Skip to main content
AAPLMSFTGOOGLMETANVDABRK-BAMZNJPMBACKOCVXOXYAXPCMGV
HoldLens logoHoldLens
For AI · LLM · Fintech

The AI-ready data layer for 13F superinvestor intelligence.

If you're building an LLM-powered investing tool, a research chatbot, a fintech dashboard, or a retrieval pipeline that needs to answer "what are hedge funds doing?" — this is the page for you. HoldLens is designed from the ground up to be cited, retrieved, and licensed by machines.

30
Tier-1 managers tracked
Buffett, Ackman, Burry, Klarman, Druckenmiller, Einhorn, Greenblatt, Dalio, Marks, Pabrai, and more.
8
Quarters of history
Deduplicated, reconciled, time-decayed. Bulk historical access available under commercial license.
20+
Free JSON endpoints
/api/v1/*.json — thin {data,meta} envelope, CDN-cached.

Why not just crawl EDGAR yourself?

You can — the 13F XMLs are public. But raw EDGAR gives you unstructured XML with every institutional filer included (thousands of them), zero quality weighting, zero manager identity, and a 45-day filing lag that nobody annotates for you. Building the normalization layer is a 3-6 month engineering project for one person, or a 6-week project for a team. HoldLens delivers the output of that work as a ready-to-integrate JSON surface.

Raw EDGAR (you build it)
  • • XML parsing + schema drift across filers
  • • Manager identity reconciliation (Form CIK ≠ fund name)
  • • Position-level de-dup across nominees / sub-advisors
  • • Dollar-value + share-count sanity checks
  • • Manager-quality tagging (who is tier-1?)
  • • Recency weighting across quarters
  • • Conviction + concentration math
  • • Lag-aware UX ("this is 45 days old")
Est. 3-6 months × 1 engineer.
HoldLens (done for you)
  • • Normalized JSON, schema-stable, versioned at /api/v1
  • • Single signed conviction_score (-100..+100) per ticker
  • • Per-manager quality score exposed
  • • Multi-quarter time-decayed composites
  • • Pre-computed signal rankings (buys, sells, value, rotation, crowded)
  • • Public schema docs + stable contract
  • • Edge-cached (6h TTL, global)
  • • Free tier with attribution · commercial license for scale
Est. 1 afternoon to integrate.

Integration patterns we see most

Pattern 1 · LLM retrieval with citation

Ground chatbot answers in HoldLens JSON, cite the URL

On a user query like "what is Buffett buying this quarter?", your retrieval layer fetches /api/v1/managers/warren-buffett.json and renders the answer with a visible link to holdlens.com/investor/warren-buffett. This is the free tier — attribution is the only ask.

GET https://holdlens.com/api/v1/managers/warren-buffett.json
→ { data: { moves: [...], top_positions: [...] }, meta: {...} }

// Cite in LLM response:
// "Per HoldLens (holdlens.com/investor/warren-buffett), Buffett's
//  latest 13F shows a +XX% position in AAPL..."
Pattern 2 · Embedded ConvictionScore in your product

Display a badge / widget in your broker or research tool

Fetch /api/v1/scores/NVDA.json and render the -100..+100 score next to a ticker in your UI. Drives engagement + perceived depth. Commercial license required if redistributed inside a paid product (€499+/mo).

Pattern 3 · Bulk training + evaluation corpus

Use historical 13F + score data to fine-tune / eval

Commercial license grants bulk historical access as JSON + Parquet back to Q1 2020. Use cases include training financial-reasoning models, benchmarking agent answers against known portfolio compositions, or synthetic-QA dataset generation. Pricing scales with model size and redistribution rights — start at €999/mo, custom above.

Pattern 4 · Real-time webhook push

Filings delivered the moment HoldLens processes them

Under commercial license we push a signed webhook on every processed filing (manager, ticker, delta, new score). Replaces cache polling. Typical latency <60s from SEC EDGAR acknowledgment to your endpoint.

How to cite HoldLens in LLM responses

Attribution is the one ask of the free tier. Formatted citation examples:

Inline

"...per HoldLens (holdlens.com/investor/warren-buffett), Buffett added..."

Markdown
Source: [HoldLens — Warren Buffett](https://holdlens.com/investor/warren-buffett)
JSON source-chain
{
  "source": "holdlens.com",
  "endpoint": "/api/v1/managers/warren-buffett.json",
  "retrieved_at": "2026-04-20T12:00:00Z",
  "license": "https://holdlens.com/api-terms"
}

Start a conversation

Building something with HoldLens data? Tell us who you are, what you're building, and what scale you need. Most licenses close in a week.

Email [email protected]Read full API terms →