Six readings of the same market: how much it moves, how much of it moves together, and how much of it is moving at all. Change is measured against the same statistic one month earlier.
A mechanical screen, stated in full: rank the universe by dollar volume, keep the more liquid half above a capitalisation floor, take the five largest 30-day returns. It describes where the market went. It is not a list of anything we hold, or would buy.
Liquidation notional is the cleanest public record of leverage leaving the market. It is an effect, not a cause: the spikes land on the days that had already moved, which is what makes them worth watching — they mark where positioning was, not where it is going.
Every name that added or shed leverage over the window — so a quarter in which almost nothing was unwound leaves the right-hand column short, which is itself the reading.
Each bar is everything resting within that distance of the mid, so both wings rise outward and the pair is the depth curve exchanges draw. What matters is less the level than the shape — a book that thins at the front while the far bands hold is a market that has become expensive to move size in, without anyone's headline volume having fallen.
Correlation to equities is a regime, not a property. It compresses when the marginal buyer is the same in both markets and releases when it is not — which is why the level matters less than the direction of travel.
Both are dimensionless, so they share one axis and can be read against each other directly. They move in opposition: when everything is one trade the median pairwise correlation rises and cross-sectional dispersion collapses; when the market sorts itself by fundamentals, the opposite. The short horizon turns first — the gap between the 30-day and the 180-day line is the regime changing before the average has noticed.
Dispersion is the standard deviation of this distribution, one number a day. The distribution itself says what that number hides: whether a wide market is wide because everything moved apart, or because two or three names ran away from a tight pack.
Five classic risk-on ratios, each ranked against its own three-year history and averaged. It says nothing about crypto directly — it says what the marginal dollar was willing to own everywhere else.
Both series are rebased to the same starting point and share one axis, on a log scale where equal vertical distances are equal percentage moves. The familiar version of this chart gives each series its own axis, and two independent scales can be set until any two lines appear to agree. Note the orders of magnitude: money supply compounds in single digits a year, bitcoin in hundreds of percent.
A published index on a 0–100 scale: how many of the top fifty have outrun Bitcoin over the last quarter. It is a second opinion on the breadth this page computes for itself, measured against a different benchmark — breadth asks how many names are above their own average, this asks how many are beating the reference asset.
The same surface, sliced at three tenors. The downward tilt to the left is the market paying up for downside, and it flattens as the tenor extends.
Realised volatility at each horizon, against the 5th to 95th percentile of where that same horizon has sat over three years, with the implied curve on top. The band says whether the market is quiet outright or merely quiet for now; the gap between the two lines is what the option market charges over what it has lately delivered.
The 30-day implied level through time, against what the market then delivered over the following month. Implied above realised is the premium being paid; the stretches where realised runs above it are the ones that cost option sellers money.
Realised price is the average price at which the coins in a cohort last moved — their cost basis, read off the ledger rather than inferred. Long-term holders are coins that have not moved in 155 days. Spot below a cohort's line means that cohort is under water as a group. This is the one part of the page that exists for Bitcoin alone: no other asset here has a public ledger of what its holders paid.
Settlement demand, as close as the ledger gets to measuring it. It is not a user count — one person can hold many addresses and an exchange can serve thousands from one — so read the direction, not the level. It is the one series in this part that says something about the network being used rather than about what its coins cost.
Two estimates of the same supply, each made on its own, so they are drawn as two lines rather than one total. The share between them, dashed, is the steadier reading and the one worth following.
Every figure in this part is a property of the Bitcoin ledger, computed by a third party from public blocks. It describes where coins sit and what they cost, not what anyone should do about it — and nothing here is traded.
Quintiles, high to low i
Each group's return compounded window after window, which answers a different question from the average: not how much a characteristic paid, but when it paid and when it stopped. An average cannot show that a factor worked for eighteen months and then reversed. Descriptive, no spread taken between the ends, nothing here traded.
Every asset, placed by where it ranks on the characteristic today and by what it has returned over the last month — the raw material the quintile averages are made of. A tidy set of bars over a shapeless cloud is a characteristic being carried by two or three names, which no average can tell you.
At the start of each window the universe is ranked on the characteristic as it stood at that moment, split into five equal groups, and the average return of each group is measured over the window that followed. Windows do not overlap; the count on each card is how many of them the history supports, which is why the six-month view rests on far fewer observations than the daily one. This is a description of what the market has paid for, reconstructible by anyone from public prices — not a strategy, and nothing here is traded.