Kasher13/Institutional-Holdings-Dashboard

📊 Institution Holdings Dashboard

SEC EDGAR 13F filings — cleaned, structured, and ready to use. 42 top hedge funds · 10+ years of history · Weekly auto-updates · Zero auth required

---

📌 Overview

This dataset contains cleaned, structured institutional holdings data parsed directly from SEC EDGAR 13F-HR filings. It powers a public intelligence platform tracking what the world's top hedge funds are buying and selling — quarter by quarter.

Everything in this dataset is freely derived from public SEC EDGAR filings.

Stat	Value
🏦 Managers tracked	42 top institutional investors
📅 History depth	10+ years (40+ quarters per manager)
🔄 Update frequency	Weekly (every Sunday via GitHub Actions)
💾 Full database	~600 MB SQLite (cache.db)
📁 Pre-built API	Static JSON files — no auth, no rate limits
🌐 Live demo	Open Dashboard →

---

🏦 Tracked Institutions

Includes top hedge funds and asset managers such as:

Berkshire Hathaway · Bridgewater Associates · Soros Fund Management · Renaissance Technologies · Pershing Square · Druckenmiller Capital · Tiger Global · Citadel · Viking Global · Coatue Management · D1 Capital Partners · Third Point · Greenlight Capital · Appaloosa Management · Baupost Group · Lone Pine Capital · Two Sigma · Point72 · Elliott Management · Farallon Capital · Jana Partners · Maverick Capital · Eminence Capital · Glenview Capital · Highfields Capital · Pzena Investment Management · Southeastern Asset Management · ValueAct Capital · Corvex Management · Starboard Value · Sachem Head Capital · Trian Fund Management · Luxor Capital · Omega Advisors · Gotham Asset Management · Icahn Associates · Armistice Capital · OrbiMed Advisors · Redmile Group · Rock Springs Capital · Venrock Healthcare Capital · Whale Rock Capital

---

📂 Dataset Contents

Kasher13/Institutional-Holdings-Dashboard/
│
├── cache.db                              ← Full SQLite database (~600 MB)
│
└── api/                                  ← Pre-built static JSON endpoints
    ├── meta.json                         ← Dataset metadata & stats
    ├── managers/
    │   ├── popular.json                  ← List of all 42 tracked managers
    │   ├── {cik}.json                    ← Manager profile + filing history
    │   ├── {cik}/history.json            ← Top holdings across all quarters
    │   └── {cik}/holdings/{year}/{q}.json ← Holdings for a specific quarter
    ├── stocks/
    │   ├── popular.json                  ← Popular tracked stocks
    │   └── {cusip}/holders/latest/latest.json ← Top institutional holders
    └── search/
        ├── managers.json                 ← Full manager search index
        └── stocks.json                  ← Full stock/CUSIP search index

CIK = SEC Central Index Key (10-digit, zero-padded, e.g. 0001067983 for Berkshire Hathaway) CUSIP = 9-character stock identifier (e.g. 037833100 for Apple)

---

🔌 API Usage (No Auth Required)

All JSON files are served via Hugging Face's global CDN — no API key, no rate limiting, no server needed.

Base URL

https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api

Python

import requests

BASE = "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api"

# ── List all tracked managers ─────────────────────────────────────────────
managers = requests.get(f"{BASE}/managers/popular.json").json()
for m in managers["data"]:
    print(f"{m['name']}  (CIK: {m['cik']})")

# ── Berkshire Hathaway's latest holdings ─────────────────────────────────
q = requests.get(f"{BASE}/managers/0001067983/holdings/2024/4.json").json()
for h in q["data"]["holdings"][:10]:
    print(f"  {h['issuer_name']:40s}  ${h['value']:>15,}  ({h['pct_portfolio']:.1f}%)")

# ── Who holds Apple? (CUSIP 037833100) ────────────────────────────────────
apple = requests.get(f"{BASE}/stocks/037833100/holders/latest/latest.json").json()
for holder in apple["data"]["holders"][:5]:
    print(f"  {holder['manager_name']:40s}  {holder['shares']:>15,} shares")

# ── Manager portfolio history (top holdings over time) ────────────────────
history = requests.get(f"{BASE}/managers/0001067983/history.json").json()
for quarter in history["data"]["periods"][:5]:
    print(quarter)

JavaScript / TypeScript

const BASE =
  "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api";

// List all tracked managers
const { data: managers } = await fetch(`${BASE}/managers/popular.json`).then(r => r.json());
managers.forEach(m => console.log(`${m.name} — CIK: ${m.cik}`));

// Citadel's portfolio history
const { data: history } = await fetch(`${BASE}/managers/0001423053/history.json`).then(r => r.json());
console.log(history.periods);

// Search index for stocks
const { data: stocks } = await fetch(`${BASE}/search/stocks.json`).then(r => r.json());
const aapl = stocks.find(s => s.cusip === "037833100");

curl

# List all managers
curl -s "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api/managers/popular.json" | python -m json.tool

# Bridgewater's profile and filings
curl -s "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api/managers/0001350694.json" | python -m json.tool

# Dataset metadata and stats
curl -s "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/api/meta.json" | python -m json.tool

---

🗂️ Data Schema

All API endpoints return a consistent wrapper:

