Machine-learning data pipelines, quantitative credit models and analytical tools — for institutions that run their own credit risk analysis.
Machine-learning pipelines that clean, match and structure traditional financial, alternative and on-chain data into model-ready credit datasets, with entity resolution, data-quality checks and anomaly detection built in.
AI-enhanced probability of default (PD), loss given default (LGD), expected credit loss and portfolio credit risk models, with macro scenario analysis and stress testing. Clients run the models on their own portfolios; every model version is documented in a version-controlled audit trail, and on request a cryptographic hash can be recorded on-chain.
Software that brings quantitative credit analytics into institutions' own workflows: scenario and sensitivity analysis, model monitoring and back-testing, and secure API access to the models. Results are generated from the data and settings each client chooses, for that client's internal use.
AI-driven data processing, quantitative credit models and analytical tools for institutions' own credit risk analysis
Machine-learning pipelines that turn financial, alternative and on-chain data into model-ready credit datasets:
AI-enhanced models that clients run on their own portfolios, with version-controlled audit trails:
Software that embeds the models in clients' own workflows:
Clients access the models through a secure API or a client-specific environment and run them on data they select. Results are generated for that client's internal use under its contract with us; they are not published, broadcast on public networks or blockchains, or provided to other clients. Where a client requests it, only a cryptographic hash of a result is recorded on-chain for audit purposes.
Quantitative credit analytics for institutions across traditional and digital-asset markets
Credit risk tools for bank risk and finance teams: machine-learning-enhanced IFRS 9 expected credit loss models, stress-testing and regulatory capital calculation tools, and portfolio monitoring dashboards with version-controlled audit trails. Credit decisions remain with the bank.
Data and modelling tools for fixed income teams covering traditional bonds and tokenised instruments: data cleansing and normalisation, cash-flow and scenario modelling, and portfolio analytics that clients configure and run themselves. Investment and credit decisions remain with the client. Kriteria Analytics does not provide investment advice, portfolio management or credit ratings.
On-chain data tools for digital-asset teams: protocol and liquidity-pool data monitoring, smart contract data analysis, and models trained on on-chain data that clients run for their own risk analysis of DeFi protocols and RWA platforms. Results are generated by each client from its own data, for its internal use.
Data tools for corporate treasury and procurement teams: supplier and counterparty data monitoring using alternative data, machine-learning cash-flow forecasting models, and data integration for blockchain-based trade-finance and receivables platforms that clients operate. Credit and commercial decisions remain with the client.
Technical papers on machine-learning credit modelling, quantitative methods and data engineering
Our first papers are in preparation. They describe analytical methods and tools and are not credit ratings. Check back soon.
A quantitative credit analytics company within the Kriteria group
Kriteria Analytics Limited provides technology and support services for AI-driven quantitative credit analytics: data processing, quantitative modelling, and the development and maintenance of analytical tools. It is a separate legal entity within the Kriteria group and a sister company of Kriteria Ratings Limited; both are wholly owned by Kriteria Holdings Limited.
Our aim is to make quantitative credit analytics practical for more institutions, with models that are transparent, documented and reproducible, so that clients can run, test and explain their own credit risk analysis.
Our technology combines machine learning with established quantitative credit methods: transformer-based NLP models that extract data from financial documents, statistical and machine-learning PD and LGD models, Monte Carlo portfolio simulation, and version-controlled audit trails with optional on-chain hash anchoring. It is designed to support clients' own model governance.
What sets Kriteria Analytics apart:
To find out how our data, models and analytical tools can support your institution's own credit risk analysis, please get in touch.
For enquiries about our data, models and analytical tools. Kriteria Analytics does not accept credit rating requests.
