Fitch-dv01 Non-QM Benchmark

The Fitch-dv01 Non-QM Benchmark provides a more holistic view of the U.S. Non-QM market. Powered by loan-level data, the benchmark is anonymized, standardized, and AI-ready. Built from dv01’s role as Loan Data Agent and from Fitch Ratings–rated transactions, the benchmark aggregates approximately 330,000 loans, representing more than $155 billion in original loan balances and $93 billion in outstanding balances since 2019. <p><br/></p> The benchmark reflects dv01’s data standardization model, which includes consolidation of hundreds of borrower loan docs into 15 consistent reporting fields, while preserving underlying reported values for consistent analysis. <p><br/></p> Delivered in a consistent, analysis-ready schema, the dataset enables advanced modeling, benchmarking, and AI-driven insights into origination trends, borrower credit quality, delinquency, prepayment, and loss dynamics. Users can query, visualize, and integrate the data directly within Snowflake to support machine learning pipelines, portfolio optimization, and risk analysis workflows. Start your free trial today with a 1% sample of our benchmark, made available in your Snowflake warehouse immediately.

搜集汇总

背景与挑战

背景概述

Fitch-dv01 Non-QM Benchmark是一个美国非合格抵押贷款市场的标准化基准数据集，基于约33万笔贷款数据，总原始贷款余额超过1550亿美元，自2019年起提供。该数据集经过匿名化和AI就绪处理，支持分析贷款趋势、信用质量和风险动态，可用于机器学习、投资组合优化和风险分析工作流。