What a Reg D Score means, and what it doesn't
The Reg D Score is a single number that summarizes how many risk signals a Regulation D offering shows, derived by a rules-based engine and traceable point by point. Knowing what that number represents, and what it deliberately does not represent, is what makes it useful in real diligence work.
What the score is
The Reg D Score is a reproducible 0-100 risk score for a Regulation D offering. It runs in one direction: higher means more risk signals were found, lower means fewer. It is not a grade for the issuer, a probability of loss, or a verdict on the deal. It is a structured count of the concerns that a defined set of rules looks for, put on a common scale so that one offering can be compared against another on consistent terms.
Two offerings scored on the same day with the same inputs will receive the same number, because nothing about the score is left to judgment at the moment of scoring. That property, determinism, is the foundation everything else in this article rests on. The score is meant to be a stable, defensible starting point for analysis, not the last word on it.
A measure of signals, not of guilt
A high Reg D Score says the engine found a lot of things worth examining. It does not say the offering is fraudulent, and a low score does not say it is safe. The number points your attention; it does not replace your judgment.
How it is built
The score is computed by a rules-based engine that evaluates an offering across six domains. Each domain contributes flags, and each flag adds a defined number of points to the total. Together the domains cover the places where risk tends to show up in private placements.
- Disclosure integrity: whether the filing's own facts are complete, consistent, and internally coherent
- Behavioral and issuer patterns: how the issuer and the offering behave relative to common patterns
- Principal background: the people behind the deal and what the public record shows about them
- Enforcement and litigation exposure: regulatory actions, proceedings, and related history
- Structural factors: how the offering and the entity are constructed
- Adverse media: negative coverage and public signals tied to the offering or its principals
Language models play a deliberately narrow role here. They help pull data out of documents and they draft the written explanation that goes with the score. They never set the number. The points come from the rules engine alone, which is what keeps the result reproducible rather than dependent on how a model happened to phrase things on a given run.
Why determinism and the derivation appendix matter
Every point in the score is traceable to a specific flag, and those flags are laid out in a score-derivation appendix that comes with the result. If an offering scores higher than you expected, you can open the appendix and see exactly which flags fired, in which domain, and how many points each added. The score is auditable. It is not a black box that hands you a number and asks you to trust it.
A deterministic engine plus a line-item appendix is what makes the score usable in a professional setting. Reproducibility means the same inputs always produce the same output, so the score can be re-run, cited, and defended. Traceability means a reviewer can question any part of it, disagree with a particular flag, and weigh it accordingly. The point is not to end the conversation about an offering but to give that conversation a precise, shared starting point.
Auditable by design
If you cannot see why a score is what it is, you cannot rely on it. The derivation appendix exists so that every point can be inspected, questioned, and, where warranted, set aside by the analyst.
Confidence, and the missing-information toggle
A risk score alone can mislead if you do not know how much information it was built on. That is why every result also carries a separate confidence score, showing how complete the underlying information was when the offering was scored. A high risk score computed on thin information is flagged as low confidence, which is a very different situation from the same risk score computed on a full set of documents.
Confidence is something you can improve. Uploading the deal documents and confirming which documents actually exist gives the engine more to work with and raises confidence in the result. The score and its confidence are meant to be read as a pair: the score tells you how concerning the picture looks, and the confidence tells you how much of the picture you are actually looking at.
An optional toggle controls how gaps in information are handled. With "treat missing information as a risk signal" turned on, gaps count against the score, so the absence of something you would expect to see is itself treated as a concern. With it turned off, the engine scores only what is affirmatively known and lets missing data lower confidence instead of inflating risk. Both modes are legitimate, and they answer different questions. The conservative setting asks how the deal looks if you assume the worst about what is missing. The strict-evidence setting asks what the available facts alone support.
Reading the bands: risk and confidence together
Results are placed in labeled bands, each with a plain-language meaning and explicit caveats, so the raw number is never presented without interpretation. The band tells you, in words, what a score in that range generally indicates and what it does not. The caveats are part of the output, not fine print bolted on afterward.
The habit that matters most is reading the band alongside the confidence score rather than on its own. A concerning band on high confidence is a strong cue to dig in. The same band on low confidence is mainly a cue to gather more information before drawing any conclusion. A reassuring band on low confidence is the easiest result to over-trust and deserves the most skepticism, because it may just reflect how little was known rather than how sound the offering is.
- Concerning band, high confidence: a well-supported signal to investigate closely
- Concerning band, low confidence: gather more information before concluding anything
- Reassuring band, low confidence: the easiest result to over-trust, so treat it as preliminary
- Any band: read the plain-language meaning and caveats that come with it, not just the number
What it does not mean, and where it fits
Be explicit about the limits. The Reg D Score is decision support for due diligence. It is not investment advice, not a rating, not an audit, and not a guarantee. It does not establish that fraud has occurred, and it does not predict returns. A score is a draft for analyst review: a structured, traceable summary of signals that a human is expected to read, question, and build on, not a conclusion to be acted on by itself.
In a diligence workflow the score sits near the front. It triages a set of offerings or frames a single one, telling you where to spend attention first by showing which domains lit up and how strongly. From there the derivation appendix turns each flag into a concrete thread to pull, confidence tells you whether you have enough information to trust the picture yet, and the toggle lets you test how the result changes under a stricter or more conservative reading of the gaps.
Used this way, the score does what a good instrument should: it makes the analyst faster and more thorough without pretending to be the analyst. The number points you toward the questions worth asking; the answers still live in the offering documents, in independent verification, and in professional judgment.
Key takeaways
- The Reg D Score is a deterministic, reproducible 0-100 risk score where higher means more risk signals. It is not a rating, an audit, or a prediction of returns.
- A rules-based engine sets every point across six domains. Language models only pull data and draft the explanation, never the number.
- Every point is traceable to a specific flag in the score-derivation appendix, so the score is auditable rather than a black box.
- Read the score with its separate confidence score. A high risk score on thin information is flagged low confidence, and uploading documents raises confidence. The missing-information toggle decides whether gaps count as risk or just lower confidence.
- Treat the score as decision support and a draft for analyst review that sits at the front of diligence. It points to the questions, and the answers live in the documents and independent verification.
Run this analysis automatically
The Reg D Score® canvasses EDGAR filings, principal background, enforcement records, and adverse media, then returns a cited, reproducible risk assessment.
This article is for general educational purposes only and does not constitute legal, financial, or investment advice. EDGAR-INSIDER® is not a regulator, auditor, or investment advisor. Independently verify all information and consult qualified professionals before acting on it.