Hedge Fund Tracker

Scrapes SEC 13F filings for a configurable universe of hedge funds, parses holdings, enriches with stock-level data, and surfaces cross-fund analytics through a Streamlit dashboard.

Structure mirrors clo_scraper: multi-module Python, SQLAlchemy ORM, analytics module, Click CLI. Retargeted from CLO trustee reports to institutional equity holdings.

Setup

pip install -r requirements.txt

cp .env.example .env            # then edit .env and put your name + email in SEC_USER_AGENT

python cli.py init-db
python cli.py seed
python cli.py refresh           # pulls 13Fs from EDGAR for every seeded fund
python cli.py map-tickers       # OpenFIGI: CUSIP -> ticker
python cli.py enrich-stocks     # yfinance: sector, market cap, last price
python cli.py rankings          # sanity check
streamlit run dashboard.py

Modules

hf_tracker/
├── config.py                       # paths, DB URL, SEC user agent
├── db.py                           # SQLAlchemy engine + session
├── models.py                       # Fund, Filing, Stock, Holding
├── scrapers/
│   ├── edgar.py                    # 13F pull + parse
│   └── stock_data.py               # OpenFIGI + yfinance enrichment
├── analytics/
│   └── fund_tracker.py             # FundTracker (analog to ManagerTracker)
├── data/
│   └── funds.yaml                  # seed universe
├── cli.py                          # Click commands
└── dashboard.py                    # Streamlit app

Data flow

SEC EDGAR ──13F XML──▶ scrapers/edgar.py ──▶ models.Holding (CUSIP only)
                                                    │
                       OpenFIGI ◀──────── scrapers/stock_data.py
                       yfinance ◀──────── scrapers/stock_data.py
                                                    │
                                                    ▼
                                         analytics/fund_tracker.py
                                                    │
                                                    ▼
                                          dashboard.py / cli.py

FundTracker analytics

• consensus_holdings(top_n) — most-held stocks across all funds, with fund count + total $ value
• new_positions(fund_id) / closed_positions(fund_id) — QoQ delta
• concentration(fund_id) — HHI + top-10 share
• overlap(fund_a, fund_b) — Jaccard + weighted portfolio overlap
• rankings() — funds sorted by 13F-reported value

Known caveats

• SEC rate limits: ≤10 req/s, and a User-Agent identifying you is required. Scraper sleeps 150 ms between calls.
• 13F value units: changed from "thousands of dollars" to "dollars" in mid-2022. Current code treats raw value as dollars. If you sync filings before Q3 2022, multiply by 1000 or add a date-conditional in parse_information_table.
• CIK accuracy: seed CIKs in funds.yaml are best-effort; verify on EDGAR before trusting.
• yfinance: unofficial, occasionally flaky. Swap for Financial Modeling Prep or Alpha Vantage if needed — schema is the same.
• Only long equity: 13F-HR captures long US-listed equity, ADRs, and certain options. It does not capture shorts, fixed income, derivatives notional, or international books. Fund-level "AUM" derived here is therefore reported-13F AUM, not true AUM.

Extending

• Add fund to universe: append to data/funds.yaml, run python cli.py seed then refresh --cik <CIK>.
• Add analytics: extend FundTracker in analytics/fund_tracker.py.
• Add stock data source: drop a new module in scrapers/ and call from the CLI.