周东野

Everyone thinks liquidation losses are borne by the traders who get liquidated. This time with XPL, $556,000 in losses landed on a public vault.

At noon on November 1, in 6 minutes, $XPL fell 19%.
But 8 minutes before the crash, 10 accounts withdrew about $1.62 million in collateral from Hyperliquid. The less collateral they have, the smaller the drop needed to trigger liquidation.
Seconds later, all these accounts were completely liquidated. Under the rules, the platform’s HLP vault took over their long positions and sold them, losing $556,000.
And the money in HLP comes from ordinary people’s deposits.

There’s currently no evidence that they caused the crash. But the rules are clear: the yield you earn from the vault already includes covering other people’s liquidations.
There’s currently no evidence that they caused the crash. But the rules are right there: the yield is yours, the liquidation losses are everyone’s.
On the same day, $BTC also edged up about 1%. Bitcoin holding steady doesn’t mean leveraged altcoins are safe.

Would you still deposit your money in a vault that steps in to cover other people’s liquidations?
Yes / No?

Binance was hacked for $6.3 billion. On the first day withdrawals reopened, $3.64 billion left.

Withdrawals were suspended for a few days, then gradually resumed on Sunday. On the first day, $4.65 billion was withdrawn and $1.01 billion deposited. Net outflows accounted for 2.1% of its $171 billion in assets—$2 withdrawn for every $100.

Honestly, that figure was lower than I expected.
A hack on the scale of $6.3 billion would be catastrophic for any traditional financial institution. Binance saw outflows of just 2.1%. Its liquidity held up, and users didn’t flee.

CZ said: “98% of people chose to stay.”
That 98% is just another way of presenting the same data above. It’s a clever bit of spin, to be sure, but the figure itself isn’t fabricated.

There’s also a difference in how the figures are calculated: DefiLlama puts the 24-hour outflow at $4.05 billion, while Binance says that includes funds it moved into its new hot wallet.
The difference between the two figures depends on how you count.

The stress test is over—for now. But compensation is only the first step. Whether Binance can shore up security is what will really determine whether it can remain the industry leader.

💁💁Borrowing a line CZ used over the past couple of days: “Trust isn’t earned through words in good times; it’s earned through actions in adversity.”

The bank run appears to have subsided for now, but the warning bells left by this security crisis have not stopped ringing.
For Binance, honoring its commitments through compensation is only the first step. The key to how far it can go from here is whether it seizes this opportunity to overhaul its internal controls and risk management framework.

After all, the market may give an industry leader one chance to patch things up, but it won’t give it a second chance to make the same mistake with the same vulnerability.

@币安中文社区 @Yi He @Binance Wallet @币安Binance华语

#币安已处理46.5亿美元提款 #币安第一阶段15个网络已恢复提现

A new week is about to begin. I didn’t make the top 1,000 either of the last two times—all because you guys are so fierce. This is tough! #币安神算子

Over the past 24 hours, Binance quietly processed $4.65 billion in on-chain withdrawals. The system not only stayed online, but Gas fees didn’t even see any sharp, unusual spikes.

Many people panicked when sharing this figure, thinking that big money was running for the exits. But seasoned insiders who really understand how on-chain settlements work know that this was actually a liquidity stress test on a whole different level.

The last time the market faced this kind of single-day net withdrawal pressure was during the confidence crisis triggered by FTX’s collapse and the U.S. Department of Justice settlement.

But the underlying flow of funds behind this wave of withdrawals is completely different this time.

Tracking on-chain whale addresses shows that these $4.6 billion didn’t simply vanish. Instead, the funds followed three very clear paths:

The first was market makers rebalancing to hedge risk. Arbitrage between exchanges by market makers stalled following the earlier blowup, prompting major MMs such as Wintermute to quickly move liquidity back to cold wallets or decentralized lending protocols. This is a classic move to avoid risk exposure; they’ll return to the market once volatility subsides.

The second was institutional funds moving to compliant custody accounts. As yields on institutional-grade staking protocols in the Ethereum and Solana ecosystems have surged, large amounts of idle assets previously held on Binance’s CEX are being withdrawn in batches and moved into self-custody ecosystems to capture risk-free yield spreads.

The third was panic selling driven by retail FUD. But it accounted for less than 15% of total withdrawals.

