Understanding Your ARC AI Token Report: What Every Score Means

When ARC AI generates an intelligence report on a token, it produces six scores. Each measures a different dimension of risk and opportunity. Understanding what each score represents — and how they interact — makes you a significantly better analyst.

Overall Score (0–100)

The overall score is a weighted composite of all other signals. Think of it as a single-number answer to the question: how trustworthy and valuable is this token?

• 70–100: Strong signal. On-chain health is good, team is credible, momentum is real.
• 40–69: Mixed signal. Some positive indicators but notable risks. Proceed with scrutiny.
• 0–39: High risk. Multiple red flags detected. Do your own research before engaging.

The overall score deliberately weights scam probability heavily — a token with a perfect builder score but 80% scam probability will still score low overall.

Scam Probability (0–100%)

This is the most important number in the report. It represents the likelihood that this token is a rugpull, honeypot, or coordinated pump-and-dump scheme, based on on-chain analysis of the contract, liquidity, holder distribution, and social activity.

• 0–20%: Low risk. Contract is clean, liquidity is locked, holder distribution is healthy.
• 21–50%: Moderate risk. Some warning signs present. Could be a young project that hasn't established trust yet.
• 51–100%: High risk. Multiple scam indicators detected. Approach with extreme caution.

A scam probability above 60% is a hard pass for most risk profiles.

Builder Trust Score (0–100)

Builder trust measures the credibility of the team behind the token. ARC AI's Builder agent analyses GitHub activity (commit frequency, contributor count, code quality), wallet history of linked deployer addresses, LinkedIn and social presence, and cross-references against known scam deployer databases.

A high builder trust score does not guarantee a token will succeed — but a low one is a serious warning sign. Anonymous teams with no verifiable history and no GitHub activity score poorly here regardless of other signals.

Momentum Score (0–100)

Momentum measures whether a token is gaining or losing traction right now. It combines 24-hour and 7-day price action, volume trend, social mention velocity, new wallet acquisition rate, and DEX trading activity.

Momentum is intentionally short-term. A high momentum score on a new token means it is attracting attention — but that attention could be organic or manufactured. Always read it alongside scam probability.

Hype Score (0–100)

Hype is distinct from momentum. It measures coordinated social amplification — the kind that comes from paid promotions, airdrop farming, and shill campaigns rather than genuine community interest. A high hype score with low builder trust and high scam probability is the classic signature of a pump-and-dump.

High hype on a credible project is a neutral-to-positive signal. High hype on an unknown project is a red flag.

Smart Money Score (0–100)

Smart money tracks whether wallets with a proven track record of profitable trades are accumulating or distributing this token. ARC AI maintains a database of high-performing wallets across Solana — when these wallets buy, the smart money score rises. When they sell, it drops.

This is one of the most predictive signals in the report. Smart money does not guarantee gains, but it is a strong indicator of conviction from experienced market participants.

Reading signals together

No single score tells the full story. The most useful patterns to look for:

• High overall + low scam + high smart money: Strong long candidate. Experienced wallets are buying a credible project.
• High momentum + high hype + high scam: Classic pump-and-dump setup. Avoid.
• Low builder trust + low scam: Brand new project with no track record but no active red flags. Watch, do not buy yet.
• High builder trust + low momentum: Credible project that hasn't found product-market fit yet. Could be an early opportunity or a dead project.

Run a report on any token at arcai.io/analyze — no login required for your first analyses.