Where the numbers come from.
Every source is mapped into one canonical data model, with provenance on every field.
Data sources
Mapped into one canonical data model, with provenance on every field- Berkeley VROD
- Offsets database
- Issuances and retirements across the major registries
- CarbonPlan
- Research & ratings
- Independent project assessments and methodology research
- CAD Trust
- Registry metadata
- Harmonised project and unit records shared by registries
- Carbonmark
- Marketplace
- Listings and live prices for tokenised and registry credits
- Regen Network
- Registry & marketplace
- Ecological credits with on-chain issuance
- Toucan · KlimaDAO
- On-chain pools
- Tokenised credit pools and on-chain pricing
- EU ETS
- Compliance market
- Reference prices from the regulated EU emissions market
Methods in plain words
Canonical data model
One shared structure for certificates from every source, so the same credit listed in three places is recognised as one.
Fair value
A machine-learning estimate of what a credit should cost, given its standard, type, vintage, risks and co-benefits.
SHAP
A method that splits a model’s estimate into contributions per feature, shown in euros, so users can see why a value is high or low.
Anomaly detection
Flags listings whose price or registry history looks unusual compared with similar credits, as a prompt for human review.
Smart procurement routing
Weighs price against evidence quality, vintage and project risk across venues and OTC quotes, and can split one order across counterparties.
Human in the loop
The agent prepares recommendations and orders. A person always approves before anything is bought.
Source data: CarbonPlan OffsetsDB (carbonplan.org/research/offsets-db). Project types incorporate the Voluntary Registry Offsets Database, Berkeley Carbon Trading Project, CC BY 4.0. Signals and summaries are derived by QAVCM. Data as of 2026-10-02.