Buy Crypto
Markets
Spot
FuturesANT
Earn
Event Center
Rewards Hub
More
100%+ PoR
Overview Exchange reserves measure the elasticity of sellable supply rather than the direction of price, and the eight days following October 29 delivered the strongest confirmation of that on record.Overview Exchange reserves measure the elasticity of sellable supply rather than the direction of price, and the eight days following October 29 delivered the strongest confirmation of that on record.

Bitcoin Exchange Reserves Tracker: BTC Inflows vs. Outflows

Overview

 
Exchange reserves measure the elasticity of sellable supply rather than the direction of price, and the eight days following October 29 delivered the strongest confirmation of that on record. Between November 1 and 2, users withdrew 25,793 BTC from Binance, exceeding the 18,270 BTC stolen from the platform by 7,523 coins or 41.2%, while bitcoin fell from $81,690.49 on October 9 to $75,463 on November 6, a decline of 7.62% and the lowest level since September 16. The largest single exchange outflow event on record came with a fall, not a rally.
 
CryptoQuant's series puts aggregate exchange reserves near 2.67 million BTC, down roughly 130,000 from about 2.8 million in February and equal to roughly 13.35% of circulating supply. That level remains a multi-year low, but this update stresses something else: readings from different providers diverge by as much as 700,000 BTC, or 26.2% of the lowest figure, so the metric supports comparisons of change within one series and never comparisons of level across providers.
 
 

Key Takeaways

 
Aggregate exchange reserves stand near 2.67 million BTC on CryptoQuant's series, a multi-year low for that dataset, while other series over the same window show 2.83 million and 3.37 million, a spread of 700,000 BTC. Binance's tracked bitcoin balance fell from roughly 675,500 before the incident to roughly 652,300 on November 2, down 23,200 coins or 3.43% across four days, of which about 15,894 moved in the 24 hours after withdrawals reopened in phases. Of the 25,793 BTC that left Binance, 984 arrived at Coinbase and 1,422 at Kraken, together 2,406 coins or just 9.3%, leaving 90.7% that did not reach either venue. Bitcoin trades at $75,463, down more than 6% on the day, alongside a 5.5% fall in the Nasdaq and 6.4% in the Nasdaq 100, with more than $1 billion of derivatives positions liquidated over 24 hours.
 

Reserves at Multi-Year Lows While Price Slides

 
The divergence existed in October and has widened over the past month.
 
On CryptoQuant's reserve reading, aggregate exchange reserves stood near 2.67 million BTC on October 13 against roughly 2.8 million in February, a decline of nearly 130,000 coins this year. Against a circulating supply just above 20 million, exchange-held coins represent roughly 13.35%.
 
Price moved the other way. Bitcoin opened at $81,690.49 on October 9, recovered above $83,000 in mid-October and reached a monthly high of $87,402.34, then fell to roughly $76,200 on October 30 and closed the month near $77,500, down 7.3%. On November 6, TradingView data from Bitstamp showed $75,463, down more than 6% on the day, the lowest since September 16, after briefly breaking the 50-week moving average at $75,733 and settling just $270 below it. That is 7.62% below the October 9 level and 13.66% below the monthly high.
 
The macro backdrop that day shows this was not internal to crypto. The Nasdaq fell 5.5%, the S&P 500 4.3%, the Dow 3.4% and the Nasdaq 100 6.4%, while ether fell 7.5% and XRP and Solana each 5.5%. Gold rose 0.6% to $4,094 an ounce, WTI crude moved above $90 a barrel and Brent gained 2.5% to roughly $97. CoinGlass recorded more than $1 billion of derivatives positions liquidated over 24 hours, most of them bets on higher prices, with Cointelegraph reporting nearly $500 million in the first hour of US trading.
 
Reserves falling, price falling and liquidations concentrated on the long side all held at once, which establishes that the reserve level alone is not a reason to buy.
 

The Binance Event Was This Metric's Largest Stress Test

 
The October 29 incident produced something close to experimental conditions for testing the reserve metric, because it generated an outflow of unprecedented size in a very short window.
 
