ML · AI · MCP ML-native · Agent-native · Auto ABS

Machine learning
for the life of your loans.

Walfactor is the ML-native, agent-native analytics platform for the Auto ABS market. Data feeds, dashboards, and production models for bond investors, issuers, regulators, and fintechs — from origination through securitization.

Explore the platform See the agent demo

Built by engineers and practitioners who've shipped credit systems, scaled data platforms, and deployed ML at JPMorgan Chase Bank of America SoFi Netflix AMD

API Request Also via MCP

# FICO stratification for AART 2024-1 curl https://api.walfactor.com/v1/strat \ -H "Authorization: Bearer sk_..." \ -d '{"deal": "AART 2024-1", "strat": "credit_score", "snapshot": "2025-06-30"}' # Response: 23ms # 68,736 loans · ML-scored · live

DealScope Output Live

FICO Bucket	Count	% Bal
750+	18,204	32.1%
700 – 749	22,871	28.7%
650 – 699	14,328	19.4%
600 – 649	8,102	12.3%
< 600	5,231	7.5%

WA FICO: 724 · WA APR: 5.60% · Pool Factor: 7.4%

32M+

Auto loans & counting

600+

Active ABS deals

19K+

ABS-EE filings parsed

1B+

Cross-sectional datapoints

Clean data. Transparent models.
ML you can actually audit.

Verified, high-granularity datasets and production-grade models — purpose-built for the modern auto finance and securitization ecosystem.

Explore the platform

NEW Agent-Native

Your Claude. Your Cursor.
Our data.

One install. Every Auto ABS deal, every strat, every snapshot — exposed as tools your agent can call. No dashboards required, no glue code.

Terminal · MCP Client

▸ search_deals(issuer="Ally", limit=3)

▸ compare_deals(ids, strat="fico")

▸ stratify("AART 2023-1", "fico", snapshot="latest")

DealScope Result Live

AWAITING QUERY

Computing distribution…

FICO Distribution · 3 Deals Compared

Deal	WA FICO	Sub-600	Subprime Tail
AART 2024-2	741	3.8%
AART 2024-1	738	4.2%
AART 2023-1	712	9.1%	← worst tail

AART 2023-1 carries 2.4× the sub-600 concentration of the 2024 vintages. Recommend excluding or re-pricing the junior tranche before including in the cohort.

⊞

Works with what you use

MCP

Claude Code, Cursor, Zed, or any client that speaks the Model Context Protocol. Your existing agent workflow stays unchanged — our data just shows up as tools.

Clients supported5+

✓

Same data. Same auth.

Unified

Every dashboard metric is also a tool. One API key, one schema, one source of truth. No reconciliation between dashboard and agent — they see identical data.

Tools exposed8

⊘

Bundled with Enterprise

Included

MCP access is included with Enterprise at no additional cost. Shares your existing API auth and rate limit. Install in 30 seconds with one command.

Enterprise$499/mo

Install MCP Server View docs

See it in action

Don't take our word for it.

Three workflows, three audiences, one platform. Every example below uses real production endpoints.

Use Case #1

Portfolio manager screens new deals

Compare FICO distributions across 3 Ally deals in one click. No Bloomberg terminal needed.

# Compare deals side by side GET /v1/compare?deals=AART-2024-1,AART-2024-2,AART-2023-1 &strat=credit_score &snapshot=latest # 3 strats · aligned buckets · 89ms

Live Output

FICO Distribution Comparison

AART 2024-1 · WA FICO738

AART 2024-2 · WA FICO741

AART 2023-1 · WA FICO712

Δ Best vs Worst+29 pts

Sub-600 Concentration4.2% → 7.1% → 11.3%

Use Case #2

Data engineer builds a DQ monitor

Pull 30+ DPD rates across your entire portfolio via API. Feed it into your prepayment and default models.

import walfactor client = walfactor.Client("sk_...") # DQ rates for all deals in shelf dq = client.pivot( group_by="ticker", metrics=["pct_30plus", "pct_60plus"], filters={"shelf": "AART"} )

Live Output

Delinquency — AART Shelf

AART 2024-2 · 30+ DPD2.1%

AART 2024-1 · 30+ DPD3.4%

AART 2023-1 · 30+ DPD5.8%

AART 2022-3 · 30+ DPD8.2%

Shelf Avg 30+ DPD4.88%

Use Case #3

Analyst runs vintage analysis before earnings

12 stratification types, any snapshot. Origination year, LTV, state, vehicle make, FICO, delinquency — all sub-second.

