Methodology

Version 1.0 — 28 May 2026.

Universal Asset Owners publishes journalism, research, scenario analysis, and data products. The methods we use differ across those products and are set out here so that a reader can judge the work on its inputs rather than its conclusions.

Daily Brief and intelligence reporting

The Daily Brief is a same-morning intelligence product covering moves by sovereign wealth funds, large pensions, endowments, foundations, insurance balance sheets, and the asset managers and policymakers in their orbit.

Inputs. Fund annual reports and quarterly updates; regulator filings (SEC, FCA, MAS, DFSA, FINMA, OSFI, ASIC and equivalents); central bank statements; multilateral institution publications (IMF, World Bank, OECD, BIS); press releases and public conference remarks; reporting by named outlets that we link.

Process. A morning slate is assembled, fact-checked against primary documents, edited, and reviewed against our Editorial Standards. Pieces that cite a primary source carry a link to it. Pieces that draw on a paywalled or registration-walled source say so.

AI assistance. The drafting and research workflow uses large language models for first-draft writing, summarisation, and source surfacing. Every published piece is edited and signed off by a human; the AI-assisted line on the post discloses the assistance.

The Probability Desk and scenario work

The Probability Desk is our afternoon scenario product. It assigns probabilities to discrete, named outcomes over stated time windows ("a closure of the Strait of Hormuz of more than 48 hours within twelve months"), shows the evidence that informs each probability, and re-scores when new evidence lands.

Inputs. Public reporting, official statements, treaty and legislative text, primary economic data, satellite and shipping data where available and licensed, expert work cited by name, and our own market and capital-flow tracking.

Process. Scenarios are stated as a named outcome over a named window. Probability ranges are assigned by a structured method (a base rate from the historical record where one exists, plus a small number of stated up- and down-weights with their sources). Where a probability is shown without a stated base rate, the post says so. A separate review pass checks each scenario for recency drift, dated framing, source quality, and overreach before publication. Flagship Probability Desk reports are subject to a stricter standard set out in the report.

What we do not do. We do not publish a probability without showing what informs it. We do not run a Monte Carlo simulation on a free post and present the output as the work; full-simulation reports are flagship products and disclose their model. We do not present probabilities as forecasts of the future; they are the platform's reading of the evidence at a stated moment.

UAO Research — long-form

Long-form research pieces are published in the UAO Research stream. Each piece states its research question, the time window of the analysis, the datasets used, and any limitations. Where a piece introduces a framework or index, the methodology for that framework is published with the piece.

Rankings, indices, and scoreboards

Where Universal Asset Owners publishes a ranking, index, or scoreboard, the methodology is published alongside it and includes:

The universe of funds, institutions, or instruments considered, and how it was defined.
The data window and the data sources, named.
The scoring rubric — what is measured, how it is weighted, and how ties are broken.
Any exclusions and the reason for them.
The version of the methodology and the changelog of substantive revisions.

A ranking that cannot meet this standard is not published.

Charts

Every chart published on the platform has:

A source line naming the primary dataset and the date of the snapshot.
The time window shown.
Where the chart presents a calculation we performed (a ratio, an average, an index), the formula or rule used.

Where a chart is generated programmatically, the source data and the rendering version are recorded.

Surveys

Where we publish survey results — an Asset Owner Sentiment Survey, an Allocator Outlook, or similar — the methodology section discloses:

The fielding window.
The sample size and the eligibility criteria for respondents.
The response rate where it can be calculated.
The fielding method (online, telephone, in-person).
Any incentives offered to respondents.
Whether any party other than UAO sponsored the survey, and what role they played.

AI in our workflow

The shorter summary is on the Editorial Standards page. The expanded version:

Drafting and research use general-purpose large language models. The model in use is recorded with the piece and disclosed on the post line.
Speech synthesis for the daily podcast uses commercial text-to-speech models. The voice used is a synthesised voice; the post discloses this.
Video generation uses a combination of commercial generative-video tools, stock licensed footage, and in-house compositing. Where AI-generated imagery is used to depict an event or place, the depiction is general rather than specific and the post discloses the use.
Charts are rendered programmatically in Python from the cited source data. We do not "AI-render" charts; the chart you see is plotted by software from the source numbers.
A human edits and signs off every piece before publication. No piece is published purely by automated workflow without that sign-off, even when the slate is built without human help.