Bitcoin exchange reserves: what outflows really mean

Bitcoin exchange reserves are the total BTC sitting in wallets that analysts attribute to centralized exchanges: the hot, cold and deposit addresses of Binance, Coinbase, OKX, Kraken and the rest. Traders read the number as the stock of coins that can be sold at short notice. The fewer bitcoins held on exchanges, the less supply can hit the order books quickly, and sustained outflows are usually read as coins moving into long-term storage. According to CryptoQuant data, exchange reserves fell from roughly 2.8 million BTC in February 2026 to about 2.67 million BTC by October 2026. This guide covers how the metric is built, why two data providers can disagree by a factor of two, what inflows and outflows meant in past cycles, and why spot ETFs changed what the chart actually measures.

How exchange reserves are calculated

The Bitcoin blockchain is public, but addresses carry no names. To measure reserves, analytics firms first label the addresses that belong to exchanges, then sum their balances. CryptoQuant's documentation defines the metric as "the total amount of coins held in the exchange addresses" and notes that reserves are "the accumulated result of Exchange In/Outflow & Netflow": each day deposits are added and withdrawals subtracted.

Where the labels come from is spelled out in Glassnode's exchange data transparency notice. There are three sources: addresses verified by the exchanges themselves, public tags that pass quality checks, and clustering algorithms that track how coins move between an exchange's own wallets. Three properties follow, and they rarely make it into headlines:

1. It is a floor, not a total. Glassnode states that "reported exchange balances can largely be considered as lower bounds of the true balance." Unlabeled addresses are simply not counted.
2. History gets revised. When a new exchange wallet is identified, its balance is added to every past day. The chart you saw a month ago may look different today. Glassnode's notice says plainly that "reported exchange balances may undergo retrospective revisions."
3. It is not the exchange's own accounting. In its Proof-of-Reserves research, Glassnode writes that exchange balance metrics "do not reflect the officially reported numbers by the exchanges themselves" and carry no 100% guarantee.

A derived metric, the Exchange Supply Ratio, divides reserves by total supply, per CryptoQuant's definition. It is more useful than raw coin counts because Bitcoin's supply grows with every block.

Why CryptoQuant, Glassnode and Santiment show different numbers

Open three dashboards on the same day and the figures will not match, and not by a few percent. CryptoQuant's chart showed about 2.68 million BTC on October 3, 2026, the lowest reading in three years for that dataset, and about 2.67 million by mid-October. In the same days Santiment put the share of supply on exchanges at just 6.5%, which works out to about 1.3 million BTC.

The gap is in the label sets. One provider counts Coinbase's custody wallets as part of the exchange, another separates them; one tracks dozens of small venues, another only the large ones. That means you can only compare a provider's chart with its own history, never absolute levels across services. A headline such as "exchange reserves at the lowest since 2018" almost always means the lowest point on one specific company's chart.

The shares and the coin conversion in the table are our own arithmetic, based on 20.1 million BTC in circulation per CoinGecko on October 30, 2026.

What bitcoin exchange outflows mean

The classic reading is simple. People withdraw coins when they do not plan to sell soon: keeping a large balance on an exchange carries more counterparty risk than a cold wallet. Persistent outflows are therefore read as accumulation, persistent inflows as preparation to sell. CryptoQuant frames it the same way: a rising netflow trend indicates selling pressure, a falling one indicates buying pressure.

The best-known example is November 2022. After FTX collapsed, users pulled coins off platforms en masse. Glassnode's report of November 21, 2022 found Bitcoin "flowing out of exchanges at a rate of -172.7k BTC per month" and said the 30-day net position change of exchange balances had "reached a new all-time-high for outflows this week." That was not a price signal. It was a flight from counterparty risk, with BTC near its yearly lows at the time.

A recent example is October 7, 2026, when Santiment counted a net outflow of about 24,073 BTC in a single day, the largest daily withdrawal in seven months. Santiment's analysts wrote: "Outflows alone guarantee nothing, but falling exchange supply strengthens the bullish setup." That is the honest version: an outflow is an argument, not a forecast.

The opposite case came on January 22, 2026, when Cointelegraph reported that 16,653 BTC were sent to exchanges over two days, against a typical January daily netflow of roughly −2,000 to +2,000 BTC, and Bitcoin's push toward $90,000 "failed to hold." A large inflow is a reason for caution, especially when the coins arrive from old addresses or from miners.

Why the metric works worse since the ETFs

US spot Bitcoin ETFs have traded since January 2024, and they changed what reserves mean. The funds keep their coins with custodians, and eight of the eleven original ETFs chose Coinbase as custodian, as Glassnode noted. When an investor sells BTC on Coinbase and buys ETF shares, the coin physically moves from an exchange address to a custody address. On the chart that is an outflow, even though the coin is still in the same building and just as easy to sell through the fund.