# No code needed — use DealScope # 1. Pick deal: AART 2017-1 # 2. Select snapshot: 2020-08-31 # 3. All 12 strats in ~5s # Or via API: POST /v1/strat {"deal_id": 551, "snapshot": "2020-08-31"} # 12 strats · 68,736 loans · 4.9s

DealScope View

AART 2017-1 — Origination Year

201641.2% of bal

201528.7% of bal

201418.3% of bal

20178.1% of bal

Other3.7% of bal

$1.51B OLA · $102.6M ending bal · 45 snapshots

The Walfactor platform

Engineered for scale, transparency, and depth.

From ABS-EE ingestion to production ML models — every layer purpose-built for the auto finance ecosystem.

⊞

Data sources

Ingestion

Standardized ingestion of loan-level and securitization data across lenders and issuers, with schema alignment and metadata tracking.

Records normalized32M+

✓

Data integrity

Validation

Cross-checks against prospectuses, servicer tapes, and deal documents ensure modeling-grade consistency and auditability.

Checks passed92%

≡

Storage & streaming

Infra

Streaming pipelines with Redpanda, columnar storage on ClickHouse, PostgreSQL/Timescale for analytics and surveillance.

Throughput1B+ rows

◈

ML models

Production-grade prepayment, default, and CNL models with full explainability. Governance aligned to SR 11-7 and consumer credit standards.

Models in prod4+

⇄

API & MCP

Agent-native

REST + GraphQL endpoints for loan- and deal-level queries. Native MCP server — every tool callable from Claude Code, Cursor, Zed, or any MCP client.

p95 latency< 250ms

▦

Dashboards

Loan-, pool-, and deal-level dashboards with interactive filters, benchmark overlays, and scenario tools. Same data as the API.

ViewsLoan · Deal · Pool

Pricing

Transparent pricing, no surprises.

Free dashboards for everyone. API access for teams. MCP + raw data for enterprise.

Explorer

$0/mo

Full dashboard access, forever free.

DealScope — all 12 strat types
Pivot table — all dimensions
All 600+ deals
Snapshot history
Community support

Get started free

Pro

$99/mo

API access for your team.

Everything in Explorer
REST API — 10K req/day
Python SDK
Loan-level CSV exports
Custom strat definitions
Priority support

Start 14-day trial

Enterprise

$499/mo

Unlimited API + MCP + raw data + SLA.

Everything in Pro
Unlimited API requests
MCP server — Claude Code, Cursor, Zed
Raw loan-level DB access
Custom deal ingestion
Webhook alerts (DQ spikes, new deals)
99.9% SLA + dedicated support

Contact sales

FAQ

Questions.

Where does the data come from?

SEC EDGAR ABS-EE (XML) filings and servicing data, parsed and loaded nightly. Every loan-level data point the issuer reports — FICO, LTV, delinquency, payments, balances, vehicle info, and more.

How fast are the queries?

Dashboards and strats run on ClickHouse over 1B+ rows. Single-strat queries return in <100ms. Full 12-strat DealScope loads in ~5 seconds. Pivot tables: sub-second. API p95 latency <250ms.

What's the difference between free and paid?

Free gets you full interactive access — every deal, every strat, every snapshot. Pro unlocks the REST API and Python SDK. Enterprise adds the MCP server for AI coding assistants plus raw database access and an SLA.

What is the MCP server?

Model Context Protocol is an open standard for exposing data and tools to AI agents. Walfactor's MCP server lets Claude Code, Cursor, Zed, and any other MCP client call Walfactor tools directly — no glue code. One install and every deal, strat, and snapshot becomes callable from your agent. Included with Enterprise.

What about ML models?

Walfactor includes production prepayment, default, and CNL models with full explainability. Governance aligned to SR 11-7. Models are available via API to Pro and Enterprise tiers, with detailed validation documentation.

Do you support non-auto ABS?

Auto ABS is live today. Student loans, credit cards, and equipment leases are on the roadmap. Same infrastructure, different parsers.

Machine learningfor the life of your loans.

Clean data. Transparent models.ML you can actually audit.

Your Claude. Your Cursor.Our data.