Signal Sources
What feeds the default-risk score
The score is a composite built on 40+ behavioral signals drawn from objective financial data sources — bank transaction records and processor settlement data. These are not self-reported or member-submitted; they are observed behavioral patterns derived directly from financial account activity.
NSF frequency
Non-sufficient-funds events observed in linked bank transaction data. A rising NSF rate signals cash-flow deterioration before it is visible in statements.
Days cash on hand
Derived from bank balance trajectory relative to average daily outflows. Low or declining days-cash is a leading indicator of payment stress.
Processor chargeback velocity
Rate and trend of chargebacks coming through processor settlement data. Elevated chargeback velocity often precedes a revenue shortfall.
Industry-specific stress markers
Behavioral patterns weighted by merchant industry vertical, because the same balance fluctuation means different things for a restaurant versus a contractor.
Revenue trajectory
Month-over-month and trailing-90-day revenue trend from bank and processor settlement data. Used as a positive signal for renewal probability and as an early-warning negative when trajectory turns.
Stacking exposure
Detection of concurrent advance positions visible through transaction pattern analysis. Multiple simultaneous MCA obligations are the leading structural cause of merchant default.
Update Cadence
Scores are updated weekly, not monthly or on demand
Merchant financial health can deteriorate meaningfully within a single week. Comet refreshes every merchant's score on a weekly cycle so the signal you see reflects current behavior — not a snapshot from 30 or 60 days ago.
Weekly
Score refresh cadence for every merchant in your portfolio
40+
Behavioral signals composited into each merchant's score
Live
Revenue trajectory signal updated as bank and processor data comes in
Methodology Contrast
Observed transaction data vs. crowdsourced submissions
The fundamental distinction between Comet's methodology and a crowdsourced default-reporting model is the origin of the data. Comet's inputs are observed — pulled from financial data feeds that reflect what is actually happening in a merchant's accounts. Crowdsourced models depend on what members choose to submit.
Comet
Observed behavioral signals
✓ Objective
Signals are derived from bank transaction records and processor settlement data — financial behaviors that are recorded automatically and are not subject to the submitter's intent, timing, or selectivity. A merchant can't opt out of having an NSF event; it either happened or it didn't.
Crowdsourced model (e.g. DataMerch)
Member-submitted reputation records
Requires member action
Records exist only when a member submits them. A merchant who defaulted but whose funder never submitted a record is invisible. Records can be delayed, incomplete, or reflect only the funder's perspective. Positive signals — who will fund well — are structurally absent.
Scope and Limitations
What this methodology does and does not claim
Comet's score is a risk indicator, not a guarantee. It is derived from observed financial behavior and is designed to surface patterns that correlate with merchant stress — it does not predict the future with certainty, and no underwriting signal does. The score is one input into your funding decision, not a replacement for your underwriting judgment. Signals are based on data available at the time of each weekly refresh; material changes between refresh cycles may not be immediately reflected.