Signal Sources

Behavioral signals of creditworthiness

Everyone who manages money leaves a data trail: prepaid top-ups, on-time utility payments, and regular trips between home and work. Supersymm reads that trail.

Signal Categories

Five behavioral signal categories

Each category reflects a distinct aspect of financial behavior. Signals are processed only after explicit borrower consent.

Mobile Activity

Prepaid recharge frequency, consistent airtime spending, and device stability over time. Top-up frequency and consistency across months indicate financial discipline, especially among borrowers using prepaid rather than postpaid mobile plans.

Why it predicts repayment: Regular prepaid top-ups require budgeting, the same skill associated with on-time loan payments. Irregular or lapsed top-ups can indicate income instability before a formal payment is missed.

Utility Payments

The regularity and promptness of electricity and water payments. In many low-income households, utilities take priority over discretionary spending and often rent. We measure both payment regularity and lead time before the due date.

Why it predicts repayment: Utility payments closely resemble loan repayment: they are fixed, recurring, and obligatory. Consistent utility payments show the repayment behavior lenders need to assess.

Geolocation Patterns

Regularity of home and work locations, derived from app-permissioned coordinate data. The signal measures how consistently a borrower appears at the same places, times, and days across a 90-day window.

Why it predicts repayment: Residential stability is associated with lower credit risk across demographic groups, including borrowers without formal credit files. Reliable travel between home and work reflects a stable routine that correlates with meeting financial obligations.

Peer Transfers

Mobile money transfer cadence, including frequency, consistency, and size relative to inferred income. This captures payments to family, landlords, and service providers through mobile platforms. Transfer patterns reveal active financial management in informal economies.

Why it predicts repayment: Reliable peer transfers indicate that a borrower manages formal and informal obligations on schedule. This activity is not visible to the bureau but can predict formal credit behavior.

App Engagement

Session regularity and usage frequency in financial apps used by borrowers on mobile-first lender platforms. This is available to lender partners whose borrowers opt into app-level data sharing. Session timing and frequency indicate active financial awareness.

Why it predicts repayment: Borrowers who monitor their financial app tend to stay more engaged with their financial position, which correlates with proactive repayment management. The signal is strongest among digital-native borrowers and consumer fintech customer bases.

Consent Model

Supersymm processes only signals the borrower has explicitly authorized the lender to share. The API ingests no data without a confirmed consent flag in the request.

The lender owns the consent flow: the loan application records disclosure and authorization. Supersymm's API requires consent_confirmed: true in every scoring call. The API gateway rejects requests without confirmed consent before signal processing.

Lenders remain accountable for borrower disclosures under applicable data protection requirements, including LGPD. Supersymm does not store raw behavioral data. It processes only derived features, which are purged 12 months after the last lender query for a borrower ID.

What Supersymm does not store

Supersymm never persists raw call records, location coordinates, or transaction histories. The pipeline extracts features, including aggregated patterns such as "recharge frequency over 90 days", then discards the underlying raw data. Supersymm retains the score and derived features; source data remains in your system and on the borrower's device.

Data Quality

Measuring signal quality

90-day

Default signal window

Signals are weighted by recency across a 90-day rolling window, so recent behavior counts more. Shorter windows are available for borrowers with rapidly changing profiles.

3+ signals

Minimum for full confidence score

A full confidence score requires at least 3+ signals. Scores based on fewer available categories receive a Medium or Low confidence band.

Bureau cohort

Validation benchmark

Early-access pilot data validates signal predictive power against bureau-scored populations. Score bands are calibrated so each band indicates a default probability comparable to the same band in bureau scoring.