The Accumulation Trend score boils down what each holder bracket has recently done to its stack, leaving one figure. That means the question of whether the network is adding or shedding can be settled without following half a dozen lines simultaneously.

At one extreme every bracket is buying with conviction; at the other every bracket is letting go. The centre describes stacks that have barely shifted, which is a different animal from a market that cannot make up its mind.

What it actually measures

Each cohort’s recent change is scored and the scores are blended, so the reading reflects breadth as well as direction. A high score means the buying is spread across sizes rather than concentrated in one band.

The series is deliberately calmed down, since untreated daily movements in holdings jump between states from one session to the next. Calming it is what stops the label being decided by chance. The bill for that is delay, and a real change of direction shows up here once it is already under way rather than at the moment it begins, which is what any measure built to name conditions costs.

The divergences from price are the most useful readings on the chart

The score can sit high while price falls, and that combination is far more informative than agreement. Accumulation into weakness describes a market where balances are growing without price having to reward anybody for it.

The reverse case is equally worth catching. Distribution during an advance says supply is being handed over into strength, which is the configuration that precedes trouble more often than a falling price does. Reading the score against price rather than on its own is where the value sits.

Cohort Flow Matrix: the relative scale is a decision, not a convenience

Each cell reports what one cohort did over the days it covers. Reading down a column says whether the whole size range moved together; reading across a row says whether one cohort has been consistent.

One substantial owner shifts more coin in an afternoon than the entire population of small accounts manages over weeks. Left in raw quantities that arithmetic leaves the top row as the only thing with any tint at all, and everything beneath it uniformly empty.

The divergence the view exists to show is therefore invisible on the scale that looks most honest. Ranking each cohort against its own behaviour is what makes a small cohort’s decisive week legible next to a large cohort’s ordinary one, and choosing that as the default is a judgement about what the chart is for.

What it does not tell you

Balance changes are not purchases. Coins moving between an entity’s own addresses register as a cohort gaining or losing, and no on-chain measure separates that from a decision to buy.

The blend also hides who is doing the work. A middling score can come from a genuinely balanced market or from large cohorts buying while small ones sell, and the headline number cannot tell you which.

Breadth is also not size. A score built from every cohort moving slightly is treated much like one built from a few cohorts moving decisively, so the reading answers how widely rather than how much. Pair it with a measure of size before concluding anything about the scale of what is happening.

How to read it

Strong Accumulation. Broad, decisive buying across the cohorts.

Accumulating. More cohorts adding than shedding.

Neutral. Balances broadly unchanged across the network.

Distributing. More cohorts shedding than adding.

Strong Distribution. Broad, decisive selling across the cohorts.

The Supply dashboard keeps Accumulation Trend alongside Wallet Distribution, Apparent Demand and Retail Holdings.

Common questions

What sits at each end of the range?

At one extreme every bracket is buying with conviction, at the other every bracket is letting go. The centre describes stacks that have barely shifted.

Why is the line calmed down?

Because untreated daily movements in holdings jump between states from one session to the next. Calming the series is what stops the label being decided by chance.

Can it read strongly while the market drops?

Yes, and those are the readings worth stopping for. Buying into a falling market is an entirely different animal from buying into a rising one.

Does a rising balance mean a purchase?

Not necessarily. Coins moving between an entity’s own addresses register the same way, and no on-chain measure separates that from a decision to buy.

What does a middling score hide?

Who is doing the work. A balanced market and one where large cohorts buy while small ones sell produce the same headline reading.

Why is the relative scale the default?

Because one substantial owner shifts more coin in an afternoon than the entire population of small accounts manages over weeks. Left raw, only the top row carries any tint and the split cannot be seen at all.

Why is the palette anchored rather than adaptive?

So a tint reports the same thing at every setting, and a single exchange reshuffle cannot wash the colour out of everything else.