The scale needs establishing first. The attack ran between roughly 16:05 and 16:20 US Eastern time on October 29, lasted about 20 minutes, spanned ten blockchains and took roughly $6.3 billion, making it the largest crypto theft on record, surpassing the $1.5 billion Bybit incident in 2025. The bitcoin portion was 18,270 BTC valued near $1.5 billion at the time, implying roughly $82,102 per coin. Around $3 billion was frozen, all of it stablecoins frozen by issuers and partners, with roughly $1.5 billion unfreezable and roughly $1.8 billion remaining with the attackers. Binance said the attackers used EIP-7702 against its hot wallet and attributed the entry point to a compromised engineer's laptop. TRM Labs linked the incident to North Korea-affiliated actors.
 
Withdrawals reopened in phases on November 1, and the flow data that followed is the core of this update.
 
Item
Amount
Timing
Source
Bitcoin stolen
18,270 BTC
October 29
Consolidated disclosures
Bitcoin withdrawn by users
25,793 BTC
November 1 to 2
CryptoQuant
Binance tracked bitcoin balance
675,500 to 652,300 BTC
Pre-incident to November 2
A single tracker
Of which the latest 24 hours
About 15,894 BTC
After withdrawals reopened
Same
Platform-wide net flow
$4.649B out against $1.01B in
November 1, 06:00 to 15:30 UTC
Binance disclosure
Bitcoin arriving at Coinbase and Kraken
984 and 1,422 BTC
Within 24 hours
Whale Factor
 
Withdrawals exceeding the theft deserves separate comment. The 25,793 BTC withdrawn exceeds the 18,270 stolen by 7,523 coins, or 41.2%, meaning what left the platform was not only the stolen portion but also voluntary user withdrawals. At current prices, 25,793 BTC is worth roughly $1.946 billion.
 
More informative is where those coins went. On Whale Factor's tracking, Coinbase gained 984 BTC and Kraken 1,422 over the same window, together 2,406 coins or just 9.3% of what left Binance, leaving 90.7% that did not reach either venue. That indicates the outflow was most likely a genuine move into self-custody rather than a rotation between platforms, and for the reserve metric that distinction is decisive: rotation does not change the aggregate, while a move to self-custody reduces it.
 
21Shares' November outlook gives a consistent figure, citing bitcoin outflows of 30,000 coins worth roughly $2.35 billion on November 1 alone, and noting that sellable supply falls if those coins move into self-custody.
 
Then the conclusion. Aggregate reserves are falling, the largest single exchange outflow on record just occurred, more than nine tenths of those coins went to self-custody, and price fell 7.62% over the same window to a near seven-week low. If exchange outflows were a bullish signal, this data should have produced the opposite result. It did not.
 

What the Metric Actually Measures

 
Reserves are the quantity of bitcoin sitting at addresses labelled as exchange addresses at a point in time. The measure captures the scale of supply that could be sold quickly, which is to say the elasticity of supply, and captures neither holder intent nor price direction.
 
CryptoQuant's metric documentation and Glassnode's exchange data transparency notice both make the same point: this data depends on address attribution, and attribution changes when an exchange restructures its wallets, activates new consolidation addresses or adjusts custody arrangements. A labelling change appears on the chart as an inflow or outflow while nothing economic has happened on-chain.
 
This incident supplied a concrete example. Binance's SAFU insurance fund bitcoin wallet fell from roughly 15,000 coins to roughly 7,500, the decline following a single 7,499 BTC transfer, with the remaining 7,500 worth about $583 million at the time. That transfer moved coins between exchange-controlled addresses, and whether it registers as a reserve change depends entirely on which addresses the tracker classifies as exchange reserves. The same on-chain action can be an outflow, an internal transfer, or invisible, depending on the provider.
 

Readings Differ by 700,000 BTC

 
This is where using the metric goes wrong most often.
 
Series
Reading
Timing
Note
CryptoQuant aggregate
About 2.67 million BTC
October 13
A multi-year low for that series
Another cited aggregate series
2.83 million BTC, from 2.85 million
Reported November 4
Provider and timing not stated
Pre-incident level cited by 21Shares
About 3.37 million BTC
Before October 29
Coverage differs from both above
Binance alone, tracker
652,300 BTC
November 2
Not audited
Binance alone, CoinGlass
604,843 BTC
October 31
Explicitly stated as not directly comparable to the row above
 
The three aggregate readings differ by as much as 700,000 BTC, equal to 26.2% of the lowest, while the two Binance readings differ by 47,457 coins or 7.8% despite timestamps only two days apart.
 
