Accountability

Methodology

How the risk score is calculated, what it does not mean, and why we decline to rank exchanges.

Risk ratings are statements of opinion. Our risk ratings are expressions of opinion, derived from the published methodology and based on the specific facts listed and linked in each rating. They are not statements of fact, not financial advice, and not an allegation that any person or project has engaged in criminal or fraudulent conduct. Ratings may rely on limited or unverified information, reflect data at a point in time, and can change without notice.

What the number means

The score runs from 0 to 100, where 0 means the fewest risk signals detected and 100 the most. It is a risk score. It is never a safety score, and we will not describe any token as safe. A low score means that at the moment we last checked, few of our automated checks fired. It says nothing about whether a project will succeed, whether its team is honest, or whether you will lose money.

We publish the number alongside every flag that produced it, each linked to the primary on-chain evidence, so that you can disagree with our weighting and reach your own conclusion from the same facts.

The five sub-scores

Sub-score	Weight	What it looks at
Contract authority & code risk	30	Mint authority, freeze authority, ownership renounced, upgradeable proxy, hidden mint, blacklist functions, pausable transfers, modifiable tax, honeypot simulation, unverified source, similarity to known scam contracts.
Liquidity risk	25	Whether the liquidity pool is locked or burned — and when the lock expires, not merely that one exists. Single-provider pools, deployer-owned liquidity, absolute depth, and liquidity as a fraction of stated market cap.
Holder & distribution risk	20	Top-10 concentration, deployer holdings, supply sniped in the first blocks, and wallet clusters funded from a common source.
Trading & market-integrity risk	15	Wash-trading signatures, buy/sell tax asymmetry, and the ratio of unique buyers to transactions.
Social & off-chain trust	10	A modifier only. A doxxed team and a real audit can lower a score slightly. They can never, on their own, produce a low-risk verdict.

The critical-flag override

A weighted average can dilute a fatal flaw. If any one of the following is positively confirmed, the score is floored at 85 regardless of everything else:

A honeypot, or a confirmed inability to sell.
Active mint authority combined with deployer concentration above 50%.
Unlocked liquidity, a single liquidity provider, and deployer-held LP tokens together.
Unverified source code on a token trading with real volume.
A modifiable sell tax currently above 50%.

Each of these must be confirmed present. A check we could not run is never treated as though it failed.

Unknown data, and why the score sometimes disappears

We never impute a missing value as its worst case. A signal we could not check is marked unknown, excluded from the average, and it lowers the confidence figure instead.

If confidence falls below 60% — meaning we could not gather sub-scores covering at least 60 of the 100 available weight — we show no number at all, only “Unverified”. The absence of a score is not an endorsement. It is an admission that we do not know.

The bands

Low risk signalsIn our opinion, no major automated red flags at last check. This is not a guarantee of safety.

Some caution advisedA few risk indicators are present. Review the flag list below before engaging.

Elevated riskMultiple red flags detected. In our opinion this token exhibits characteristics commonly associated with high-risk launches.

High riskCritical red flags detected. In our opinion there is a high risk of loss. Exercise extreme caution.

UnverifiedWe could not verify enough on-chain data to score this token. The absence of a score is not an endorsement.

Language rules we hold ourselves to

These are not stylistic preferences. Under Australian defamation law, the defence available to a published rating is honest opinion: the statement must be opinion rather than fact, on a matter of public interest, and based on proper material that is stated or linked on the page. Small token projects can sue, and so can the people behind them.

We say in our opinion. We never state as fact that a token or a person is a scam, a fraud, or a thief.
Every flag that fires is shown, with a link to the primary evidence, on the same screen as the score.
We rate tokens and projects, not named individuals.
Every High risk rating is reviewed and signed off by a person before it publishes. An algorithm cannot hold an opinion honestly.
Every rating carries a right of reply. If a project responds, we publish the response in full.
Our pump-and-dump monitor reports patterns as observations. It never asserts that anyone manipulated a market — that is a serious offence, and it is a matter for a court.
There are no comment sections on rating pages. In Australia the operator of a page publishes the comments it hosts.

Why we do not rank exchanges

Ranking exchanges is where crypto comparison sites make their money, and it is the one thing we will not do. Partly this is editorial: the moment a ranking exists, the incentive to sell a position in it exists too, and every competitor who has crossed that line now discloses that placement is paid.

Partly it is legal. A ranked recommendation about a financial product is financial product advice under section 766B of the Corporations Act. From 9 April 2027, when the Corporations Amendment (Digital Assets Framework) Act 2026 commences, a digital asset platform becomes a financial product in its own right. And under ASIC’s INFO 269, a commissioned affiliate link into such a platform is likely to be dealing by arranging — a licensable service.

So we publish facts: what an exchange lists, where it is incorporated, which currencies it quotes, and whether a regulator’s register names it. You do the ranking. See the Independence Ledger.