dawson. Get in touch

Market research, trading analytics and deal scoring

Fintech & quant

Research and analytics products on live market and property data: ranking trading strategies, reasoning about each day's market moves, and scoring real-estate deals. Plus payment and on-chain work built for clients.

Email me about this workSee all work

What ties this work together

  • Ledger firstPositions, cash and NAV are replayed from recorded orders, so every number can be traced back.
  • Every call is gradedMarket scenarios and trade ideas are checked against what actually happened.
  • Scores you can explainDeal and land scores show the signals behind them instead of a black-box number.

Top work

Ranked by how complete and substantial each one is, flagship first.

Ranks strategies by replaying ledgers

  1. Strategy orders
  2. Idempotent ingest
  3. Ledger replay
  4. NAV & positions
  5. Rank vs SPY

Research engine that tracks algorithmic trading strategies from backtest to paper trading and ranks them against SPY.

  • Authenticated, idempotent order-ingestion API for external strategy runners
  • Positions, cash and NAV derived by replaying the order ledger, never written directly
  • Splits and corporate actions applied from market data; intraday and end-of-day valuation
  • Python, FastAPI, PostgreSQL, Polygon.io; deployed on AWS
  • Python 3.11
  • FastAPI
  • Jinja2 + HTMX
  • Supabase
  • Polygon.io market data
  • Lightweight Charts
  • Live, sign-in
  • Built solo

Research to paper trading, gated

  1. Thesis
  2. Backtest
  3. Optimize
  4. Independent audit
  5. Shadow
  6. Paper

Quant strategy factory

Strategies move from a written thesis through backtest, logged optimization, an independent audit, shadow runs and paper trading; real money only by manual release.

21gate thresholds in one place

  • Each strategy needs a written thesis naming who is on the other side of the trade
  • Independent audit with adversarial review before shadow and paper trading
  • Real money only by manual release, never automatic
  • Python
  • pixi
  • Backtesting & walk-forward optimization
  • Next.js status board
  • Private code
  • Built solo
WhyNext screenshot

WhyNext

Pre-market causal analysis of global markets with three scenarios, probabilities and levels to watch; every call is graded after the close. Chinese-language.

268daily briefs published

  • Prices and headlines in, a schema-validated JSON report out
  • A deterministic template renders it, so the model never writes HTML
  • Every call is logged and graded in the evening review
  • Python
  • Pydantic
  • Jinja2
  • Polygon.io market data
  • Live
  • Built solo

More finance and analytics work

2 more, in ranked order.

  1. 4

    HouseFlipScore

    Deal sourcing for real-estate investors on live Los Angeles and Orange County parcel data, with explainable scoring and after-repair valuation.

    houseflipscore.com

    • Next.js 16
    • Prisma
    • PostgreSQL
    • Auth.js v5
    Visit
  2. 5

    Ranch Alpha

    Land valuation research: price per acre against county medians, water, access and catalysts, with flagged fatal issues.

    • Next.js 16
    • SQLite
    • Leaflet
    • LLM extraction
    Visit

Commissioned on-chain payment work

Built for clients, who own the code and are responsible for product rules and compliance.

Other areas

The rest of the work, by area.

Applied AI16 projectsAI education5 projectsFull-stack products11 projectsAll workEvery project in one list