Methodology

Overview

Every token is rated on two independent axes rather than a single verdict, because merit and risk are not opposites. The Goldcoin score (0–100) measures merit: fair distribution, sound supply, decentralization, security, real development and real usage. The Shitcoin score (0–100) measures red flags: insider control, admin powers, unlock overhang, missing utility, hype dependence and track record. A token can score high on both — that is the point.

How scores work

Each axis is a weighted sum of category subtotals (seven on merit, six on red flags), and each category is a weighted sum of indicators scored 0–100. Indicators are tagged measured (fetched from live data: market figures, node counts, developer activity, on-chain authorities) or assessed (researched by hand, with a source link on every row). The two axes are computed independently, then the token is placed in a quadrant: Goldcoin (high merit, low risk), Goldturd (high both), Midcoin (low both) or Shitcoin (low merit, high risk).

Weights

Category weights are published per methodology version and only change with a version bump recorded in the changelog. Each axis sums to 100.

Data sources

Market data comes from CoinGecko, fee and TVL figures from DefiLlama, developer activity from the GitHub commits API, Bitcoin mining and node data from mempool.space and Bitnodes, and token authorities from public chain RPCs. Curated facts — launch method, insider allocations, exploits, governance — are researched per token and every row links to its source. Every external answer is cached in our database; if a source rate-limits or goes down, the last good answer is used instead of a guess.

Our worldview

Fair distribution, credible supply, removed admin keys and demand that survives a bear market. Everything else is marketing. We penalize tokens that exist only to be sold, teams that can mint or freeze at will, and valuations that depend on a story rather than throughput. We reward boring things: years in production, permissionless participation, and fees paid by people who actually use the thing.

Data confidence

Every token page shows a data confidence rating of high, medium or low. This measures how much of the methodology actually had data when the score was computed — it says nothing about the project itself. When a data source is unavailable or a fact can't be verified, it is shown as unknown rather than as zero. The current score still uses only known evidence, while the red-flag range shows how high the score could be if every unknown red-flag indicator proved severe. On the quadrant map, a dashed vertical line shows that possible range. If at most 10% of the scoring weight was missing, confidence is high; 10–30% missing is medium; anything more is low. A low-confidence score is still honest — it just rests on fewer verified facts, and it will improve as the missing data fills in on later runs.

Changelog

1. Community votes — coin pages ask "Is this verdict fair?" with 👍 / 👎 counts. Votes are community opinion only and never feed into any score; to change a score, dispute a fact with evidence.
2. v0.8 — added "Full bear markets survived" to the Security category, next to years in production. A coin gets credit for a bear market only if it was live before the peak and still running after the bottom (2014–15, 2018, 2022): 0 → 0, each one +40, capped at 100. Being old is not the same as having survived a crash.
3. v0.7 — unknown red-flag indicators no longer look like proven zero risk. Token pages show them explicitly, and the map shows a range from the score supported by known evidence to the worst-case score if all unknowns were severe. Holder concentration now means the top 10 beneficial holders after excluding identified exchanges, mining pools, contracts and burn addresses. Kaspa's estimate was corrected to about 10% after removing known exchange wallets; Litecoin remains unknown because no comparable attributed dataset is available.
4. v0.6 — added a "Security budget" category (10% of merit): security spend (rewards ÷ market cap per year: ≥ 1.5% 100, 0.75–1.5% 70, 0.3–0.75% 40, under 0.3% 10), fee share of rewards (≥ 25% 100, 10–25% 70, 3–10% 40, under 3% 15), security trend (hashrate or staked share: growing 100, flat 50, falling 0) and real reward to securers (PoW miner margin or PoS yield minus inflation: positive 100, break-even 50, negative or none 0). Tokens on a host chain get half the host's score. Fair launch, decentralization, supply, security, tech and real usage were trimmed to make room.
5. v0.5 — added a "Programmability" indicator under Tech & development: full smart contracts (EVM or a native VM) 100, limited scripting or tokens only 50, none 0. Scored on the base layer's own capability; tokens on a host chain inherit the host.
6. v0.4 — added a "Usability as money" category (10% of merit): base-layer confirmation speed, with a top tier for blockDAGs that merge parallel blocks instead of orphaning them, and the typical transaction fee. Only L1 counts; L2 claims do not. Other merit categories were trimmed to make room.
7. v0.3 — added a "Consensus mechanism" indicator under Decentralization: proof of work 100, permissionless proof of stake 65, delegated / small elected set 30, permissioned 0.
8. v0.2 — first live methodology. All 50 tokens re-scored from measured and sourced data; per-source caching added so rate limits never fabricate numbers; score history recorded on every run.
9. v0.1 — initial draft methodology with demo placeholder scores.