Have you ever stared at an exchange reserve report and wondered whether the headline is really about confidence, or just about people parking coins in a different bucket for a few weeks? I had that feeling again when the latest Bybit snapshot landed. Mainstream assets on the platform were listed around $19.6 billion. Bitcoin and Ether in customer accounts ticked higher. Tether did the opposite and slid more than eleven percent from the previous cut. That mix is not a morality play. It is a balance-sheet photograph taken at 03:00 UTC on September 23, independently reviewed, and easy to over-read if you treat it like a diary of trader intent.
What The Latest Bybit Reserve Snapshot Actually Shows
Proof of reserves is a point-in-time test. It compares selected customer liabilities with assets sitting in published wallets. It does not tell you who bought what, who withdrew, or who simply rotated from a stablecoin into a volatile coin because the chart looked less boring that week. Still, the September figures are detailed enough to be useful if you read them slowly.
The exchange reported 56,131 BTC in user assets against 58,721 BTC in wallets. That produces a Bitcoin reserve ratio of 104%. Ether user holdings sat at 551,868 ETH, with wallet assets at 570,018 ETH, for a ratio of 103%. Covered mainstream assets rose from about $18.1 billion in the prior report to $19.6 billion. Every one of the fifty tokens in the current set was described as fully covered or better.
A reserve ratio above 100 percent means reported wallet assets exceeded the customer balances included in that snapshot. It is not a yield. It is not a bonus. It is spare coverage on the line items the report chose to show.
I’ve found that people skip that last sentence. They see 110 percent next to USDT and assume the platform is “safer by ten points.” Coverage above par is comforting. It is not a full audit of every corporate obligation sitting off that page.
Bitcoin Inched Higher Without Breaking The Ratio
From the August 26 snapshot, Bitcoin balances held for users rose about 0.24%, landing near 56,100 BTC in round terms. The reserve ratio stayed at 104 percent. That is a small move. Small moves still matter when you stack them across months.
Go back to June and the picture gets louder. User Bitcoin on the same venue was reported around 49,309 BTC then. Set that next to the September 23 figure and you get more than 6,800 additional BTC in customer assets over that stretch. Proof-of-reserves data cannot say whether those coins were bought, deposited from self-custody, or shifted from another venue. It only says the liability line got heavier and the wallet line stayed ahead of it.
In my experience, that is the honest way to talk about exchange balances. You can notice the direction. You should refuse the fan fiction about “the crowd finally believing.” Maybe they believed. Maybe they needed collateral. Maybe they just left coins on the book after a trade. The spreadsheet does not pick a story for you.
Ethereum Rose Faster Than Bitcoin In This Window
Ether was the livelier of the two majors. User holdings increased 2.96% to nearly 551,900 ETH. Wallet holdings of 570,018 ETH lifted the reserve ratio one point, from 102 percent in an earlier cut to 103 percent now. That is still a tight spread. Tight can be fine when the wallets are public and the tree of liabilities can be checked.
Why Ether moved more than Bitcoin in this particular month is guesswork. Funding rates, airdrop farming, staking wrappers, simple preference. None of those motives appear in a reserve PDF. What does appear is a larger customer ETH stack and slightly thicker coverage.
Perhaps the most interesting aspect is how ordinary the increase looks once you stop treating every basis point as a manifesto. People hold Ether on exchanges for trading. They also hold it because withdrawing and restaking is a chore. Both explanations can be true on the same Tuesday.
Why USDT Holdings Fell While Coverage Improved
Tether went the other way. User holdings dropped 11.46% from the previous report to roughly 3.59 billion USDT. That was the cleanest decline among the large line items. Wallet holdings were still about 3.96 billion USDT, or roughly 364 million tokens above the liability figure. The reserve ratio moved to 110%, up from 105 percent in July.
Does a smaller USDT balance mean users fled stables into Bitcoin and Ether? It might. It might also mean they withdrew cash-like tokens to another venue, paid down margin, or rotated into a different dollar token. The report records a level. It does not record a path.
USDe did not rhyme with USDT. User balances of that token jumped 62.41% to about 516 million. That is a loud month-to-month change sitting right next to a double-digit drop in Tether. If you like narratives, you can braid them. If you like evidence, you keep them as two separate facts.