The biggest mistake in trading is jumping to conclusions based on every little sign.

The fact that the exchange processed $4.6 billion in reserves smoothly, without delays or withdrawal restrictions, essentially validates its 1:1 reserve ratio and the resilience of its settlement layer. Compared with smaller platforms that tout high yields but could restrict withdrawals at any moment, the fact that funds can be withdrawn in large amounts at any time actually demonstrates the resilience of the entire underlying settlement architecture.

The uncertainty in the market ahead isn’t about how much more money will be withdrawn from Binance, but which sector this $4.6 billion in liquidity will flow into first to start compounding returns again once it settles on-chain.

Watching where large on-chain transfers land is far more valuable than blindly following the crowd and turning bearish.

#币安已处理46.5亿美元提款

b14g Protocol Attacked: Analysis of the Theft of 97 Million CORE

The b14g protocol was hacked, and 97 million CORE tokens were stolen. Technical analysis suggests the attacker may have exploited a vulnerability in the protocol. The team has suspended the contract and is tracking the funds. Beware of the risks of double staking, and monitor key levels. #Bitcoin# #SecurityIncident#

Binance hot wallet transfers out another 780 million USDC, with cumulative asset outflows reaching $3.3 billion

Binance’s hot wallet has continued transferring out large amounts of assets, moving another 780 million USDC today and bringing the cumulative total to more than $3.3 billion. Technical analysis shows weakening support at key levels, so the flow of assets from the hot wallet and the market’s reaction warrant close attention.

The hacker behind the Binance hack sold about $5 million worth of the small-cap token PROM into an almost-empty pool, getting just $71 in return.

Because on-chain pools like Uniswap price tokens based on the ratio of the two assets in the pool, and there was almost no ETH on the other side, dumping 908,000 PROM barely brought in any money.

The price gap didn't disappear—the PROM was immediately snapped up by arbitrage bots.

But overall, the hacker wasn't stupid. According to Lookonchain, he has sold more than 130 altcoins for about $279 million, almost all of it in ETH, because stablecoins can be frozen, while nobody can freeze ETH.

The lesson for us is this: market cap is just a quoted price. How much you can sell depends on how much money is there to buy.

Could you get back half of your altcoin's market cap if you dumped it all at once?

#币安将恢复第一阶段提现 #币安已覆盖60亿美元被盗损失

Everyone says ETF outflows mean a sell-off, but this time the data tells the opposite story.

On October 30, U.S. spot BTC ETFs saw net outflows of $643 million in a single day—more than the $440 million they bought over the previous four days combined. Yet $BTC fell just 0.9% that day, and edged up slightly the next day.

The real plunge came the day before: BTC fell 4.1% on October 29, while ETFs were still making small net purchases.

In other words, this time the price fell first, and ETF money pulled back afterward. These investors seem to follow the price, redeeming only after the drop.

【ETFs aren’t the ones leading the sell-off; they’re more like the last to leave after turning out the lights.】

So by the time you see a headline about “large ETF outflows” and decide to cut your losses, you’re often already a step behind. What matters more is whether outflows continue for several days—and whether the price breaks below its previous low.

The most talked-about topic in today’s briefing is $HYPE . The breakdown is in the poster.

Do you think BTC is holding firm at the bottom, or is this calm before a further drop?

#美国现货ETHETF连续23日净流出

Stop letting yourself be swept up by the blind hype that “November is bound to surge.” Understanding what these historical figures actually show can help you avoid half the pitfalls.

Across the 16 Novembers from 2010 to today, BTC rose 10 times and fell 6 times, delivering an impressive average return of 34.59%. At first glance, the win rate is over 50% and the average is astonishing—but these figures hide a major outlier trap:

The median return is actually just 10.26%.

That 34.59% figure is almost entirely due to two extreme rallies that pulled the average sharply higher: 2013 (when BTC surged more than 450% in a single month) and 2020 (the bull market’s main rally). Once you exclude years with extreme bull runs, November usually falls into one of two camps: either a moderate continuation of gains or a slow, grinding pullback.

The data doesn’t lie. Just two key factors shape November’s price action:

The quarterly liquidity “handoff effect” If October has already delivered a major shakeout or breakout, November is likely to follow the trend and accelerate. But if October has exhausted buying demand, November often brings institutional profit-taking and hedging ahead of year-end.

