ON-CHAIN METRIC

Exchange Reserves

How much Bitcoin sits in addresses attributed to exchanges—and how reliable are those labels?

Open the Exchange Holdings dashboard

Bitcoin exchange reserves estimate the BTC held in addresses attributed to trading venues. They are best read as labelled on-chain balances, not audited statements of every exchange asset or a direct measure of sell orders.

Coins on an exchange may be easier to trade, but coins in self-custody can also be transferred quickly. Location changes potential access to liquidity; it does not prove intent.

How exchange balances are estimated

Providers combine addresses disclosed by exchanges with public labels, clustering heuristics and ongoing monitoring. Coverage and false-positive controls differ, so reserve estimates can disagree.

Wallet labels are updated over time. Historical series may be revised when a provider discovers new addresses or reclassifies old ones.

Why balances move

User deposits and withdrawals matter, but internal wallet reshuffles, custody migrations, collateral movements and business changes can produce large flows too.

ETF and institutional custody have added another layer: coins can leave an exchange-labelled wallet for a custodian without representing retail self-custody or immediate spot demand.

How to read reserves

Longer trends in the share of supply held on consistently tracked venues are more robust than one day’s move, but no timescale removes label risk completely.

Split the data by venue and confirm large moves with transaction context. A broad move can be more informative than one venue’s reshuffle, but it still does not prove a market-wide decision.

How to read it

Large outflow. Labelled exchange balances fell quickly; investigate destination and label changes.

Outflow. Net BTC left the tracked venue set.

Flat. Tracked balances changed little.

Inflow. Net BTC entered tracked exchange addresses; sale, collateral and custody remain possible.

Large inflow. A substantial arrival that needs venue and transaction context.

Exchange balances update inside the Exchange Holdings dashboard, with per-venue flows and concentration views.

Common questions

Do falling reserves mean investors bought Bitcoin?

No. Withdrawals can reflect self-custody, custody migration, internal operations or other transfers.

How are exchange wallets identified?

Through disclosed addresses, public labels and clustering heuristics. The result is an estimate that can be revised.

Why does the venue split matter?

A move isolated to one venue is more likely to reflect that venue’s activity. Broad moves still need confirmation.

Did ETFs invalidate the metric?

No, but custody flows make the old exchange-versus-self-custody interpretation less complete.

What is the safest use?

Track consistently defined balances over time, read them as estimates and investigate large changes before assigning intent.

ON-CHAIN METRIC

Exchange Reserves

How much Bitcoin sits in addresses attributed to exchanges—and how reliable are those labels?

Open the Exchange Holdings dashboard

Bitcoin exchange reserves estimate the BTC held in addresses attributed to trading venues. They are best read as labelled on-chain balances, not audited statements of every exchange asset or a direct measure of sell orders.

Coins on an exchange may be easier to trade, but coins in self-custody can also be transferred quickly. Location changes potential access to liquidity; it does not prove intent.

How exchange balances are estimated

Providers combine addresses disclosed by exchanges with public labels, clustering heuristics and ongoing monitoring. Coverage and false-positive controls differ, so reserve estimates can disagree.

Wallet labels are updated over time. Historical series may be revised when a provider discovers new addresses or reclassifies old ones.

Why balances move

User deposits and withdrawals matter, but internal wallet reshuffles, custody migrations, collateral movements and business changes can produce large flows too.

ETF and institutional custody have added another layer: coins can leave an exchange-labelled wallet for a custodian without representing retail self-custody or immediate spot demand.

How to read reserves

Longer trends in the share of supply held on consistently tracked venues are more robust than one day’s move, but no timescale removes label risk completely.

Split the data by venue and confirm large moves with transaction context. A broad move can be more informative than one venue’s reshuffle, but it still does not prove a market-wide decision.

How to read it

Large outflow. Labelled exchange balances fell quickly; investigate destination and label changes.

Outflow. Net BTC left the tracked venue set.

Flat. Tracked balances changed little.

Inflow. Net BTC entered tracked exchange addresses; sale, collateral and custody remain possible.

Large inflow. A substantial arrival that needs venue and transaction context.

Exchange balances update inside the Exchange Holdings dashboard, with per-venue flows and concentration views.

Common questions

Do falling reserves mean investors bought Bitcoin?

No. Withdrawals can reflect self-custody, custody migration, internal operations or other transfers.

How are exchange wallets identified?

Through disclosed addresses, public labels and clustering heuristics. The result is an estimate that can be revised.

Why does the venue split matter?

A move isolated to one venue is more likely to reflect that venue’s activity. Broad moves still need confirmation.

Did ETFs invalidate the metric?

No, but custody flows make the old exchange-versus-self-custody interpretation less complete.

What is the safest use?

Track consistently defined balances over time, read them as estimates and investigate large changes before assigning intent.