{
  "data": { ... },
  "cached": true,
  "static": true
}

managers/popular.json — Manager list

{
  "data": [
    {
      "cik": "0001067983",
      "name": "BERKSHIRE HATHAWAY INC",
      "display_name": "Berkshire Hathaway",
      "filing_count": 43,
      "latest_period": "2024Q4"
    }
  ]
}

managers/{cik}/holdings/{year}/{quarter}.json — Quarterly holdings

{
  "data": {
    "cik": "0001067983",
    "period": "2024Q4",
    "filed_date": "2025-02-14",
    "total_value": 267000000000,
    "holdings": [
      {
        "cusip": "037833100",
        "issuer_name": "APPLE INC",
        "shares": 300000000,
        "value": 70000000000,
        "pct_portfolio": 26.2,
        "put_call": null,
        "investment_discretion": "SOLE"
      }
    ]
  }
}

Holdings fields:

Field	Type	Description
cusip	string	9-char stock identifier
issuer_name	string	Company name as reported to SEC
shares	integer	Number of shares held (×1000 per SEC convention)
value	integer	Market value in USD (×1000 per SEC convention)
pct_portfolio	float	Percentage of total portfolio value
put_call	string|null	"Put", "Call", or null for equity
investment_discretion	string	"SOLE", "SHARED", or "OTHER"

stocks/{cusip}/holders/latest/latest.json — Institutional holders

{
  "data": {
    "cusip": "037833100",
    "issuer_name": "APPLE INC",
    "period": "2024Q4",
    "holders": [
      {
        "cik": "0001067983",
        "manager_name": "BERKSHIRE HATHAWAY INC",
        "shares": 300000000,
        "value": 70000000000,
        "pct_portfolio": 26.2
      }
    ]
  }
}

meta.json — Dataset statistics

{
  "data": {
    "managers_count": 42,
    "filings_count": 1820,
    "holdings_count": 485000,
    "cusips_tracked": 12000,
    "earliest_period": "2013Q4",
    "latest_period": "2024Q4",
    "last_updated": "2025-03-23T00:00:00Z"
  }
}

---

💾 Full Database Download

For bulk analysis, download the complete SQLite database (~600 MB):

https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/cache.db

SQLite schema:

-- Tracked institutional managers
CREATE TABLE managers (
    cik TEXT PRIMARY KEY,
    name TEXT,
    display_name TEXT
);

-- Individual 13F filing periods
CREATE TABLE filings (
    id INTEGER PRIMARY KEY,
    cik TEXT,
    period_of_report TEXT,   -- e.g. "2024-12-31"
    filed_date TEXT,
    total_value INTEGER
);

-- Individual stock positions per filing
CREATE TABLE holdings (
    id INTEGER PRIMARY KEY,
    filing_id INTEGER,
    cusip TEXT,
    issuer_name TEXT,
    shares INTEGER,
    value INTEGER,            -- ×1000 USD (SEC convention)
    put_call TEXT,
    investment_discretion TEXT
);

Example SQLite query:

import sqlite3, urllib.request

# Download once
urllib.request.urlretrieve(
    "https://huggingface.co/datasets/Kasher13/Institutional-Holdings-Dashboard/resolve/main/cache.db",
    "cache.db"
)

conn = sqlite3.connect("cache.db")

# Top 10 most widely held stocks (latest quarter)
rows = conn.execute("""
    SELECT h.issuer_name, COUNT(DISTINCT f.cik) as holders, SUM(h.value) as total_value
    FROM holdings h
    JOIN filings f ON h.filing_id = f.id
    WHERE f.period_of_report = (SELECT MAX(period_of_report) FROM filings)
    GROUP BY h.cusip
    ORDER BY holders DESC
    LIMIT 10
""").fetchall()

for name, holders, value in rows:
    print(f"{name:40s}  held by {holders:2d} managers  ${value/1e6:,.0f}M")

---

🔄 Update Pipeline

Data is refreshed automatically every week:

Every Sunday 00:00 UTC
  └─ GitHub Actions: crawl SEC EDGAR → parse 13F XML → store in SQLite
  └─ Generate static JSON files for all managers/stocks/search
  └─ Upload cache.db + JSON to this Hugging Face dataset
  └─ Deploy updated dashboard to GitHub Pages

---

🌐 Live Dashboard

Explore the data interactively — no setup required:

→ Open Institutional Holdings Dashboard

Features:

• Portfolio comparison between any two quarters (NEW / EXITED / INCREASED / DECREASED)
• Historical stacked area charts (Top 30 holdings evolution)
• Stock holder intelligence — who owns what and how positions changed
• Full-text search by manager name, CIK, stock name, or CUSIP
• Portfolio allocation pie charts

---

📡 Postman Collection

Download the full API schema for Postman or Insomnia: Vantage_API_Postman_Collection.json

---

☕ Support

If this dataset saves you time, consider buying me a coffee — it keeps the weekly crawls running!

---

⚖️ License & Disclaimer

• Code: MIT License
• Data: Sourced from SEC EDGAR — free and public domain
• Disclaimer: For informational and educational purposes only. This dataset reflects what institutions have reported to the SEC — not real-time positions. Nothing here constitutes financial or investment advice.

---

🙏 Credits

Built by KitTran1307 · Developed with Gemini · Data from SEC EDGAR