On-chain analysis: what addresses evidence

The chain shows completely what happened — never who did it, or why.

This lesson has had no expert review. It was written for this platform and against the evidence it cites; nobody has gone through it independently.

Learning objectives

You can say which questions on-chain data answers and which it only suggests.
You can read a holder distribution as a statement about liquidity rather than as a verdict.

Check your prior knowledge

Answer these for yourself before reading on. Wherever you hesitate is where this lesson pays off.

What is in a transaction, and what is not?
How many addresses can one person hold?
How would you tell that two addresses belong together?

Core concept

Fact

Complete and anonymous at once

Every transaction is permanently visible: amount, time, addresses involved, contract called. That is a level of data an outsider never has in the traditional financial system. At the same time, nowhere is it recorded who stands behind an address or what intent a movement had. Both hold simultaneously, and most fallacies arise from carrying only the first half forward.

Assumption

Addresses are not people

One person can hold any number of addresses, and one address can hold for many people — every exchange and every custodian does exactly that. Any metric counting addresses therefore makes a silent assumption about that relationship. Clustering techniques group addresses into presumed entities by behavioral patterns; the result is a reasoned conjecture with an error rate, not a register extract.

Interpretation

Concentration is a statement about liquidity

When a few addresses hold most of a token, nothing follows about their intentions — but something does follow about the market: a sale from one of them meets a depth that does not grow to accommodate it. The useful formulation is therefore not “whales might sell” but: how many days of trading volume does the largest single holding represent?

Risk

Proximity in time is no proof of causation

A large movement shortly before a price drop is an observation. It can mean foreknowledge, a reshuffle between the same owner's addresses, a margin call, or coincidence. On-chain data does not separate those cases. Presenting it in a report as the explanation passes an interpretation off as an observation — the error this platform's statement types exist to prevent.

Definitions

Clustering: A technique grouping addresses into presumed entities by behavioral patterns.
Holder distribution: How a token's supply is spread across the addresses holding it.
Active addresses: Number of distinct addresses with at least one interaction in the period.

Model

Read addresses and amounts — observation
Set aside contracts and known custodians
Attempt attribution; mark the result as conjecture
Convert into days of volume — a statement about liquidity
State unresolved attributions explicitly

From raw finding to a defensible statement — Steps 3 and 5 belong together; doing 3 without 5 reports a conjecture as a finding.

Formulas

Largest holding in days of volume

Days = largest single holding × price / average daily volume

largest single holding: quantity held by the largest non-contract address
daily volume: average trading volume over a representative period

Limit: Assumes the holding belongs to one entity and is tradable. Contracts, custodial omnibus addresses and locked holdings distort the figure in both directions.

Worked example

Ten addresses, three explanations

Observation: 10 addresses hold 61 % of circulating supply
Identifiable among them: 3 contract addresses (staking, bridge, treasury)
Remainder: 7 addresses with no recognizable attribution

The three contracts are no selling pressure — they hold on behalf of many, or are locked. That leaves 7 addresses whose attribution is open: they may be seven owners, or one.

“61 % across ten addresses” becomes “up to X % across at most seven independent owners, attribution not established”.

Reading: The second wording is longer and the only one that holds. The difference between them is precisely the work an on-chain analysis consists of.

Retrieval

Nothing — addresses are the usual proxy.Addresses are not people; the figure can be distorted in both directions by one person with many addresses or one custodian with a single address.The figure is set too low.

“Ten whales control the token.”“61 % sits across ten addresses, three of them identified contracts; attribution of the rest is not established.”“The distribution is healthy.”

Exercise on real data

Read that metric's limitations and record why it is marked “not tracked” here rather than appearing with an invented source.

The “active addresses” metric in the catalog →

Compare which market statements this platform's data supports — and which require an on-chain survey of your own.

Dimension 1: market →

Application

You are to check whether a position in a token can be built. Which three on-chain findings do you gather — and which one decides?

Related case studies

Institutional reading

Asset management: How many days of volume does the largest third-party holding represent — and your own planned position?
Bank: Which of these findings could be documented with a named method and an error rate?

Metrics in this lesson

Key takeaways

The chain shows completely what happened — never who did it, or why.
Any metric counting addresses silently assumes how addresses relate to people.
Concentration holds as a statement about liquidity, not as one about intentions.