The divergence comes from coverage: providers track different exchange lists and handle derivatives venue balances, custodial sub-accounts and internal consolidation addresses differently. There is therefore exactly one correct use, which is comparing change within a single series. Taking one provider's level against another's produces a difference with no meaning.
 

Inflows Are Not Selling and Outflows Are Not Conviction

 
This was the original judgment in this piece, and the data above now tests it in both directions.
 
The outflow side is covered: the largest single outflow on record accompanied a 7.62% decline. The trigger here was a security incident rather than improving holding intent, so reading it as conviction inverts the signal entirely. It also establishes that the driver has to be identified first, because an outflow triggered by a security incident, a regulatory action or a custody migration carries a completely different implication from one triggered by improving holding intent, and on-chain data cannot distinguish between them.
 
The inflow side requires the same discipline. Binance's tracked balance was actually rising before the incident, from roughly 663,100 BTC on October 6 to roughly 675,500 immediately before, an increase of 12,400 coins. 21Shares read that pre-incident build as a potential headwind on the grounds that more coins on exchanges means more sellable supply. The direction of that reasoning holds, but it is equally not evidence of selling: coins move onto exchanges for custody adjustments, market-maker inventory and derivatives margin requirements, none of which is a sale.
 
From September 20 to November 2, Binance's tracked balance fell from 704,800 BTC to 652,300, a decline of 52,500 coins or 7.45%. That window contains both the pre-incident build and the post-incident drop, and price fell across the whole of it. A reserve change in one direction cannot explain that path, which is precisely the boundary of this metric.
 
On the other channels between flows and price, the bitcoin ETF flow tracker and the guide to funding rates and open interest handle two further dimensions, and using the three together is more reliable than reading any one alone.
 

Putting Reserves Into a Working Framework

 
Reserves suit judgments about position sizing and volatility, not about direction.
 
Tighter sellable supply means an order of a given size produces a larger price impact, so when reserves sit low the same amount of capital corresponds to higher realised volatility, which bears directly on order type and position size. A $100 million order moves the book more in a 2.67 million BTC environment than in a 3.37 million one, whichever way price is heading.
 
Direction requires other inputs. ETF flows ran roughly $2.6 billion of net inflows in September and turned to roughly $290 million of net outflows in October, including about $380 million of net outflows on October 30, the day after Binance suspended withdrawals. On rates, the Federal Reserve held its target range at 3.75% to 4.00% on October 28 by 10 votes to 2, with both dissents preferring a further hike, and the December 8 to 9 meeting brings a new dot plot with hike odds running between 75% and 87.5% across sources. The 10-year yield was roughly 5.32% on October 8 and roughly 5.05% by October 28.
 
Put together, the role of reserves is to tell you how far price moves once a direction appears, not when it appears. On how the various drivers interact, the complete framework for what moves bitcoin's price sets them out systematically, while the guide to market cycle indicators supplies a longer-horizon reference.
 

Risks, Scenarios and the Dates Ahead

 
Three scenarios warrant separate preparation.
 
Supply tightens further while demand returns: reserves keep falling while ETF flows turn back to net inflows. This is the only combination in which a low reserve level does useful work, because it supplies both a bid and a constrained offer. The test is daily ETF net inflows turning positive and staying there, not reserves falling on their own.
 
Reserves rebuild: coins move back onto exchanges and sellable supply widens. The driver has to be identified before the implication can be read, since custody migration and market-maker inventory are not selling pressure. 21Shares noted reserves had risen to roughly 3.37 million before the incident, and a repeat of that direction implies lower volatility rather than higher.
 
Trust in Binance is not fully restored: the platform says it has replaced roughly 95% of the stolen assets and completed a zero-knowledge proof of reserves snapshot on November 2 covering ten assets, stating that user balances are backed at least 1:1, with Yi He saying the full zero-knowledge proof takes roughly 40 to 50 hours after the snapshot completes. Per The Block's report, that snapshot published neither per-asset balances nor a reserve ratio beyond the 1:1 claim. On DefiLlama data, Binance's tracked total assets fell from roughly $144.9 billion on October 29 to roughly $128.3 billion on November 2, a decline of roughly $16.6 billion or 11.46%, though that figure combines price declines, the stolen amount and withdrawals and cannot be read as withdrawals alone.
 
