How We Work
Our Methodology
Rigorous, transparent, and accountable. Every Crestmore research deliverable is produced through a documented four-stage process — and we publish our forecast accuracy on an annual basis.
The research industry has a transparency problem. Too often, institutions publish forecasts without explaining their assumptions, use models they never document, and quietly revise history when calls go wrong.
At Crestmore, we take a different approach. Our methodology is fully documented, our data sources are disclosed, and we publish an annual forecast accuracy review — including the calls we got wrong. We believe that accountability is the foundation of trust, and trust is the foundation of long-term client relationships.
Below, we describe in detail how each of our four research stages works, what tools and data sources we use, and how we ensure quality across all published output.
Primary Data Collection
The foundation of every Crestmore report is primary research — not secondary synthesis.
Our analysts conduct structured primary research programmes for every coverage area. This includes:
Industry Interviews
We conduct 20–40 primary interviews for each major research report, with a structured questionnaire protocol designed to surface non-consensus views. Interview subjects include: producers, traders, consumers, regulators, logistics providers, and financial market participants.
Field Research
For commodity and country-risk coverage, our analysts conduct regular in-country visits — attending industry conferences, visiting project sites, and meeting government officials. In 2025, our team completed field research visits across 19 countries.
Proprietary Surveys
We maintain a proprietary sentiment and conditions survey across selected industry verticals, conducted quarterly. Survey results are published exclusively to clients and feed directly into our quantitative models.
Trade and Official Data
Primary interview data is cross-referenced against official trade databases, company filings, regulatory disclosures, and — where available — satellite and remote-sensing datasets. We maintain subscriptions to over 40 specialist data providers.
Quantitative Modelling
Every Crestmore forecast is built on a documented, proprietary model — not expert guesswork.
Our modelling team has developed and maintained in-house supply-demand, price, and risk models across all coverage sectors. Key features of our modelling approach:
Supply-Demand Frameworks
For commodity sectors, we maintain detailed supply-demand balance models extending to a 10-year horizon. These are built bottom-up from individual project and asset-level data, aggregated to market-level with macroeconomic overlay.
Scenario Architecture
We publish three-scenario forecasts (base, bull, bear) with explicit probability weights and clearly-stated assumption differentials. This allows clients to assess sensitivity to key variables rather than anchoring on a single point estimate.
Country Risk Scoring
Our proprietary country risk scores are constructed from 48 weighted sub-indicators across political, regulatory, fiscal, social, and external risk dimensions. The scoring framework is reviewed and recalibrated annually.
Backtesting and Calibration
Our models are backtested against historical outcomes on a rolling 5-year basis. Where systematic biases are detected, we disclose them and adjust the methodology accordingly.
Expert Review & Calibration
No research leaves Crestmore without a structured peer review.
Every research deliverable undergoes a two-stage review process before publication:
Internal Peer Review
Each draft report is reviewed by a senior analyst who was not involved in its preparation. The reviewer assesses: analytical consistency, data accuracy, logical coherence of conclusions, and appropriate expression of uncertainty. All review comments are documented.
Methodology Audit
For reports involving new model outputs or significant methodology changes, an independent methodology audit is conducted by our Research Standards Committee — a small panel of senior analysts and external advisers who review but do not contribute to the research itself.
External Expert Consultation
For technical subjects outside our core team's expertise, we engage with a network of 200+ specialist consultants and academics. External contributors are disclosed in the report appendix.
Actionable Delivery
Research that sits unread has no value. We design our deliverables to be used.
Decision-Ready Formats
Every report begins with a one-page executive summary structured around: key findings, central scenario forecast, primary risks, and investment implications. Senior executives can read this page and act on it; analysts have the full report for deeper review.
Analyst Briefings
All published reports are available for client briefing calls with the lead analyst. For retainer clients, these calls are available on demand; for report subscribers, scheduled calls are available within 5 days of publication.
Data Downloads
The datasets, model outputs, and scenario tables underlying each report are available as Excel downloads for all clients. We believe clients should be able to work with our numbers — not just read them.
Annual Forecast Accuracy Review
Every January, we publish a full review of the prior year's forecasts — including our misses. We track our price forecasts, demand estimates, and country risk scores against realised outcomes, and publish mean absolute error metrics for each sector. This report is available to all clients and prospective clients on request.
We believe forecast accountability is non-negotiable for a firm that asks institutions to make decisions based on its work.
Request our latest accuracy review →