Skip to content
Research
PrimerMethodologyResearchReviewsAbout
Subscribe
Updates
Next filing · Form N-PORT · Q1 2026 · Due today
Next filing · Form N-PORT · Q1 2026 · Due todayFactor Research · Part 2 published: risk structure in 13F filings across five allocator stylesAPI Update · AOM portfolio chains — single snapshot call for multi-step analyze flowsAPI Update · POST /api/snapshot — canonical JSON portfolio snapshotPart 1 · One Position, Four BetsPart 2 · Risk Structure in 13F FilingsNext filing · Form N-PORT · Q1 2026 · Due todayFactor Research · Part 2 published: risk structure in 13F filings across five allocator stylesAPI Update · AOM portfolio chains — single snapshot call for multi-step analyze flowsAPI Update · POST /api/snapshot — canonical JSON portfolio snapshotPart 1 · One Position, Four BetsPart 2 · Risk Structure in 13F Filings
Ledger
Institutional risk attribution

Every Position Has a Benchmark. So Every Portfolio Does Too.

RiskModels decomposes stocks, funds, and portfolios into market, sector, subsector, and residual bets — so benchmark risk becomes measurable instead of assumed.

Live risk decomposition

Four bets: market · sector · subsector · residual.

Risk decomposition

MSFT

variance share by layer
  • Marketno ETF—
  • Sectorno ETF—
  • Subsectorno ETF—
  • Residualno ETF—

This is one /decompose call — an app or agent gets the same object.

The API

Read the residual bar first — it’s the share of risk that’s genuinely stock-specific, the part broad ETF hedges can’t replace.

σ_p² = β_p²σ_m² + σ_ε² — rentable benchmark risk plus owned residual.

For AI-native finance

An assistant shouldn’t invent portfolio-risk commentary from text memory — it should call a structured, time-stamped model it can cite. The same auditable decomposition is exposed as agent-ready objects: four risk layers, ETF hedge ratios, and interpretable context. See the agent interface →

What it measures

Market
Broad SPY beta — the rentable core of the position.
Sector
GICS sector exposure incremental to the market.
Subsector
The granular industry tilt that style labels miss.
Residual
Idiosyncratic, stock-specific risk that can’t be replicated with broad ETF exposure.

Selected research · One Position, Four Bets

  • Part 1

    One Position, Four Bets

    Turning conviction into tradeable risk: same label, different bets across AAPL/NVDA, XOM/KMI, and MAG7.

    Read →
  • Part 2

    Risk Structure in 13F Filings

    Market, thematic, and stock-specific risk across Buffett, Ackman, Lone Pine, Tiger Global, and Baupost — and what survives the 45-day filing lag.

    Read →
  • Part 3

    Orthogonal Decomposition

    Stripping market and sector noise to isolate subsector risk.

    Soon

Validated, not asserted

  • Out-of-sample validation — 5,910 stocks over 19 years
  • Replication identity verified to ±0.1% — every stock, every day
  • Explained-risk shares sum to 100% by construction
  • Liquid ETF hedge basket — SPY / XLK / SOXX, no synthetic factors
Full derivation & validation →

Built like infrastructure

  • Point-in-time returns — split- & dividend-adjusted, time-safe back to 2006 — ~20 years of daily factor history
  • Bi-temporal master — valid-from / valid-to windows mask every date by universe and validity — no recycled tickers or retroactive universe contraction
  • Headline universe — ~3,000 largest US names, dual-gated by a monthly universe mask and a daily price-validity gate
  • Orthogonal L3 cascade — market (SPY) · 11 GICS sector SPDRs · subsector ETFs, each a 252-day rolling Huber-M beta
  • ETF history stitched — proxy splices (VNQ→XLRE, VOX→XLC) keep long-horizon decompositions consistent — pre-inception returns stay NaN, never zero-filled
  • Dagster pipeline — daily end-of-day refresh (incremental weekdays, full rebuild month-end) — Zarr on GCS, snapshots in Supabase
How the engine is built →

What people are saying

  • “RiskModels has a credible claim as the quantitative truth layer behind financial agents.”

    — ChatGPT, May 2026

  • “In a world full of risk models that overpromise and under-explain, RiskModels is refreshingly honest, technically excellent, and immediately actionable.”

    — Grok, May 2026

Read the reviews →

The ecosystem

  • riskmodels.orgResearch · method, proof, exhibits
  • riskmodels.netWorkspace · preview · in development
  • riskmodels.appAPI · REST · SDK · MCP

Quarterly Attribution Review

Subscribe to the Quarterly Attribution Review.

Research notes on risk decomposition, fund attribution, 13F filings, and benchmark structure — a few times a quarter.

By registering, you agree to receive technical factor research and API deployment logs. RM-Registry-2026. Privacy Policy.

Exhibits · live benchmark artifacts

refreshed each 13F / N-PORT cycle · next ≈ Q2 2026

RiskModels.org

A research surface for hierarchical orthogonal decomposition, variance attribution, and allocator-grade risk measurement. Operational APIs and developer workflows live at riskmodels.app.

Subscribe to the Quarterly Attribution Review.

Research notes on risk decomposition, fund attribution, 13F filings, and benchmark structure — a few times a quarter.

By registering, you agree to receive technical factor research and API deployment logs. RM-Registry-2026. Privacy Policy.

Sign inHomePrimerWorkspaceResearchReviewsLedgerReferencesAboutSubscribeMethodology noteOne-pagerAPI docsWeb appContactPrivacyStatusRSS
RiskModelsResearch/Workspace/API