On dates, October payrolls publish on November 6, October CPI on November 10, MSCI decides on November 11 whether to remove non-operating treasury companies from its indices effective after November 30, and the Federal Reserve meets on December 8 and 9. In the current environment each of those matters more to price than the reserve level does.
 
Readers wanting to set reserve changes against price reactions around those dates can mark the event timestamps on the intraday chart on the BTC price and market data page. Spot and futures markets on MEXC typically see short-lived shifts in liquidity and spreads through dense data windows.
 

Exclusive View from the MEXC Editorial Team

 
The past eight days supplied the cleanest counterexample this metric has on record. If exchange outflows represented holding conviction and low reserves represented tight supply and therefore an upside case, then the largest single outflow ever recorded alongside multi-year low reserves should have produced a rally. The actual result was a fall from $81,690.49 to $75,463, down 7.62%, with the 50-week moving average broken. Anyone treating reserves as a directional signal stood on entirely the wrong side of those eight days.
 
What the market is most likely to misread is reading the 25,793 BTC of withdrawals as users voting with their feet on some view. It is first of all the direct consequence of a security incident, exceeding the 18,270 BTC stolen by 41.2%, driven by damaged trust in one platform rather than any change of view on bitcoin. The detail that matters is where those coins went: only 2,406 reached Coinbase and Kraken combined, or 9.3%, leaving more than nine tenths that did not. That establishes a genuine move into self-custody rather than a rotation between platforms, so aggregate sellable supply did fall. Supply falling while price falls is not a contradiction, because reserves never determined direction.
 
What genuinely warrants watching is the series rather than the level. The three cited aggregate readings differ by 700,000 BTC, 26.2% of the lowest, and the two Binance readings differ by 7.8% despite being two days apart. At that dispersion, any claim that reserves hit a new low has to name whose series it comes from, or the next article citing a different provider can reach the opposite conclusion. Comparing change only within a single series is a more valuable discipline than any threshold level.
 
For cross-asset investors, November 6 located this metric precisely. The Nasdaq fell 5.5% and the Nasdaq 100 6.4% while bitcoin fell more than 6% alongside them, with more than $1 billion liquidated over 24 hours and most of it on the long side. On a day like that, whether exchanges hold 2.67 million or 3.37 million bitcoin explains almost nothing about the price; what determines it is the discount rate across risk assets, and that is set by the December meeting and the November 10 inflation print. The usefulness of reserves is in telling you how far price travels once those external variables turn, making it a measure of magnitude rather than of timing.
 

FAQ

 

How much bitcoin is on exchanges now?

 
On CryptoQuant's series, aggregate exchange reserves stood near 2.67 million BTC on October 13, down nearly 130,000 from about 2.8 million in February and equal to roughly 13.35% of circulating supply. Readings vary substantially between providers, however, with other series over the same window showing 2.83 million and 3.37 million, a spread of 700,000 BTC. Any citation of this figure has to name the series.
 

Why do reserve figures differ so much between sources?

 
Because coverage differs. Providers track different exchange lists and handle derivatives venue balances, custodial sub-accounts and internal consolidation addresses differently. Two readings for Binance alone differ by 47,457 BTC, or 7.8%, with timestamps only two days apart. The correct use is comparing change within a single series, since comparing levels across providers produces nothing meaningful.
 

How much did the Binance incident affect exchange reserves?

 
Users withdrew 25,793 BTC from Binance between November 1 and 2, exceeding the 18,270 stolen by 7,523 coins or 41.2%. The platform's tracked bitcoin balance fell from roughly 675,500 before the incident to roughly 652,300 on November 2, down 23,200 coins or 3.43% across four days, of which about 15,894 moved in the 24 hours after withdrawals reopened. Platform-wide, $4.649 billion was withdrawn against $1.01 billion deposited between 06:00 and 15:30 UTC on November 1.
 

Where did those bitcoin go?

