ML Model Performance

About this dataset

ML Model Performance is a Free dataset in the Risk & Failures group of the BankRegReports Data API. It is sourced from BRR ML model.

Production model metadata (per horizon) plus held-out backtest results: AUC-ROC (ranking accuracy), AUC-PR and Brier score. A horizon whose held-out window holds only one outcome (for example every bank-quarter failed) cannot be scored: its backtest row keeps the sample and failure counts, leaves the three metrics null and explains why in metric_note.

Endpoint

GET /api/v1/risk/model/

Example request

curl "https://api.bankregreports.com/api/v1/risk/model/" \
  -H "Authorization: Bearer $BANKREG_API_TOKEN"

Responses are JSON in a {"data": ..., "meta": {...}} envelope. Get a token on the setup page.

When to use it

ML Model Performance sits in the Risk & Failures group of the Data API. Reach for it when your work needs this data straight from the API: it comes back as JSON you can load into a notebook, a spreadsheet or a scheduled pipeline, with the dataset key model_performance naming it the same way in the REST path, the Python SDK and the OpenAPI document.

Access and authentication

This dataset is in the Free tier. Send your API token in the Authorization header as a Bearer token, using a GET request to the endpoint above. The Free plan is the first plan that includes it. Check the plan comparison for request limits, and the data update schedule for when new filings reach the API.

Related datasets

• High-Risk Banks: Banks ranked by the exploratory failure-risk screening score (uncalibrated, not a probability of failure).
• Failure Risk Score: Exploratory 1-year failure-risk screening score (not a calibrated probability) + safety verdict for one bank.
• Recent Failures: Most recent bank failures (FDIC).
• Failures by Year: Bank failure counts and total assets by year.

Keep going

Browse every dataset, follow the setup walkthrough to make a first call, use the Excel add-in for formulas, or connect an assistant through the MCP server.

Python SDK

Install with pip install bankregreports and set BANKREG_API_TOKEN to your key.

import bankregreports

brr = bankregreports.BankReg()  # reads BANKREG_API_TOKEN
result = brr.dataset("risk_model")