Crypto
The structure board
Every one of the largest 200 freely floating crypto assets, placed on two readings taken from price and volume alone: turbulence — how wide its outcomes run — and turnover — whether there is a market underneath the price. Refreshed every four hours. This page describes structure and nothing else: never a recommendation, never a forecast, and never a claim about direction.
What the market did. Over the last 90 days the middle of this panel moved -31.7%, and the typical asset sits -48.4% below its own 180-day high. Bitcoin itself is -0.0% over 30 days and -16.5% over 90. These are measurements of what happened, not a view on what happens next.
This reading is behind. The newest panel is 462 hours old and the board refreshes every four. The capture pipeline has not delivered a newer reading, and we would rather say so than let the timestamp sit here unexplained.
| Coin | Sector | Market cap | Turbulence | Band | Turnover | Depth | Zone | 30d | 90d | From 180d high |
|---|
How to read it
Up the page is wider. Turbulence is annualised volatility over the last 30 days of daily closes, banded at 60%, 90% and 140%. Crypto runs hot by nature, so Elevated is not treated as turbulent here — 60 to 90 percent annualised is an ordinary crypto tape, and counting it as turbulent would make the reading meaningless.
Right on the page is thinner. Turnover is 30-day median traded volume divided by market capitalisation: how much market there is underneath the price. It is cut at frozen quartiles — 0.0013, 0.0166 and 0.0724 — fitted once on a survivorship-included panel and deliberately never re-fitted per run. A threshold that moves with the population stops being a measurement.
The two together make four zones. Steady is calm with a real market underneath. Wide is a large range with real depth. Quiet is calm and lightly traded. High-risk is a large range and lightly traded. The names describe structure and nothing else — “Quiet” says a price is calm and thinly traded, not that it is safe, and not that it is dangerous.
The two readings are measured independent, which is what makes this a plane rather than one number drawn twice.
Price sits beside structure, not inside it. The return columns are what the price did over the last 30 and 90 days, and the drawdown is against the highest close in the 180 days the capture holds — not an all-time high. None of it feeds the band, the turnover reading or the zone. They are placed together because a reader needs both, and kept separate because mixing an outcome into a structural measurement is how a measurement stops being one.
What is left out, and why
Both exclusions happen at the universe boundary, before the ranking is taken, and the ranking keeps paging further down the market-cap list until it holds a full 500. This is not a haircut: the board is 500 assets, not 500 minus the exclusions. What an excluded token costs you is its place, not the slot — the next asset down takes it.
Pegged tokens. A token engineered to hold a dollar or euro peg has no variance state to read: its volatility is near zero by construction, not because the market is calm. Leaving them in would park a tenth of the board permanently in the calmest corner. They are detected empirically — a series sitting within 2% of its own median 90% of the time — rather than by a name list, so obscure and de-pegged-then-repegged tokens are caught too.
Wrapped and staked derivatives. stETH, wBTC and their kind are claims on another asset, and the trading happens on the underlying. Their turnover is thin by design rather than by neglect, which is a different thing entirely and would read as the same.
Assets with too little history. Under 45 days of daily closes, no band is issued. The panel abstains rather than guessing, and says so in the status column.
Provenance
Measured from daily price and volume history, captured every four hours and committed to a public record. This page reads the newest panel. The bulk downloads — JSON and CSV — run about a day behind it, deliberately, and each file states its own lag in a delayHours field rather than leaving you to work it out. Every row, every measurement and both sets of cut points are in there unaltered, so any reading on this page can be reproduced from the file rather than taken on trust. What the downloads leave out is the vendor's own price, market cap, rank and 24-hour change: those are licensed market data we display to you here, not ours to hand on in bulk. The file is the analysis, and it joins to any price feed you already have on symbol and date.