Private Market Fundraising Signals

Discover companies raising capital from SEC Form D filings. Backfill, filter, and track amendments, all from a single SQLite database.

Quick start

git clone https://github.com/AdityaKotari/pmft-tool
cd pmft-tool

# 1. Install dependencies
npm install
cd pipeline && uv sync && cd ..

# 2. Initialize the database
npm run db:init

# 3. Start the dev server
npm run dev

Open http://localhost:3000. The onboarding wizard walks you through setting your EDGAR identity and running your first backfill.

Prefer no local toolchain? docker compose up --build does the same thing. See HOWTORUN for Docker, scheduled pulls, and troubleshooting.

What it does

• Pulls SEC Form D filings (private capital raises) via edgartools
• Filters out pooled investment funds (the majority of Form D volume) using industry-group flags and entity-name heuristics
• Tracks amendments: detects when companies update their raise amounts, offering targets, or contacts in Form D/A filings
• Exports leads with primary contacts (officers/directors) ranked by role seniority

Architecture

fundraises-oss/
├── src/
│   ├── app/                    # Next.js 16 App Router
│   │   ├── api/                # REST API routes (better-sqlite3)
│   │   ├── dashboard/          # Lead browser with dynamic filters
│   │   └── onboarding/         # First-run wizard
│   ├── components/             # React components
│   └── lib/                    # DB, query builder, types, schema
├── pipeline/                   # Python ingestion pipeline
│   ├── fundscraper/            # EDGAR access, parsing, fund detection
│   ├── migrations/             # Alembic schema migrations
│   └── tests/                  # pytest suite (no network needed)
└── data/                       # SQLite DB (gitignored)

Read path: TypeScript → better-sqlite3 → SQLite. Instant queries.

Write path: TypeScript spawns Python subprocess → edgartools → SEC EDGAR. Backfills are idempotent (dates already in run_log are skipped).

API

Endpoint	Description
GET /api/health	DB status, filing count, date range
GET /api/schema	Available columns, filter presets, enum values
GET /api/leads	Paginated, filtered, sorted leads
GET /api/summary	Aggregate stats (total raised, median, amendments)
GET /api/amendments	Per-filing signals (new, amended, amount changed)
POST /api/backfill	Start a backfill for a date range
GET /api/backfill/progress	SSE stream of backfill progress
GET /api/backfill/coverage	Date coverage, gaps, history

Filter presets

Filters are named presets, not hardcoded tiers. The default preset is Technology (software, infrastructure, telecom). Switch to:

• Biotech & Pharma: biotech and pharmaceutical companies
• All Operating Companies: everything except funds/REITs/SPVs
• Everything: no filters, includes investment vehicles

The filter UI renders dynamically from the /api/schema columns, so adding new filterable fields requires no UI code changes.

Development

npm run dev        # Next.js dev server on :3000
npm run build      # Production build
npm run db:init    # Initialize/migrate the database
npm run db:backfill -- --from 2026-01-01 --to 2026-01-31  # CLI backfill

# Pipeline tests (no EDGAR access needed)
cd pipeline && uv run pytest

Stack

• Frontend: Next.js 16, React 19, TypeScript, Tailwind v4, TanStack Table
• Backend: Next.js API routes, better-sqlite3
• Pipeline: Python 3.12+, edgartools 5.35.0, SQLAlchemy 2.0, Alembic, DuckDB
• Database: SQLite (WAL mode, single file)

Data quality

• Contacts are officers/directors listed on Form D, not vetted fundraising contacts. Form D does not include email or phone.
• Amounts use the latest filing per company. Amendments are tracked separately. Pooled investment funds, REITs, and SPVs are excluded from operating-company totals.
• Industry groups come from the SEC's own taxonomy ("Other Technology," "Biotechnology," etc.) and are not enriched.

Known limits

• Form D gives a person's name + city/state, NO email; enrichment needed for contact info
• A filing proves a capital raise, not hiring or visa sponsorship
• edgartools is pinned to 5.35.0 (releases frequently, APIs break)

Contributing

npx tsc --noEmit                 # TypeScript
npm run build                    # Next.js build
cd pipeline && uv run ruff check && uv run mypy fundscraper/ && uv run pytest

Code layout: src/app/api/ (REST routes), src/app/dashboard/ (lead browser), src/components/ (data table, filters, backfill panel), src/lib/ (DB, query builder, schema), pipeline/fundscraper/ (ingest, adapter, filters, analysis, store).

Adding a filterable column: migrate the model (cd pipeline && uv run alembic revision --autogenerate -m "add X"), populate it in pipeline/fundscraper/adapter.py, add it to COLUMNS in src/lib/schema.ts, and, if filterable/sortable, to FILTERABLE_*_COLUMNS in src/components/filter-panel.tsx and ALLOWED_SORT_COLUMNS in src/lib/query-builder.ts.

Dependency policy: edgartools is pinned exactly (it ships often and breaks APIs); no framework dependencies in Python; flag cost before adding a new npm package.

License

MIT