Methodology
The Agentic Series · reviewed quarterly
The Inventing Quality Index is a rules-based, systematic equity index. It screens U.S.-listed companies for durable business quality and holds a concentrated set of the strongest names, rebalanced on a regular schedule.
Built by AI agents that read company filings. The rules that select and weight are fixed and published. What the agents do, and what they never do, is set out in full under The Role of AI Agents.
Beta — methodology in active fine-tuning. This index is in beta; the construction methodology described here and the constituent weights it produces are being actively refined and may change between rebalances.
Index Objective
The Index seeks durable U.S. businesses that compound value over time — companies that earn strong returns on the capital they employ, grow on their own merits, and consistently return cash to shareholders. Each candidate is evaluated against a consistent set of factors, and the highest-quality names are admitted.
What the Index Looks For
Companies are assessed across several dimensions of business quality and resilience. Among the factors considered:
- Returns on capital — how efficiently a business turns invested capital into profit, favoring companies that sustain high returns.
- Durable, organic growth — growth driven by the underlying business rather than one-off acquisitions or financial engineering.
- Balance-sheet strength — modest leverage and the financial flexibility to weather downturns.
- Cash generation & shareholder returns — steady free cash flow and a consistent record of returning cash through dividends and buybacks.
- Valuation discipline — a preference for quality available at a reasonable price rather than at any price.
- Risk & stability — lower fundamental and price-based risk, with an eye to structural threats to the business.
These factors are combined into a single quality assessment used to rank candidates. The emphasis is on businesses that score well across the board rather than excelling on one dimension alone.
The Role of AI Agents
Most of the Index’s factors are computed directly from reported financials and market data. Two forward-looking dimensions are read out of the filings by an AI agent panel:
- Growth analysis — assessing the durability and organic nature of each company’s growth outlook from its disclosures and reported results.
- Risk analysis — surfacing and characterizing the qualitative risk factors a company discloses, so that structural threats to the business are reflected in its score.
These agent assessments are converted into standardized factor scores and combined with the quantitative factors under the same rules-based framework. The agents inform the score; they do not override the rules that govern selection and weighting.
| What the agents read | SEC filings — 10-K, 10-Q, 8-K — and reported results. |
|---|---|
| What the agents produce | Two bounded scores per company: growth durability, qualitative risk. Nothing else. |
| What is deterministic | The eligibility screen, every financial factor, the composite formula, the ranking, the weights, the position cap, the retention band, the calendar. |
| What a human acknowledges | Methodology changes, with an effective date in the change log. |
| What the agents never do | Trade. Set or adjust a weight. Choose a threshold. Add or remove a constituent directly. See prices. Write the methodology. |
Construction
The Index holds a concentrated portfolio of the highest-ranked eligible companies. Holdings are weighted to reflect both company size and quality, with a ceiling on any single position so that no one name dominates the Index. Partnerships and other ineligible structures are excluded.
Rebalancing
The Index is reviewed and rebalanced on a regular calendar schedule. A retention preference for existing holdings limits unnecessary turnover from small changes in rank, while a periodic full refresh keeps the roster aligned with the current screen. Between rebalances, the Index monitors holdings for material deterioration and can reduce or remove a position when warranted.
Data & Governance
The Index draws on public company filings, market data, and AI language-model analysis to assess each company. The methodology is versioned, and material changes are recorded with an effective date. All inputs — including model outputs — are subject to error, revision, and delay.
This page provides a high-level overview of the Index methodology and is not a complete specification.