USDC coverage looked even chunkier. The snapshot showed about 334.8 million USDC in user assets against roughly $748.3 million in wallet holdings, a reserve ratio of 223%. Excess coverage that large can reflect operational buffers, incoming flows not yet matched in the liability cut, or simply how the exchange warehouses that particular stablecoin. Again, the ratio is not interest paid to you.
| Asset | User holdings (approx.) | Reserve ratio | Month-to-month note |
| Bitcoin | 56,131 BTC | 104% | User balances up 0.24% |
| Ethereum | 551,868 ETH | 103% | User balances up 2.96% |
| USDT | 3.59 billion | 110% | User balances down 11.46% |
| USDe | 516 million | At or above 100% | User balances up 62.41% |
| USDC | 334.8 million | 223% | Heavy wallet surplus |
Look at that table for ten seconds and then look away. The story is not “stables are dead.” The story is “this venue’s customers held fewer USDT and more of a few other things on one Wednesday morning.” That is already enough without dressing it up.
Fifty Tokens, A Rotating Cast, And A $19.6 Billion Headline
The September disclosure covers the fifty highest-AUM tokens on the platform. Ten names in the current set include SUI, PUMP, LIT, CAP, SPX, ZEREBRO, XPL, USDTB, PENGU and ZRO. Reserve ratios for that group ran from 101 percent to 125 percent. XPL sat at the top of that pocket at 125 percent. LIT followed at 121 percent. Selection is based on how much users actually hold, not on marketing taste.
Here is where sloppy recaps go wrong. This should not be framed as a simple jump from forty tokens to fifty in September. Fifty-token coverage had already been announced in July after a different batch of additions. September swapped in another ten for the current set. So the climb from $18.1 billion to $19.6 billion in reported mainstream assets should not be blamed automatically on “we added ten more tickers.” Composition changed. Prices changed. Balances changed. All three can move the dollar total.
- Coverage is scoped to selected high-AUM tokens, not every obscure listing.
- New names enter when user holdings justify a seat at the table.
- A rising dollar total can come from prices, deposits, or a different mix of coins.
- Each listed token was reported at or above 100 percent coverage in this cut.
I like the rotating-cast detail more than the headline number. It reminds you that “mainstream assets” is a defined basket. Change the basket and you change the sum. That is bookkeeping, not magic.
How Customers Can Check If Their Balance Was In The Tree
Bybit uses a Merkle-tree model so a user can test whether their own balances sat inside the liability set for that snapshot. You can view a Merkle path in-account or run open-source verification code. Each leaf represents a customer balance without publishing the neighbor’s laundry.
Wallet addresses are published separately so anyone can inspect assets held against those liabilities. Tools exist for checking ownership and balances across supported chains. An independent reviewer examines the recurring reports. The combination is recurring disclosure, proof of liabilities, wallet-ownership checks, and third-party review. That stack is stronger than a screenshot of a hot wallet. It is still not a statutory audit of the whole company.
Proof of reserves verifies covered customer liabilities against specified reserve assets. It does not inventory every corporate debt, lawsuit, or off-scope wallet.
If you only remember one limitation, remember that one. Point-in-time is a feature and a bug. It freezes a morning. It cannot freeze the next withdrawal spike.
Context After The 2025 Breach Without Turning This Into A Horror Recap
Bybit started publishing these reports in December 2022. The latest snapshot arrives more than a year after the February 2025 incident, when roughly $1.46 billion in Ether-related assets were stolen. After that event, the same reviewer verified that at least 1:1 backing across assets in the February 2025 reserve report had been restored within 72 hours. More recently, U.S. court support has backed efforts to trace funds, including orders that let the exchange seek information from U.S.-linked platforms.
That history belongs in the same article as the September ratios because trust is sticky. People who lived through that week will read 104 percent on Bitcoin with a different pulse than someone who joined last month. Fair enough. A rebuilt ratio is not a time machine. It is a current claim, checkable against wallets, sitting on top of a scar.
I will say this plainly. Recurring public wallets plus a Merkle path is the minimum I now expect from a large venue. Anything less feels like asking customers to take a vibe check. Anything more, such as a full external audit of the operating company, is still rare and still worth wanting.
What The USDT Drop Does Not Prove
Let me labor this point because social feeds will not. An 11 percent decline in user USDT on one exchange is not a referendum on Tether’s reserves. It is not proof that “everyone is risk-on.” It is not proof that Bitcoin demand is back. It is a change in one liability line at one firm at one timestamp.
Could some of those dollars have funded the extra Ether? Sure. Could they have left the building? Also sure. Could they have become USDe or USDC or a perpetuals margin slice? You already know the answer. All of the above is available.
- Treat each asset line as its own fact before you weld them into a thesis.
- Compare ratios across reports, not just dollar totals that mix price and quantity.
- Remember June-to-September Bitcoin growth is a multi-month path, not a one-week miracle.
- Use the Merkle tools if you actually keep size on the venue. Screenshots of headlines are not verification.
