What Our Grades Measure Today, and What They Do Not

Why publish this at all

Rating products tend to describe their methodology in aspirational terms: the full list of things the model considers in principle, with no indication of which are actually wired up. That is how you end up trusting a number that turns out to be three price columns in a trenchcoat. We would rather be boring and specific.

How the numbers are scaled

Every pillar score and the composite itself are percentile ranks within the rated universe, not marks out of a fixed total. A 90 means "ahead of 90% of rated coins on this measure". We started with absolute thresholds and abandoned them, because real market data almost never reached the upper ones: two thirds of the universe landed on the same grade and the top bands were mathematically unreachable. A scale where the top bands never appear is not a scale.

The tradeoff is that grades are relative. In a weak market the least-weak coin still tops the table. We would rather state that plainly than imply an absolute standard the data cannot support.

What is measured right now

Fundamentals (35%) blends two things: position in the market-cap table, and supply dilution, meaning the share of maximum supply already circulating. A coin with most of its supply already issued has less future emission overhanging holders than one with a fraction released. Coins with no supply cap are scored neutrally with a slight caution, since uncapped emission is a real but not automatically disqualifying property.

On-chain health (25%) reads real chain and protocol data from DefiLlama. Its heaviest sub-weight is fee revenue: fees actually paid by users over 24 hours, annualised and measured against market cap. That is the closest thing crypto has to an earnings yield, and it is the hardest number on this page to fake, because paying fees costs money. Second is value secured, meaning total value locked on the chain or in the protocol, again against market cap. Third is 24-hour turnover, the market-structure proxy that used to be the whole pillar.

Roughly half the universe has fee or TVL data; the other half genuinely has none. Bitcoin has no protocol earning fees, and neither does a memecoin. Rather than score those zero, each sub-measure is ranked only against the coins that report it and the weights renormalise, so an asset is judged on the evidence that exists for it. The on-chain panel on every coin page names which measures were used, and says so explicitly when the answer is turnover alone.

Momentum (20%) blends trailing 7-day and 30-day returns, weighted toward the week but requiring the month to agree before the score gets strong. One hot week against a bad month nets out to a middling reading, which is the intent.

Risk (20%, inverted) combines 24-hour price volatility with drawdown depth below the all-time high.

The one adjustment that is not arithmetic

Momentum and turnover together account for a large share of the weight, and a thinly traded token in a hot month can rank near the top of both. So an A carries one additional condition: fundamentals must be at least median. A coin in the bottom half of the universe on market standing and tokenomics is capped at B regardless of how high its composite runs. The composite is left at its computed value rather than quietly reduced, which means where the cap bites you can see it, as a score in the nineties sitting beside a B.

What is not measured yet

Active address growth, holder concentration, and exchange net flow are genuine on-chain measures and none are in the score today. Fee revenue and value secured are, which is the substantive change since this article was first published.

On fundamentals, treasury runway and whether value accrues to the token rather than a foundation are absent. On risk, unlock schedules, insider concentration, and audit status are absent. Each requires a data source beyond what is currently wired up, and each will be added as its own pillar upgrade, announced when it lands.

What this means for how much to trust a grade

Reasonably, for what it is. The inputs that are live are real, refreshed every fifteen minutes, and applied identically to every coin in the universe with no analyst override and no way for a project to influence its own score. The grade is a fast, consistent, mechanical read on market structure and recent behaviour.

It is not yet a fundamental analysis of a protocol business. A coin can score well here and still have a treasury running dry or an unlock cliff next month, because we cannot currently see either. Use the grade to narrow a hundred coins down to a handful worth real diligence, then do the real diligence.

How you will know when this changes

Each coin page lists the exact input behind every pillar score. When a new data source goes live, that panel changes to name it, and this article gets updated. Where the panel says a coin is scored on turnover alone, that is because no protocol or chain reports fees or value secured for it, and turnover is genuinely all the on-chain pillar has to go on.