Every Antelux score is a weighted composite of four pillars, converted to a 0–100 scale and mapped to a letter grade. The formula is mechanical (no analyst overrides, no pay-for-play), and every coin page shows the exact inputs behind its grade.
Emission schedule & inflation · Treasury runway · Real fee / revenue generation · Team & backer quality
Active address growth · Holder concentration · Exchange net flow · Staking / lock-up ratio
Trailing 30/90-day return vs. sector · Volume consistency · Drawdown recovery
Unlock cliffs · Insider concentration · Contract & audit status · Governance centralization
Each pillar is itself a composite of sub-factors. Here is what every one measures and why it earns its weight.
The largest weight, because durability compounds. Fundamentals ask: if the market closed for a year, would this network still matter when it reopened?
| Sub-factor | What it measures |
|---|---|
| Emission schedule & inflation | Forward supply growth vs. demand sinks; cliff-heavy schedules score worse than smooth, predictable ones. |
| Treasury runway | How long the project can fund development at current burn without selling tokens into weakness. |
| Real fee / revenue generation | Fees paid by actual users, and whether any of that value accrues to the token rather than only to insiders. |
| Team & backer quality | Shipping track record and whether backers are long-horizon funds or fast-flip syndicates. |
Price can be painted; blockspace usage is harder to fake. This pillar reads the chain directly for signs of genuine, distributed activity.
| Sub-factor | What it measures |
|---|---|
| Active address growth | Sustained growth in distinct active users, filtered against airdrop-farming spikes. |
| Holder concentration | Share of supply in the top wallets; broader distribution scores higher. |
| Exchange net flow | Multi-week flow trends weighed in context (see our Education article on reading net flow). |
| Staking / lock-up ratio | Supply voluntarily locked signals holder conviction and reduces liquid sell pressure. |
Deliberately third in weight: momentum tells you about timing, not quality. It stops a structurally strong coin in a downtrend from looking like a table-pounding buy today.
| Sub-factor | What it measures |
|---|---|
| Trailing 30/90-day return vs. sector | Performance relative to the coin's own category; beating a rising sector counts more than riding it. |
| Volume consistency | Steady participation across weeks beats a single headline-driven spike. |
| Drawdown recovery | How quickly the asset reclaims levels after market-wide corrections. |
The veto pillar. Risk is inverted: a high risk reading pulls the composite down, so a token cannot earn an A with a major unlock cliff ahead, no matter how strong everything else looks.
| Sub-factor | What it measures |
|---|---|
| Unlock cliffs | Scheduled supply hitting the market in the next 90 days, sized against circulating supply and typical volume. |
| Insider concentration | Team and early-investor wallets: how much they hold and how they've historically behaved at unlocks. |
| Contract & audit status | Audit coverage, exploit history, and admin-key powers that could freeze or mint. |
| Governance centralization | Whether a small group can unilaterally change protocol parameters or treasury spending. |
Scores are percentile ranks against the rest of the rated universe, not marks against fixed thresholds. Say a coin ranks in the 92nd percentile on fundamentals, 40th on on-chain health, 62nd on momentum, and 91st on risk. Those blend by weight:
That blend is then itself ranked against every other coin's blend, and the resulting percentile is the published composite. The second pass matters: averaging four percentiles pulls hard toward the middle, so without it the top and bottom grades would be mathematically unreachable and almost everything would land in one band.
So a composite of 97 does not mean "97 points out of 100". It means the coin scores higher than 97% of the universe on the weighted blend.
Because the composite is a percentile, each band maps to a share of the universe. Grades are relative: an A means best-in-class among what is currently rated, not that an asset has cleared some absolute bar.
| Grade | Composite | Share | What it means |
|---|---|---|---|
| A | 95–100 | Top 5% | Best-in-class across the weighted pillars right now. |
| A− | 85–94 | Next 10% | Strong nearly everywhere, one pillar short of the top. |
| B+ | 70–84 | Next 15% | Comfortably above average, with a visible tradeoff. |
| B | 50–69 | Next 20% | Upper half. Solid on some pillars, ordinary on others. |
| C | 25–49 | Next 25% | Below the median. Mixed signals, real diligence needed. |
| D | 10–24 | Next 15% | Weak on most pillars relative to peers. |
| F | <10 | Bottom 10% | Bottom of the universe on the weighted blend. |
Relative. Pillar scores are percentile ranks within the rated universe, and the composite is a percentile too, so an A means top 5% of what we currently rate rather than clearing a fixed bar. This is a deliberate choice: thresholds fixed in the abstract sound rigorous but in practice real market data rarely reaches them, which pushes almost every coin into a single band and makes the grade useless for comparing. The tradeoff is honest and worth stating: in a broadly weak market, the best of a weak field still earns an A. Read the grade as "how does this compare to the alternatives", not "is this objectively excellent".
Market inputs refresh on our servers every 15 minutes, around the clock. Because grades blend slower-moving structural signals with live market data, expect small day-to-day drift and occasional band changes after genuinely new information, not constant flickering.
Usually something did happen, just quietly. A widening 24h swing lifts measured risk, a fading week erodes momentum, or volume dried up relative to market cap. Because scores are relative, a coin can also move because its peers moved while it stood still. The “How this grade was built” panel on every coin page shows exactly which input drove it. Grades also carry a buffer: a composite has to clear a band boundary by a margin before the published grade moves, so a coin sitting on a threshold does not flip back and forth.
No. Grades are computed mechanically from the pillar inputs. There is no analyst override, no partnership program, and no way to sponsor a score. The same formula runs on every coin in the universe.
Because warnings get ignored. Folding risk into the composite at 20% means a scheduled unlock cliff mathematically caps the grade: you can't get an A- with a red-flag pillar. We wrote up the full reasoning in the blog post “Why We Weight Risk at 20% (and Invert It)”.
Grades are computed from live public market data (via CoinGecko). Fundamentals blends market-cap rank with real tokenomics: the share of max supply already circulating, so coins facing heavy future emission score lower. Momentum blends trailing 7-day and 30-day returns, so one hot week can't fake a trend. Risk combines 24-hour volatility with drawdown depth, adding a penalty beyond 70% below all-time high. On-chain health currently uses volume relative to market cap; deeper on-chain inputs (active addresses, holder concentration) are next on the roadmap, and every coin page shows exactly which inputs drove its grade.
The top 100 by market cap, with stablecoins excluded (a pegged asset can't be meaningfully graded on momentum or fundamentals). The universe re-forms automatically as coins enter and leave the top 100.
See the methodology applied to a live coin, input by input.
Open a live scorecard