Industry benchmarks

Use this collection to select financial-risk benchmarks and evaluate AMLOK for a specific industry. Consumer Lending and Business Lending define product-level cohorts. Financial Institutions and Fintech describe institution and distribution contexts. Insurtech covers trade credit, surety and premium finance. A loan can belong to several contexts; count it once in a consolidated portfolio. Each page separates published market evidence from AMLOK evaluation methods. A market delinquency rate is not a model accuracy score. The measured results page contains completed AMLOK research measurements; these industry pages do not add unexecuted model results.

Explore by industry

Industry	Segments	Primary assessment
Consumer Lending	Personal, auto, mortgage, student, rent	Affordability, payment performance and changes in household obligations.
Business Lending	SMB, commercial, business financing, MCA, working capital	Reconciled cash flow, repayment structure, leverage and liquidity.
Financial Institutions	Banks, credit unions, alternative lenders	Portfolio composition, credit performance and operational validation.
Fintech	Digital lenders, embedded finance, platforms	Distribution effects, evidence completeness and financial-event reconciliation.
Insurtech	Trade credit, surety, premium finance	Counterparty, performance and financed-premium exposure.

Dated public reference points

These observations are contextual anchors, not approval thresholds or predictions for an AMLOK customer. Preserve each source’s denominator and reporting period.

Segment	Published observation	Population and interpretation
Canadian mortgages	0.24% delinquency in Q4 2025	CMHC’s reported national mortgage delinquency measure for loans at least 90 days past due. See the CMHC release.
Canadian student loans	8% three-year default rate for the 2021-2022 repayment cohort	Value-weighted federal student-loan measure; default follows more than 270 days of arrears. See the Canada Student Financial Assistance annual report.
Canadian small-business financing	49% of surveyed debt-financing borrowers intended to use it for working or operating capital in 2024	Financing-purpose context, not a default or loss rate. See ISED’s 2014-2024 credit-condition trends.

These examples retain their original dates. They are not presented as the latest available release. Refresh a benchmark from its source before using it in a current portfolio comparison.

Benchmark contract

Record these fields with every numerical comparison:

Field	Required information
Population	Geography, product, borrower type, origination vintage, selection criteria and exclusions.
Observation unit	Borrower, account, facility, invoice, policy, bond or exposure amount.
Outcome	Exact event, past-due threshold, measurement horizon and cure treatment.
Denominator	Eligible accounts, borrowers, principal, exposure, insured sales or earned premium.
Time	Decision cutoff, observation window, outcome maturity and publication date.
Evidence	Source URL, table or field identifier, extraction date, transformation and known gaps.
Uncertainty	Sample size, confidence interval, missingness and censoring.

Do not compare a borrower-level 60-day arrears rate with an account-level 90-day rate. Do not infer a loan’s default probability from an approval rate, rental vacancy rate or aggregate banking return.

AMLOK evaluation design

For Insights, freeze the evidence available at the underwriting cutoff. For Signal, evaluate new information against the previously available baseline and measure the timing and usefulness of a change signal. A later repayment outcome must not enter the input snapshot. Compare the same eligible cohort, evidence and decision date across a tabular baseline, the AMLOK system under evaluation and any eligible frontier-model comparator. Keep borrowers and related entities out of both training and evaluation sets. Report classification, calibration, evidence quality, abstention, latency and cost separately. See the evaluation protocol, frontier comparison and document benchmarks. Institutions define lending and intervention policies. These research pages introduce no servicing workflow, automatic credit decision, new paid product or public API endpoint.

Research program

Data and models

Risk analysis