📊 Live static build. Every chart below is computed from the trained model's exported coefficients, running in this browser tab — nothing is fetched from a backend. Portfolio-wide totals are precomputed over the full 396,030-loan dataset; the explorer and "both ends" examples use a 160-loan sample.
Credit Portfolio Analytics
How the loan book scores today, in plain terms — where risk concentrates, whether the score is trustworthy, and what's driving it. Technical model detail lives in the collapsed panel at the bottom.
Risk tier mix
Every loan is grouped into one of four risk tiers based on its score. Actual historical default rates climb tier by tier — proof the score ranks risk correctly, not just a label.
Tier
Loans
Volume
Actual default rate
Loan explorer
Browse individual scored loans from the sample — search, filter, and see the same plain-English reasoning an underwriter would see, computed live from the model. Click a tier segment above to jump here filtered.
Loan
Amount
Purpose
Grade
Term
Risk score
Tier
Why
Policy
Loans at both ends
Concrete examples, not just statistics — clean Low-risk cases next to clear Very-High-risk cases, with the same reasoning underwriters would see. Outlier records (e.g. implausible income or public-record values) are filtered out so these stay representative.
✓ Low risk examples
‼ Very High risk examples
Where the risk concentrates
Default rate by segment — same scoring, sliced four ways.
By LoanTap grade
A (safest) → G (riskiest) — LoanTap's own tiering, shown against actual outcomes
By loan purpose
Top categories by volume; remainder folded into "Other"
By home ownership
Renters default at a meaningfully higher rate than owners
By loan term
60-month loans default at roughly double the rate of 36-month loans
All segments, ranked
Sortable by any column. Rows outlined in red cross the 25% default-rate watch threshold.
Dimension
Segment
Loans
Volume
Default rate
Status
Default rate over time
By year the loan was issued. The most recent years are understated — see note below the chart.
Year
Loans issued
Actual default rate
What's driving risk
Plain-English read of the model's strongest signals — translated from coefficients, not shown as raw numbers.
Pushes risk up
↑ LoanTap's own grade / sub-grade assignment (expected — that's its job)
↑ Higher debt-to-income ratio
↑ 60-month terms, vs. 36-month
↑ Higher revolving credit utilization
↑ More open credit accounts
Pushes risk down
↓ Higher annual income
↓ Longer overall credit history (more total accounts)
↓ Longer employment tenure
⚠Flagged for governance review, not shown as a stated driver: a few signals — interest rate, and the "high-risk grade" flag — point the statistically "wrong" way once grade/sub-grade are already in the model. That's a known overlap artifact (grade already encodes the interest rate that was set), not a real causal effect, and shouldn't be read as "higher interest rate means lower risk."
For analysts — model metrics & methodology ▶
Underwriter WorkbenchCase-level review with plain-English reasoning and an AI-generated summary (grounding-checked before display).