Dhuel FisherProducts / interfaces / systems
Menu
All projects

Personal project / Independent project / March 2024–present

Stock App

A personal finance workspace that turns raw brokerage and market data into a small number of clear, explained decisions, with every side effect kept under my control.

Active personal use; 2,000+ commits

2,000+
commits since March 2024
600+
Python test files
800+
TypeScript test files
56
API routes behind about 20 pages

The problem

Checking dividends, cost basis, spending, and long-term goals by hand across accounts is slow and easy to get wrong. Most tools show numbers without explaining why a suggestion exists or how trustworthy the data behind it is.

My role

Sole designer and developer.

Inside the product

Stock App home screen with a sidebar of money and plays pages, a Do now card, first actions, and a tax-year summary. All numbers are masked.
The owner workspace: Home leads with what needs attention today, then gives each page one clear next step.
Spending month-closeout cards and a net-versus-reserve-target line chart, with values masked.
Spending closeout: the month reads in plain language first, then the trend shows net cash climbing past the reserve target.
Two dividend bar charts, year by year and current year by month, with explanatory copy and pace cards. Values are masked.
Dividend tracking from the Python core: year-over-year growth, with copy that says exactly what each chart includes.
  1. Sources
    • Brokerage exports
    • Market data
    • Spending records
  2. Python core
    • Dividend checks
    • FIFO cost basis
    • SQLite
  3. Workspace
    • Next.js pages
    • API routes
    • Source-health checks
  4. Owner
    • Explained decisions
    • Review and confirm
    • Read-only by default
From source data to a decision. Nothing reaches a recommendation without a source-health check, and nothing reaches the broker without my review.

What I did

  1. A Python core for the accounting

    Python modules handle dividend-transaction checks, FIFO cost-basis accounting, market data, reports, and exports, backed by SQLite and a large pytest suite.

  2. A Next.js owner workspace

    A Next.js, React, TypeScript, and Tailwind interface with about 20 pages (operations status, spending, goal tracking, insights, a stock picker, crypto, assets, and tasks) served by 56 API routes over local data adapters.

  3. Recommendations that explain themselves

    A decision contract requires every suggestion to separate a real action from an optional idea, say why it exists and what would change it, and show whether its sources are fresh.

  4. Safety before automation

    The app is read-only by default. Orders are possible only in one dedicated account, through an official brokerage integration, after I review and confirm each one. Automatic lanes stay off until I turn them on.

  5. Audited in cycles

    Repeated full-app audit cycles check routes, UI, latency, automation, source trust, and recommendation quality, and each finding is recorded so it is not repeated.

Stack

  • Python
  • SQLite
  • pytest
  • TypeScript
  • Next.js
  • React
  • Tailwind CSS

The repository is private and holds real account data. Screenshots are taken from the running app with every number and account name masked.