Every rule is written down here, so the numbers can be checked and argued with.
What is being measured
This site measures what happened to a token's price after it was mentioned:
an account's post contained a ticker or a contract address, and the token moved by some amount
over the following 1, 7 and 30 days.
The site does not judge whether a post was bullish or bearish, so these numbers
are not a "call accuracy" score. An account warning that a token looks dangerous,
followed by that token going to zero, shows up here as a large loss just the same.
Read the linked post and judge for yourself — every row links to the original.
How prices are taken
Starting price: the last trade at or before the post. A pump that happens
after the post is never used as the entry.
Ending price: the last trade at or before the post time plus the holding period.
Prices are hourly, from the public DexScreener, GeckoTerminal and CoinGecko endpoints.
A period that has not finished yet is not counted. Today's price is never passed off as
the price 30 days after a recent post.
How dead tokens are counted
When a token's pool is gone, or so thin that a position could not be sold, and no trade can be
found at the end of the period, it is counted as −100% — not as missing data.
This matters more than anything else on this page. Dropping dead tokens would delete every rug
from the record, and an account that only ever mentioned tokens that later died would come out
looking flawless. So every figure is given twice — including and excluding dead
tokens — and the distance between the two is itself information.
The reverse also holds: if a token still trades normally and a price merely happens to be missing
at one moment, that is recorded as missing data and not as a loss.
What is excluded
Ambiguous tickers. One ticker can refer to several tokens; when it is not
clear which was meant, the record is not counted. Mentions by contract address are exempt,
because an address is exact.
Repeats within 24 hours. Only the first counts. An account posting the same
token all day is one opinion, not twenty.
Non-crypto tickers, such as stock symbols, are filtered out upstream.
Each account page lists how many of its records were set aside, and why.
Ranking
Sorted by median change, worst first. A median cannot be dragged up by a
single lucky call the way an average can. Both are shown.
Accounts with fewer than 10 records are listed but not ranked,
and sit at the end.
"$100 per call ends as" is another way of writing the mean, not a separate
measure.
Known biases
The free market-data tiers only reach back so far; older posts cannot be priced.
A token matched by ticker alone can still be the wrong token of that name, even when the
match was judged reliable.
Which posts are collected depends on coverage, and is not every post an account has made,
so a sample can lean towards particular periods.
Prices are hourly, so moves within a single hour do not show up in the start or end price.
Corrections
If a record points at the wrong token, or an account's sample looks skewed, send the original
post link with an explanation and it will be rechecked and corrected on the page.
This site shows public price records only. It is not investment advice and makes no claim about anyone's intent.