The Signal Engine assigns every tracked company a Distress Probability Score (DPS) from 0 to 100 — a continuously updated estimate of financial distress based on observable public data signals.
DPS is calculated in three stages: signal ingestion and classification, temporal decay adjustment, and cross-signal multiplier application. Each stage is designed to reward convergence — multiple independent indicators pointing in the same direction — which is a far more reliable predictor of actionable distress than any single data point.
A single isolated signal rarely justifies outreach. The model is calibrated to reward signal convergence — when multiple independent data sources agree that a company is in distress. This reduces false positives in the outreach queue, which matters because a bad first impression in this market is costly.
Signals are classified into three tiers based on predictive strength. Tier 1 signals are hard, verifiable legal or financial events. Tier 2 signals are financial indicators with strong predictive power but less certainty. Tier 3 signals add behavioral context but rarely trigger action alone.
Legal and financial events sourced directly from court filings, government records, and regulatory databases. These are the highest-confidence distress indicators. 8–15 pts each
| Signal | Base Points | Source | Status | Notes |
|---|---|---|---|---|
| UCC Lien Filing — New | 15 | Utah SOS | In Progress | New secured debt filing against the company |
| UCC Lien Filing — Amendment | 12 | Utah SOS | In Progress | Amendment or continuation of existing lien |
| Federal Tax Lien | 15 | IRS / LiensDirect | Not Built | Unpaid federal taxes — severe distress signal |
| State Tax Lien | 12 | SLCo Recorder | Live | Unpaid state taxes recorded against property |
| Judgment Lien | 12 | SLCo Recorder / CourtListener | Live | Court-ordered money judgment recorded |
| Notice of Default | 15 | SLCo Recorder | Live | Pre-foreclosure filing — acute distress |
| Bankruptcy Filing | 15 | CourtListener / PACER | Live | Chapter 7, 11, or 13 filing |
| Vendor Lawsuit | 10 | CourtListener | Live | Supplier or vendor suing for non-payment |
| WARN Act Notice | 12 | DOL WARN | Live | Mass layoff or plant closure notification |
Credit and operational indicators with strong predictive power. These signals rarely trigger action alone but amplify Tier 1 signals significantly via co-occurrence multipliers. 3–8 pts each
| Signal | Base Points | Source | Status | Notes |
|---|---|---|---|---|
| FDA Recall — Class I | 8 | openFDA | Live | Serious health hazard — CPG-specific, major ops disruption |
| FDA Recall — Class II | 5 | openFDA | Live | Moderate health risk recall |
| FDA Recall — Class III | 3 | openFDA | Live | Minor violation recall |
| Debt Maturity Approaching | 8 | Manual / PitchBook | Manual | Known debt maturing within 12 months |
| Retailer Delisting | 6 | News / Manual | Manual | Removed from major retail shelf |
| Co-packer Dispute | 5 | News / Court | Live | Manufacturing partner conflict |
| Facility Closure | 7 | News / WARN | Live | Operating location shut down |
| Insurance Gap | 4 | Manual | Manual | Known coverage lapse or non-renewal |
Behavioral, market, and management signals. These add context and contribute to multiplier triggers but carry low individual weight. 2–4 pts each
| Signal | Base Points | Source | Status | Notes |
|---|---|---|---|---|
| Negative Press Coverage | 3 | Google News / RSS | Live | Adverse media — financial, operational, or reputational |
| Executive Departure | 4 | News / LinkedIn | Not Built | C-suite or key operator exit |
| Glassdoor Rating Drop | 2 | Glassdoor | Not Built | Internal culture deterioration signal |
| Supplier Change | 2 | News / Manual | Manual | Key supplier relationship disruption |
| Location Closure | 3 | News / Manual | Manual | Retail or office location closed |
| Social Media Silence | 2 | Social | Not Built | Sudden drop in brand communication |
Every signal loses weight as it ages. A lawsuit filed yesterday is far more actionable than one filed two years ago. The decay function uses a stepped model — not linear — to reflect how distress signals age in practice.
Stepped decay means signals drop sharply in the first year (reflecting how quickly distress situations evolve), then stabilize at a 10% floor. The floor is intentional: historical distress patterns are permanently informative even when the acute situation has resolved.
Signals never fully expire. A lien filed 3 years ago carries 10% of its original weight indefinitely. A company that has been through multiple lien cycles is structurally riskier than one with a clean history — even if those signals are old.
Multipliers are the most important part of the scoring model. They reward signal convergence — when multiple independent data sources agree that a company is in distress.
A company with one lawsuit scores modestly. A company with a lawsuit, a tax lien, three negative news articles, and an executive departure is a fundamentally different situation. Multipliers ensure the model reflects this difference.
Applied to both Tier 1 and Tier 2 scores when at least one signal exists in each tier. A hard legal signal combined with a financial stress indicator compounds the risk significantly.
Applied to Tier 1 score when 3 or more soft signals are present alongside at least one hard signal. Behavioral deterioration amplifies legal and financial risk.
Applied to Tier 1 score when 3 or more distinct Tier 1 signal types are present. Multiple independent hard signals converging is the highest-confidence distress pattern in the model.
Additive bonus when 3+ new signals are detected within 30 days. Rapid accumulation indicates acute deterioration — not slow-burn distress. Scales: 3 signals = +5 pts, 5+ signals = +10 pts.
Applied when known debt matures within 12 months and at least one Tier 1 signal is active. Refinancing pressure combined with legal stress creates high urgency for outreach.
Multipliers do not compound with each other. The highest applicable multiplier wins for Tier 1. The velocity bonus is always additive regardless of which multiplier applied.
DPS bands determine what the Engagement Engine does next. Automated action begins at Elevated (41+). Direct banker involvement escalates as score increases.
| Band | Action Owner | Engagement Engine Trigger | Response Window |
|---|---|---|---|
| Healthy | Agent | None — monitor only | — |
| Early Warning | Agent | Watchlist — intel package begins, warm path scan starts | No urgency |
| Elevated | Shared | Outreach sequence drafted — banker reviews intel package | Within 1 week |
| High Distress | Banker | Priority — banker personally approves each message | Within 48 hours |
| Critical | Banker | Immediate — all warm connections activated now | Same day |
The model only scores what it can observe. Companies with distress hidden behind private credit agreements, non-public disputes, or unreported executive changes will be underscored. Planned additions of D&B PAYDEX and LinkedIn monitoring will partially address this.
Companies in the 20–40 range may be healthy businesses with a single adverse event. Always review the signal breakdown before making outreach decisions based on Early Warning scores alone.