- Keep the 2025 breach in mind as context, not as a daily scare quote.
That list is unromantic on purpose. Romance is how reserve reports turn into clickbait.
How To Read Reserve Ratios Without Getting Hypnotized
A 104 percent Bitcoin ratio means the published BTC wallets exceeded customer BTC in the snapshot by four percent. A 223 percent USDC ratio means the surplus on that coin was enormous relative to the liability cut. Different coins can show different buffers because treasury practice is not uniform. Some assets sit in cold storage with slack. Some turn over faster.
Watch the direction of user balances and the direction of ratios separately. User BTC up and ratio flat can be healthy. User USDT down and ratio up can mean liabilities shrank faster than wallets, or wallets were topped up, or both. You do not need a novel. You need two columns and a calendar.
Simple reading frame: Quantity change = what customers held Ratio change = how thick the buffer looked that morning Basket change = which tokens even made the report Dollar total = price x quantity x basket Do not collapse all four into one adjective.
I’ve used that little frame for years on venue reports. It keeps me from writing “bullish” when all I know is “different.”
Why Exchange Balances Still Matter In A Self-Custody Era
Plenty of readers will shrug and say coins on an exchange are not real ownership. They have a point. Keys matter. So do order books. Liquid markets still live on venues. Collateral for derivatives still lives on venues. That is why a 6,800 BTC rise in user assets over a few months is not trivia even if you personally refuse to leave coins there overnight.
Think of exchange balances as a weather station, not a temple. The station can be wrong in a storm. It can still tell you which way the wind was blowing at 03:00. Ignore it and you fly blind. Worship it and you fly into the hill.
There is also a boring operational reason these reports keep getting longer. After several industry failures, customers asked for more than a blog post. Merkle trees, published addresses, and a reviewer on a schedule are the compromise that survived. Imperfect. Better than silence.
Stablecoin Mixes Shift Faster Than Blue-Chip Stacks
The USDT drop next to the USDe jump is a reminder that dollar tokens are not a monolith on trading platforms. One product can lose eleven percent of its user stack while another gains more than sixty. Fees, incentives, perceived peg risk, and simple interface defaults all shove balances around. Bitcoin rarely jumps sixty percent in user count on a major venue in a single reporting window unless something extreme is happening. Stables can, because they are working capital.
Working capital leaves. Working capital returns. That is their job. If you hold a stablecoin as a savings account, an exchange liability line is a strange place to look for comfort anyway. If you hold it as dry powder for the next dip, the dip will not send you a calendar invite.
Is the USDC surplus almost theatrical at 223 percent? A bit. I would rather see excess on a widely used dollar token than a thin print. Excess can also mean the liability snapshot and the wallet snapshot are not dancing to the same song that hour. Timing gaps happen. Ask anyone who has reconciled a brokerage account at month-end.
A Few Practical Checks Before You Inflate The Story
If you keep funds on the platform, run the inclusion check after each report. If the path does not match, that is a support ticket, not a social media thread. If you do not keep funds there, still glance at the wallet list when a number looks spicy. Curiosity is cheap. Blind trust is not.
Compare like with like. Bitcoin user balances versus prior Bitcoin user balances. Do not compare a fifty-token dollar total with a forty-token dollar total and call it “inflows.” Do not treat a reviewer letter as a substitute for understanding what was in scope.
And please, leave the 2025 hack in the “known history” drawer unless new facts arrive. Relitigating it every time a ratio prints 103 percent helps no one who is trying to decide whether today’s wallets match today’s leaves.
The Quiet Conclusion Hiding Under The Headlines
So where does that leave a reader who just wanted to know if Bybit users are “more bullish”? They held a little more Bitcoin, a clearer amount more Ether, and a lot less USDT on September 23 than in the prior cut. Coverage on the published set stayed at or above par, with some stables showing fat buffers. The report now talks about fifty tokens, with a reshuffled ten, and a mainstream asset total near $19.6 billion. Users can still test their own leaf. Reviewers still sign off on the recurring pack.
That is the whole omelet. No hidden truffle. I would rather you walk away slightly underwhelmed and better informed than thrilled and misled. Exchange reserve theater will keep producing numbers. Your job is to keep producing questions that the PDF can actually answer.
Will the next snapshot show USDT bouncing back while Ether cools? Maybe. Will Bitcoin user balances keep grinding higher from that June base near 49,000 toward something fatter still? Possible. I am not going to pretend the Merkle tree whispered the answer. It only confirmed that, on one September morning, the wallets the firm pointed at were still larger than the customer stacks it put in the report. For a market that has been burned by prettier stories, that modest claim is the one worth keeping.
