How we rate carbon credits
Every project is scored 0–100 and graded AAA–D on a transparent, documented rubric — the same rubric applied identically to every project, in every registry. This page explains exactly how, and what a rating does and doesn't mean.
What a CarbonDecode rating is
A rating is a decision-support signal, not investment advice. It combines public registry data with a documented rubric so that every point of every score traces to a rule you can inspect. It comes in two forms:
- — A model-based screen (the base rating on every project): produced automatically from registry data and project-type priors, calibrated across the market.
- — A project-specific review (on request): our analyst reads the project's own documents and runs a satellite/Earth-observation check, and the result is reflected in the rating. Where this has not been run, the page says so (see below).
The six dimensions
The composite is a weighted blend of six CCP-aligned dimensions. The weights reflect how much each dimension drives real-world integrity risk, and they sum to 100%.
| Dimension | What it measures | Weight |
|---|---|---|
| Additionality | Would the reductions have happened anyway? Baseline and additionality logic. | 22% |
| Over-crediting risk | Are the claimed tonnes likely inflated versus what actually occurred? | 22% |
| Permanence | How durable is the carbon, and what is the reversal exposure? | 20% |
| MRV & methodology rigor | Strength of measurement, reporting, verification, and the methodology version. | 16% |
| Double-counting & registry | Serialization, tracking, and compliance-program linkage. | 12% |
| Co-benefits & safeguards | Social and environmental co-benefits and safeguards. | 8% |
The grade scale
The composite 0–100 score maps to a letter grade. Grades of BBB and above (60+) are “investment grade” — the market's defensible floor, not a mark of excellence.
| Grade | Score | Tier |
|---|---|---|
| AAA | 90–100 | Investment grade |
| AA | 80–89 | Investment grade |
| A | 70–79 | Investment grade |
| BBB | 60–69 | Investment grade |
| BB | 50–59 | Speculative |
| B | 40–49 | Speculative |
| CCC | 30–39 | Distressed |
| CC | 20–29 | Distressed |
| C | 10–19 | Distressed |
| D | 0–9 | Distressed |
Confidence
Alongside the grade, every rating carries a confidence level reflecting how much project-specific evidence the score could draw on:
- — High — the project has issued credits and matches a well-characterized project-type profile.
- — Medium — listed, registered, or under development with a known project-type profile, but limited realized data (e.g. pre-issuance).
- — Low — little to reason about: no issuance and no applicable project-type prior.
How a score is built
Each dimension starts from a baseline, then moves up or down under transparent rules — and every adjustment appends a plain-language reason, so a score is an auditable trail, not a black box. Inputs include:
- — Project-type priors — permanence, additionality, and MRV expectations for the class of project (e.g. durable removal vs. nature-based reversal risk).
- — Registry & compliance signals — serialization strength and compliance-program linkage.
- — Buffer, vintage & verification data — buffer-pool adequacy, stale or forward-dated vintages, verification cadence, retirement ratios.
- — Hard floors & flags — override rules for serious events (e.g. a realized reversal caps permanence), plus named, severity-ranked flags.
Model-based screens & preliminary ratings
Most projects carry a model-based screen — the automated rubric above, with no bespoke document or satellite review yet performed. We label those project pages “preliminary · model-based screen” so a rating is never mistaken for a completed, project-specific due-diligence assessment.
We also flag pre-issuance projects (still under development, no credits issued): their rating anticipates type-level performance and does not reflect realized issuance, retirements, or verified outcomes. When a project-specific review is run — or when a developer shares current documentation — we update the rating and its stated basis accordingly.
Data sources & limitations
Ratings derive from public registry records — via the Berkeley Carbon Trading Project and the registries themselves — refreshed as new data publishes. The base rating does not incorporate non-public or non-final project documents unless a project-specific review has been run. Ratings are decision-support, not investment advice, and not a substitute for a buyer's own due diligence.
See it applied
Browse the ratings for every project, or pull the data programmatically.