 
Mainly into self-custody. On Whale Factor's tracking, Coinbase gained 984 BTC and Kraken 1,422 over the same window, together 2,406 coins or just 9.3% of what left Binance, leaving 90.7% that did not reach either venue. The distinction is decisive for the reserve metric, since rotation between platforms leaves the aggregate unchanged while a move to self-custody reduces it.
 

Are falling reserves a bullish signal?

 
The past eight days do not support that inference. The largest single exchange outflow on record just occurred, more than nine tenths of those coins went to self-custody and aggregate reserves sit at multi-year lows, yet price fell from $81,690.49 on October 9 to $75,463 on November 6, a decline of 7.62% and the lowest since September 16. Reserves measure the elasticity of sellable supply, not direction.
 

Do inflows to exchanges mean selling is coming?

 
No. Coins move onto exchanges for custody adjustments, market-maker inventory and derivatives margin requirements, none of which is a sale. Binance's tracked balance actually rose by 12,400 coins before the incident, from roughly 663,100 on October 6 to roughly 675,500, and price fell over that period. A reserve change in one direction cannot explain a price path.
 

Can reserve data be distorted for technical reasons?

 
Yes. The data depends on address attribution, and an exchange restructuring its wallets, activating new consolidation addresses or adjusting custody arrangements changes that attribution, appearing on the chart as an inflow or outflow with nothing economic happening on-chain. The concrete example here is Binance's SAFU insurance fund bitcoin wallet falling from roughly 15,000 coins to roughly 7,500 on a single 7,499 BTC internal transfer, where whether it registers as a reserve change depends entirely on the tracker's address classification.
 

How should this metric be used?

 
For judgments about volatility and position size, not about direction. Tighter sellable supply means an order of a given size produces a larger price impact, so a low reserve level corresponds to higher realised volatility for the same amount of capital. Direction requires other inputs such as ETF flows, the rate path and positioning structure.
 

Which dates matter from here?

 
October payrolls publish on November 6, October CPI on November 10, MSCI decides on November 11 whether to remove non-operating treasury companies from its indices effective after November 30, and the Federal Reserve meets on December 8 and 9 with a new dot plot, where hike odds run between 75% and 87.5% across sources. In the current environment each of those matters more to price than the reserve level does.
 

Disclaimer

 
This article is for general information only and does not constitute investment advice, financial advice, legal advice, tax advice or any trading recommendation. Prices of crypto assets, equities and other related financial assets can move sharply, and the on-chain data, historical figures and third-party statistics cited here cannot guarantee future results, while the reserve readings referenced are point-in-time snapshots that differ substantially between providers and will change with subsequent disclosures and attribution adjustments. Readers should conduct their own research and reach independent judgments based on their own financial circumstances, investment objectives and risk tolerance, consulting a qualified professional where appropriate. The MEXC Crypto Pulse team accepts no liability for any direct or indirect losses arising from use of the information in this article.
 

About the Author

 
MEXC Editorial Team provides the latest platform updates, product launches, trading feature guides, and industry insights. With a focus on developments in the global crypto market, it delivers transparent information, professional data, and forward-looking analysis. Both beginners and experienced traders can find firsthand updates, in-depth observations, and official announcements related to the MEXC ecosystem.
 

Research References

 
 
Want the fastest access to MEXC's latest updates? Join our official Telegram group now!
Join MEXC Community: X (Twitter) | Telegram | Discord
Account Verification: Understand KYC | How to Complete KYC
External Content Platforms: Substack | Medium | Paragraph | LinkedIn | X(News)
Market Opportunity
Bitcoin Logo
Bitcoin Price(BTC)
$58,133.49
$58,133.49$58,133.49
USD

The articles shared on this page are sourced from public platforms and are provided for reference only. They do not represent the position or views of MEXC. All rights belong to MEXC Editorial Team. If you believe any content infringes upon the rights of a third party, please contact service@support.mexc.com for prompt removal. MEXC does not guarantee the accuracy, completeness, or timeliness of any content and is not responsible for any actions taken based on the information provided. The content does not constitute financial, legal, or other professional advice, nor should it be interpreted as a recommendation or endorsement by MEXC. For expert insights and in-depth analysis, visit MEXC Learn.