Methodology
Every number on CryptoRanks comes from a measurement we can explain. This page consolidates how each ranking is built — and where its limits are.
New pools and the health score
Our collectors watch pool-creation events on the DEXes we cover (Uniswap V2/V3, Aerodrome, Raydium and others) across Ethereum, Base, Arbitrum, Optimism, Polygon, BNB Chain and Solana. For each pool we read its liquidity directly on-chain and keep re-reading it as the pool ages, building a per-pool liquidity history.
The health score (0-100) blends current liquidity, the trend versus the pool's own peak, how long the pool has survived, and light on-chain risk checks such as unverified contracts or mint and freeze authority on Solana tokens. Status moves from nascent to watching, healthy, suspicious or dead as the evidence accumulates.
Whale impact
A swap is ranked by impact, not by absolute size: impact is the swap value divided by the pool's liquidity at that moment. A $10k trade in a $20k pool moves the price far more than a $100k trade in a $5M pool. Severity tiers (notable, strong, critical) are simple thresholds on that ratio, and per-token buy/sell pressure is the share of whale volume on each side over 24 hours.
Gas tracking and levels
Gas is sampled directly from each chain's public RPC nodes roughly every 5 minutes. The cheap / normal / expensive level compares the current reading with that chain's own 24-hour range, the trend comes from the most recent samples, and the cheapest hour is the hour of day with the lowest average in the measured history. Gwei is each chain's native unit and is not comparable across chains.
Confirmed rugs and outcomes
As pools age we classify each one as survived, died or rugged. A rug is only confirmed when we find the on-chain Burn event proving liquidity was actually pulled — or when liquidity collapsed at least 90% from a peak that was once real. Pools that simply faded out count as died, never as rugs, so the headline rug rate is not inflated.
We also check whether a pool's LP tokens are burned or locked. Locked liquidity cannot be pulled, so LP-locked pools are flagged and excluded from at-risk lists. Deployer reputation aggregates these outcomes per creator wallet: rug rate = rugged / settled launches, with young pools excluded until they are old enough to judge.
Pool outcomes & deployer reputation
A rug verdict comes in two flavours, and we keep them separate. A proven rug is backed by the on-chain Burn event that removed the liquidity — we store the transaction hash. An inferred rug is a collapse of at least 90% from a peak that was once real, meaning the pool held at least $1,000 of measured liquidity before it fell. A pool that never held real money can never be labelled a rug.
A pool that simply fades counts as died, never as rugged. Most failed launches are stillborn: there was never enough money in them for anyone to steal, so calling them rugs would be an accusation without a victim — and would devalue the label everywhere it appears. Died means the pool aged past its first hour without real liquidity, or deflated without a deep collapse.
Because died and rugged are different verdicts, we also publish a failure rate: (rugged + died) / settled launches — the share of a creator's settled pools that ended worthless, whatever the mechanism. Settled excludes pools younger than one hour and pools whose liquidity we could not measure, so the denominator only contains launches with a real, countable outcome.
One nuance: a pool that died after peaking at $10,000 or more is not a stillborn — real money entered and was lost. The public rug rate still does not count it, but the creator's risk verdict weighs these heavy deaths internally, so a wallet whose launches keep absorbing money and quietly dying does not read as harmless.
The creator verdict is a ladder of rules, first match wins, each demanding a minimum track record so a thin history never raises a false alarm: Rugger (3+ settled launches, rug rate 30% or more), Junk factory (8+ launches, 5+ settled, zero survivors), Spam bot (6+ launches at 4 or more per hour, or half of them on the same pair), Risky (3+ settled, rug rate 10% or more). Wallets below every bar show as a mixed record or simply too new to judge.
Limits, stated plainly: the LP-lock check only recognises V2 LP tokens sent to burn addresses. Liquidity held in third-party locker contracts is not detected as locked, and V3 positions are NFTs we cannot assess — both show as unknown rather than safe. The same rule applies to outcomes: a reading we cannot take becomes unknown and drops out of every rate, never a verdict.
Honest limits
We measure seven chains and the major DEX factories — not every venue in existence. Public RPC nodes can lag or rate-limit, so a reading can be minutes old, and a pool can rug between two samples. Outcome labels are empirical heuristics, not legal findings about any project or wallet.
General market data on the Markets and Exchanges pages comes from the Coinpaprika aggregator and inherits its latency and coverage. When data is missing we show an empty state rather than an estimate: nothing on this site is interpolated or invented.