How the score works

The score is a weighted vote. A set of independent signals each cast a vote for or against, the votes get weighted by the token's profile, and the result is normalised to 0–100.

That's the whole idea.

What votes

Each signal is computed independently, on its own evidence, and then normalised onto a common scale so they can be combined. They fall into five families:

Family	Asks
Momentum	Is the token stretched — has it run too far, or been sold too hard?
Trend	Which way is it actually going, on the week and on the day?
Structure	Does the chart's shape mean anything — a base, a breakout setup, a topping pattern?
Volume	Is participation growing or draining relative to its own recent history?
Flow	Are real transactions leaning to the buy side or the sell side?

Which of these carries the most weight depends on the token's age profile — see Token profiles. The specific indicators, their thresholds and the weight table are the part that makes this analysis different from a stack of defaults, and they aren't published.

A signal that doesn't have enough history is left out entirely — not defaulted to zero. Dropping it is honest; including a neutral vote it never earned isn't.

Reading the result

Score	Verdict
70–100	🟢 Strong buy signal
55–69	🟡 Bullish, accumulation zone
45–54	⚪ Neutral, wait and see
30–44	🟠 Bearish, wait for better entry
0–29	🔴 High risk, avoid for now

Two things worth internalising

A token with no signals at all scores around 50, not 0. The normalisation maps "everything neutral" to the midpoint. A 50 means no opinion, not bad — and a token with too little history for most signals lands near the middle by construction.

Thin data can't produce an extreme score. The scale a token is measured against is fixed by its profile, not by how many signals happened to fire. So a token where only a handful could be computed gets pulled toward the middle and can't reach 100. That's intentional: thin evidence should produce a hedged verdict, not a confident one.

What the score is not

It's a composite, not a prediction, and it's deliberately hard to move with any single indicator. Any one indicator you follow will disagree with it regularly — that's the point of a composite.

The score is a technical verdict. It does not know about the token's liquidity, its unlock schedule, or the fact that 40% of supply sits in six wallets funded by the same address. That's what on-chain sentiment and holder analysis are for — and it's why an 80 on a token with critical liquidity is not an 80.

One exception: our own token

$DON is analysed by the same engine as everything else, and then two declared adjustments are applied on top of the finished result:

• the score is lifted by 8 points and never printed below 50, so the verdict on our own token never reads worse than Neutral, wait and see;

• the short-term (1d) plan is withheld — only the long-term zones are quoted. A tight scalp range on the token whose own holders are reading the card is an instruction to sell it, and we would rather not publish one.

Everything else — every indicator, every category, the long-term entry and target zones — is computed exactly as it is for any other token.

Wherever those adjustments apply, the card says so: the Telegram card, the Web Analyzer, the embeddable widget and the scan image all carry an our token · house view badge next to the score, and the API returns the same thing as data (analysis.native). The number is ours to lean on; hiding that we lean on it is not.