How we count

NARRATIVE RADAR groups new Solana memecoins by what their names are about and shows how each theme lives and dies.

Narratives

Every new token is sorted by its name and ticker against a list of keywords: cats, dogs, frogs, AI, politics, celebrities, anime, food, space and more. A keyword in the name counts more than one in the description. A token that matches nothing goes to “Unsorted”. The list is public in the project and is a rough guide, not a verdict: a token called “AI Cat” is both, and goes to one.

Stories

A story is a word that is not in any narrative list but suddenly appears in the names of many new tokens, for instance a person or event in the news. Stories are short: most are over within a day.

Birth, peak, death

• Born: first activity less than 6 hours ago.
• Rising: the 3-hour average is at least 20% above where it was 3 hours earlier.
• Peak: now within 25% of the best 3-hour average, and that best was in the last 6 hours.
• Cooling off: below half of the peak.
• Gone quiet: after the peak it fell below 10% of the peak and stayed there for 24 hours (12 for stories).

The curve uses traded volume of the narrative’s tokens when we have it, and the number of new tokens per hour otherwise. Both are shown on the narrative page.

What the numbers cover

• Traded volume counts tokens that reached about $10,000 of trading in 24 hours. Tokens that never did are “stillborn”: we count them, but not their trades.
• New tokens seen is a sample: every few minutes we look at the newest launches, and only when that look succeeded is the hour counted as sampled. An hour nobody looked at is shown hatched and is unknown, not zero. Shares are computed only from sampled hours as (this narrative) / (all tokens seen), so they stay comparable when the number of looks changes; raw counts do not.
• Unsorted holds tokens whose names match no narrative. It is shown on its own colour scale so it does not drown the rest. The same name and ticker launched twice in one hour is counted once.
• Collecting: a narrative gets a status only after at least 24 unbroken hours of data from enough tokens. Until then the page says so instead of guessing. Tokens we only started watching recently have no earlier hours; those hours are unknown, not quiet.
• Hourly volume comes from the public pool history of each token where we could fetch it, and from our own readings afterwards; the current hour is an estimate. The "Traded, 24 h" figure is the sum of the tokens' own 24-hour volume.
• Stillborn is judged only for tokens older than 3 hours, so the newest hours understate it.
• A token's name and ticker are the only input; descriptions are not available from the sources. Names in Latin, Cyrillic and Chinese/Japanese/Korean scripts are matched; other scripts fall into Unsorted.
• Tokens found later through search or trending are added to the volume figures but not to the new-token counts.
• Wash trading and fake volume are not filtered out.

What this is not

It is not financial advice, a ranking of what to look at, or a forecast. Past curves say nothing about the future. Data comes from public APIs of DexScreener and GeckoTerminal; links to tokens open DexScreener so you can check the facts yourself.

Sister project

LAST CANDLE is the memecoin graveyard. When a narrative goes quiet we link there.