296,000 filings since 2013. Top-ranked managers’ picks still trailed the S&P 500.
Every quarter, 8,000+ investment managers with $100m+ in 13(f) securities (mainly US-listed stocks) disclose their holdings to the SEC. This Python pipeline tracks where the capital moved, then backtests a 13F replication strategy: buying top-ranked managers’ disclosed positions the day after each filing.
- 13F FILERS · Q2 ’268,844
- NEW SPACEX POSITIONS · Q2 ’26$310bn
- FUNDS BUYING SPACEX1,649
- DATA AS OF19 Oct 2026
LARGEST NET BUYS AND SELLS · Q2 ’26
$BN · ALL FILERSExcludes spin-offs, mergers and listing moves. Share classes are summed. SpaceX listed on 12 Jun 2026, so its figure is mostly new positions (pre-IPO stakes that became reportable, and IPO allocations) valued at the quarter-end price.
NET BUYING BY MANAGER SIZE
2013–26The 10 largest managers were net buyers in 42 of 52 quarters; $100m–1bn managers were net sellers in 34. Net buying = change in shares × quarter-end price, as % of prior-quarter AUM, summed.
Skill test: in the 2015–20 ranking period (in-sample), top-ranked managers beat SPY by 11.4% a year. Out of sample, in 2021–25, they trailed it by 4.2% a year (t = −1.46, not significant).
Top-ranked = the top 10% (71 of 702) of concentrated managers ($100m–20bn, top 10 holdings ≥ 50% of the book), ranked on 2015–20 risk-adjusted return. Their disclosed positions were copied, equal weight, the day after each filing, net of costs.
HOW IT WORKS
PIPELINE- SEC EDGAR
- Parse 13F XML
- Filter managers
- Backtest vs SPY
- JSON snapshot
- This dashboard
WHAT’S NEXT
ROADMAP- Run H1: do pooled new buys and adds beat SPY over 63 days? (pre-registered)
- Forward-test the 20-manager shortlist frozen on Q2 ’26 books
- Split SpaceX into pre-IPO and IPO positions vs later buying
- Add delisted stocks to close the survivorship gap (81% price coverage today)