Methodology
How Newsie translates raw on-chain risk into a single plain-English verdict — without asking you to trust a black box.
Our approach
Every safety score on newsie.tech is calculated by a deterministic, rules-based engine that weighs signals across five risk categories. We aggregate data from multiple independent providers (DeFiLlama, GoPlus Labs, CoinGecko, block explorers, and others) and combine them using a proprietary weighting model with a real-time temporal modifier. There is no black box at the scoring layer; every point has a specific, auditable reason.
The five risk categories
We check whether the source code is verified, whether it has been audited by a reputable firm, and whether it contains the high-risk functions commonly used in exploits and rug pulls.
We measure how much real money trusts the protocol, how that trust is trending, and how the protocol behaves under recent stress — drawing on multiple independent on-chain data sources.
We assess whether the team is publicly identified, whether the codebase shows ongoing development, and whether the project has a genuine, non-bought community presence.
We test whether the token is a honeypot, how concentrated ownership is, and whether the token has hidden mechanisms — taxes, blacklists, or transfer controls — that can trap user funds.
We separate yield from real economic activity (trading fees, interest) from yield that depends on token emissions, and flag advertised APYs that historically precede collapse.
How it stays trustworthy
Deterministic, not opinion
The same protocol on the same day always produces the same score. No black box, no LLM hallucination at the scoring layer.
Multi-source by design
Every score draws on multiple independent data providers. No single point of failure — if one source is compromised or rate-limited, the verdict degrades gracefully.
Time-aware
Recent on-chain stress is weighted alongside long-term reputation. A blue-chip protocol in active trouble is not scored as if nothing had changed.
Audit-ready
Every point in every score has a specific, traceable reason. We can show — to a regulator, to an investor, or to a protocol team that disputes a score — exactly which signals fired and why.
The specific weights, factor list, and temporal modifier formula are proprietary.
Protocols, investors, security researchers, and acquirers who need to verify the methodology in detail can request full disclosure under a mutual NDA at admin@newsie.tech. The public methodology page deliberately preserves the engine's defensibility.