finance, what the equation or model is actually doing, what it assumes, where it can fail, and how I would implement it in Python. If I can’t answer those, the material stays private until I can. The goal is demonstrated understanding, not coverage.