How the labs count.
One recording, one set of words, and the weaknesses on record. Every report points here for what its terms mean.
How the market is recorded.
- Source
- The hub reads launches, trades and pool changes of the Solana launch market around the clock and writes them to disk as they arrive.
- Days
- All days are UTC days. The recording began on 25 Oct 2026, mid-day; that day and the current day are partial, drawn open in charts and left out of averages.
- Gaps
- Each outage of the recorder is logged with its start and end. Every gap is listed under Corrections and counted in the coverage report.
- Counting
- Counts run read-only on the production hub, by one script in the repository. Only sums and distributions leave the hub. Each query is saved with the numbers it made and printed under the report.
The words every report uses.
- Coin
- A token launched on the market the hub records. One row in the memory.
- Dev
- The wallet that created a coin. One person with many wallets counts as many devs.
- Priced
- A coin with at least one trade the hub turned into a market cap.
- Highest market cap
- The largest market cap, in US dollars, the hub priced for a coin.
- Rating
- The risk class the hub gives a coin: Low, Moderate, Elevated, High or Very high. It names a risk, it is no forecast.
- Lower after one hour
- The market cap one hour after the rating is at most 30% of the cap at the rating, so 70% or more lower.
- Outcome
- What the memory records a coin did afterwards: no pull recorded, ran, went quiet, liquidity pulled, still open, or in a recording gap.
- Decided
- A coin whose outcome is one of no pull recorded, ran, went quiet, liquidity pulled. Shares of outcomes count decided coins only.
- Liquidity pulled
- Liquidity removed from the coin's pool. The detection needs a sale by the dev.
- Full day
- A UTC day the recording ran from start to end.
- Base rate
- The share for all cases together. A class or group says something only where it differs from it.
- 95% interval
- The range of shares the count is consistent with, by the Wilson score method. It is wide when cases are few. A share from fewer than 100 cases never goes into a title.
What never leaves the hub.
- Addresses
- No wallet address and no mint appears in a report, a chart or a data file. A test checks every published file for them.
- Third-party data
- Raw answers of outside data providers never pass. What leaves the hub is derived by CON.
- Wallet reports
- Reports on wallets show sums only and stay unpublished until the legal review of wallet data.
Known weaknesses.
- Before 5 Nov
- Pull outcomes of coins launched before the Meteora fix of 4 Nov were measured differently. Reports on outcomes count from 5 Nov on.
- Short window
- The recording holds 20 full UTC days. Weekday and hour patterns rest on two or three occurrences per weekday.
- Many wallets
- A dev who uses a new wallet for every coin counts as many devs, so concentrations among devs are likely understated.
- Between reads
- The hub sees the market cap it priced. A spike between two reads can be missed.
- Gaps
- Whatever happened inside a gap of the recording is missing from every count.
Editions and corrections.
- Editions
- A report is frozen in its edition and reads its own saved snapshot. New data makes a new edition, with a line in the report's history. A refresh never changes a published report.
- Live data
- The market data sheet and the unit pages show the latest count and change with every refresh.
- Corrections
- Every change to data and reports is listed with its date. An error found by a reader is fixed in a new edition.
- Data files
- Every table of a report can be downloaded as CSV, with the same rows as on the page.