> ## Documentation Index
> Fetch the complete documentation index at: https://docs.chicago.global/llms.txt
> Use this file to discover all available pages before exploring further.
# Macro Report Methodology
> How Macro Reports are produced: automated data ingestion, specialist analysis, synthesis, and editorial review in a weekly process
Macro Reports combine quantitative analysis, current news research, and a consistent editorial process. Each weekly report develops a market view from the latest available inputs and compares it with the previous report.
**The short version:** Automated data pipelines feed a three-stage research process: specialist analysts, then chief-investment-officer synthesis, then editorial review. Automated grading guides draft refinement, while completeness checks assess the report's coverage.
AI research agents produce each country report through specialist analysis, synthesis, and editorial review. Seven specialist analysts cover the nine report sections, supported by a dedicated draft-review agent. The global cross-asset report uses a broader analyst panel spanning asset classes and regions.
**Research team and AI roles.** The research team sets the cross-asset pillar scores. AI agents research, draft, and synthesize the report using those scores alongside quantitative data and current news.
Quantitative inputs come from dated market and economic series. News research adds context from public sources, assessed for relevance and recency. The report combines these inputs while preserving the distinction between measured data, research judgments, and forward-looking interpretation.
## The Data Foundation
Before a single sentence is written, the data layer does its work.
**Automated ingestion.** Pipelines pull economic releases, bond market data, currency series, equity factor and sector data, and sentiment and liquidity indicators on their natural frequencies. Market prices, yields, and currencies are daily series. Economic releases follow their publication calendars. Report inputs come from the ingestion pipeline rather than from manually maintained files.
**Computation before narration.** Factor measurements, spreads, z-scores, sentiment composites, and technical readings are calculated before drafting. Analysts receive those results as the quantitative basis for their sections.
**Data availability.** Analysts receive availability information alongside their inputs and are instructed to omit figures that lack supporting data.
## The Three-Stage Process
Every count below refers to the country report; the global cross-asset report uses a larger analyst set.
A panel of seven specialist analysts covers the report's domains: news, macro, rates, currency, sentiment, equity, and allocation. Each receives a domain-specific brief and the relevant data. Together they produce the nine sections of a country report. The counts differ because a specialist can own more than one section, and the equity specialist owns the factor and sector work.
Each specialist follows the same research process:
* **Researches actively**: searches current news, queries the historical time-series library, and retrieves relevant context from the archive of past reports
* **Anchors to data**: uses the supplied quantitative inputs as the numerical basis for its section
* **Undergoes automated review**: grading against accuracy, discipline, and style criteria guides draft refinement and selection
Most of the specialists run in parallel, which is how a full multi-section report can be rebuilt from scratch every week.
The synthesis layer combines the specialist sections into a single strategic narrative:
* **Resolves tensions**: when the rates view and the equity view pull in different directions, the synthesis layer is required to name the tension and come down on one side rather than print both and leave the reader to choose
* **Builds the thesis**: identifies the one or two forces actually driving the market this week and organizes every section around them
* **Checks against history**: consults the record of prior weeks' views, so shifts in stance are acknowledged and explained rather than silently made
A final editorial layer runs refinement passes against the house style guide:
* Institutional tone written for CIO-level readers
* Probabilistic language that reflects the strength of the evidence
* Internal shorthand and system jargon stripped out
* Checks against word-count, structure, and formatting standards
Editorial review aligns the sections in tone, structure, and terminology.
## Research Consistency
### The Policy Rate Is Fetched Once
Each market's policy rate is fetched once from the data library, date-stamped, and supplied to the sections that discuss monetary policy. This gives those sections a common reference.
When the policy rate is unavailable, the relevant sections receive that availability notice and are instructed to omit the figure.
### Automated Review and Refinement
A dedicated review agent evaluates analyst drafts for factual discipline, analytical quality, and style. Its review checks source support, interpretation, and adherence to the section's editorial criteria.
Review feedback guides draft refinement and selection before synthesis and editorial review. Completeness checks assess section coverage before the report proceeds.
### Persistent Memory and Continuity
After each report is published, its key views are extracted into a structured memory. Before the next report is written, the relevant history is retrieved and placed in front of every analyst.
Prior views provide context for changes in stance and the duration of developing trends. This helps readers compare the current analysis with earlier reports.
### Temporal Discipline
Each report carries a reference date. Analysts are instructed to distinguish observed events from scheduled developments after that date.
## Why This Architecture
The three-stage design gives research, synthesis, and editorial review distinct roles. Applied across covered markets, it provides:
* **A common research process across covered markets**
* **Weekly analysis using the latest available data**
* **Dedicated automated review of analyst drafts**
## Explore Further
The product overview and the value case.
What each of the nine sections delivers.
The ingestion pipelines and data categories.
The specialized terms used in the reports, defined in plain language.
***
**About this documentation.** These pages describe how Parallax Macro Reports are produced: the data that feeds them, the process that writes them, and the terms they use. They are product documentation, not research.
They contain no investment views. Positioning labels such as Overweight, Neutral and Underweight are defined here for reference only. Any views appear in the reports themselves, which carry their own disclosures.
Parallax Macro Reports are produced by **Chicago Global Capital Pte Ltd**, regulated by the **Monetary Authority of Singapore**.