In January 2025 Glassnode quantified the effect: exchange balances had dropped from about 3.1 million BTC in July 2024 to about 2.74 million, but ETFs had accumulated about 1.69 million BTC over the same period, so exchanges plus ETFs still held about 3.04 million BTC, roughly the level of early 2024. The firm's conclusion: "we believe the majority of this decline stems from coins reshuffling into ETF wallets managed by custodians like Coinbase," not a genuine supply shortage.

Corporate treasuries add a third layer. As of October 30, 2026, our "Bitcoin in corporate treasuries" block on the daily brief showed 181 public companies holding 1.31 million BTC, or 6.25% of the 21 million cap. Those coins also left exchanges at some point, but a single board decision can send them back. See our guide to companies holding bitcoin.

What it looks like on October 30, 2026

Stack the three layers that are visible in on-chain data and in our dashboard:

That is roughly 5.3 million BTC, about a quarter of all coins, sitting where they can be sold without an on-chain transfer: on exchanges, with ETF custodians, or on corporate balance sheets. Look at the first row alone and exchange supply appears at a multi-year low. Add all three and the pool of "liquid" bitcoin has not shrunk much. It has changed hands.

The market backdrop on the same date, from our dashboard: as of 4:00 AM ET on October 30, 2026, Bitcoin traded near $78,200, down 2.1% in 24 hours, and the Fear & Greed Index had dropped to 30 ("Fear") from 51 a day earlier. October's exchange outflows happened while the price was falling, not rising, which illustrates the point: reserves describe holder intent, not the direction of the next candle. Live figures are on the Bitcoin page.

How to read exchange reserves properly

Experienced analysts use reserves as context, not as a trigger. A few rules follow from how the metric is built:

1. Watch the trend, not the day. Single-day outflows of 20,000⁠–⁠30,000 BTC happen several times a year and are often an exchange migrating its own wallets. The 30- and 90-day changes are what matter.
2. Cross-check with price and demand. Outflows during a rally suggest accumulation; outflows during a decline more often reflect fear about a venue or a move into ETFs. On their own they predict nothing.
3. Add exchanges, ETFs and treasuries together. Since 2024, sellable supply lives in three places, not one. Fund flows are covered in our guide to Bitcoin ETF flows.
4. Split by exchange. Binance and Coinbase reserves often move in opposite directions: the first reflects Asian retail and derivatives, the second US institutions. The price gap between them is the Coinbase premium.
5. Check stablecoins. USDT and USDC flowing onto exchanges is buying power that has not yet become bitcoin, and it is read as the counterweight to BTC outflows. See our piece on stablecoin supply.

Common mistakes

Comparing levels across providers. CryptoQuant's 2.67 million and Santiment's 6.5% describe the same phenomenon with different address sets. "Reserves at a record low" only means something within one chart.

Mistaking internal transfers for outflows. Exchanges routinely shuffle coins between hot and cold wallets, create new addresses and consolidate old ones. Until the new addresses are labeled, that looks like customers withdrawing. Glassnode specifically warns that early signals of large inflows or outflows should be treated cautiously.

Counting a hack as accumulation. On October 29, 2026, roughly $6.3 billion in assets was drained from Binance hot wallets, as covered in our hack timeline. On a reserves chart that is also an "outflow," but it says nothing about holder intent.

Expecting low reserves to lift the price by themselves. Scarce supply only matters when demand is growing. Reserves declined for most of 2026, yet by October 30 Bitcoin sat 38% below its October 2025 all-time high, per CoinGecko.

Ignoring OTC trades. Large purchases by companies and funds go through over-the-counter desks and never touch exchange addresses, so a share of real demand is invisible in the metric.

Frequently asked questions

Where can I see bitcoin exchange reserves for free? Basic charts are available on the free tiers of CryptoQuant and Glassnode, and also at Santiment and CoinGlass; longer history and per-exchange breakdowns are usually paid. Only compare values from the same service.

Are exchange outflows a bullish signal? Historically, sustained multi-month outflows often accompanied accumulation, but they do not move price on their own. In November 2022 record outflows occurred with BTC near its lows, and in 2026 reserves fell during a declining market.

Why do reserves fall when people buy ETFs instead of withdrawing coins? Because fund custody wallets sit outside the exchange labels. A coin moving from Coinbase to its custodian shows up as an outflow even though it can be sold through the fund just as fast.

What share of bitcoin is held on exchanges? About 13% of circulating supply by CryptoQuant's October 2026 estimate, about 6.5% by Santiment's. The spread comes from different labeled address sets, not from a data error.

This is not investment advice.