Query millions of Fannie Mae and Freddie Mac loans and pools, explore cross-month trends, and run AI-powered analysis — all from a single platform.
From raw disclosure files to production insights in one platform.
Ask questions in plain English. Get answers about delinquency trends, servicer performance, and portfolio risk backed by real data.
Full Python environment with Polars, pandas, and statsmodels. Run ad-hoc analysis directly against Parquet files and TimescaleDB.
Browse 28M+ loans per month from FNM and FRE. Filter by snapshot month range and explore UPB, rate, FICO, LTV, DTI, and more.
Explore 1M+ pools per month. Drill into weighted-average metrics, security characteristics, and cross-month performance trends.
Both agencies' monthly loan and pool disclosure files are ingested into a unified TimescaleDB hypertable partitioned by month. Cross-agency comparisons are a single query.
import polars as pl
df = pl.scan_parquet(
"data/parquet/loans/**/*.parquet",
hive_partitioning=True,
)
result = (
df.filter(
pl.col("year").is_in([2024, 2025, 2026])
)
.group_by(["agency", "year",
"month", "property_state"])
.agg([
pl.col("current_investor_loan_upb")
.mean().alias("avg_upb"),
pl.col("classic_fico")
.mean().alias("avg_fico"),
pl.len().alias("loan_count"),
])
.sort(["year", "month"])
.collect()
)Create a free account and start analyzing Fannie Mae and Freddie Mac disclosure data in minutes.
Get started