The fourth-quarter pattern in halving cycles In years following a halving, November is often a pivotal point in the middle of a bull market rally. But in non-halving years or bear markets (like November 2018, when BTC plunged 37%), November can instead be one of the months with the sharpest pullbacks.

Take a look at your current position and cost basis: are you betting on the median return of around 10% continuing, or counting on one of those rare explosive rallies?

Share your take in the comments: will you finish this November in the green or in the red?

#美国现货ETHETF连续23日净流出 #BTC

What the hell, Hornets, what kind of game are you playing? My points! I thought you had it in the bag, then you play like this in the last few minutes 😑🌶︎🐔#币安神算子

ArcBlock Founder Robert Mao Calls for a New Crypto Infrastructure

In response to security issues in the crypto industry, ArcBlock founder Robert Mao proposes rebuilding the infrastructure, emphasizing collective participation to improve asset security and user trust.#ArcBlock #CryptoInfrastructure

Bitcoin Price Holds Steady After Hack: A Market Analysis

Bitcoin was hit by a hack, but the incident did not trigger sharp price swings. Technical analysis suggests that improved market liquidity, participation by long-term investors, a clear regulatory framework, and a swift response from stablecoins were key factors.

North Korea's hackers now use AI as a phishing factory.

Microsoft says they're producing dozens of fake lure files in minutes.

A new Microsoft Threat Intelligence report says North Korean hacking groups are using AI to mass-produce phishing documents and to generate malicious code.

Why it matters
Phishing used to be limited by human time. Someone had to write each lure, one at a time. When AI can turn out dozens in minutes, the attacker's cost drops to almost zero while your inbox gets noisier. After this week's Binance hack, that is not an abstract risk.

The scam email used to take a day. Now it takes a prompt.

more worth your time:

Binance says it has covered about 80% of the stolen assets
→ The safest exchange is the one that can afford its own mistakes.

When the lure is written by AI, what's your best defense: better tools or better habits?

One story, explained properly, every day. Follow Dongye. See you tomorrow.

#币安披露黑客事件进展63亿美元损失已弥补八成

AI chose one of 19 gene therapies on its own and performed brain surgery on a mouse.
But halfway through the procedure, it refused to continue because of a calibration error. No amount of persuasion could change its mind.

The model was called Claude Sonnet 5, and it had just one goal: reduce seizures in mice with epilepsy. It chose the drug, the brain region, and the injection coordinates on its own, and made 32 observations during the surgery. Researchers say this is the first time an AI has both selected a treatment plan and carried it out on a living mammal.

Why does this matter?

Until now, surgical robots have always followed plans made by humans. Now the decision-making has been handed to AI—and it can say “no” at a critical moment. Even more interesting: the more capable Opus 5.5 and Opus 5 refused to operate, while only Sonnet 5 was willing to see it through. It turned out the model was right: a manual adjustment earlier had thrown off the control system.

The most dangerous moment for AI may not be when it goes rogue, but when it’s right and refuses to listen to you.

Four more stories worth a look today:
🔹
Privacy-focused blockchain Iron Fish announces it will end support and shut down its codebase due to an inflation bug
→ What privacy coins should fear most isn’t regulation—it’s money appearing in the ledger that no one can see.
🔹
Binance recovers around 80% of stolen funds and plans to gradually resume withdrawals
→ The money is back. Whether user trust comes back too depends on how smoothly withdrawals go this time.
🔹
TechCrunch: AI investment boom shows cracks as negative cash flow and rising debt fuel concerns
→ Big tech companies dare to burn cash because they’re all afraid of being the one who didn’t buy a ticket.
🔹
Google’s space data center prototype satellite completes its first in-orbit AI inference
→ When there’s not enough computing power, the next place to look is the sky.

If the AI on the operating table said, “I’m stopping,” would you have the doctor hit the emergency stop—or trust it for once?

Every day, we take one story worth your time and unpack it. Follow Dongye, and see you tomorrow.

#币安披露黑客事件进展63亿美元损失已弥补八成

OpenAI has slashed its fundraising target for this round by $10 billion, and its valuation has fallen from $1.4 trillion to $1.25 trillion.

On the surface, it’s about falling tech stocks and tight liquidity in private markets. But the underlying logic is brutal: the AI sector’s “endless spending” era is officially over.

Just last month, OpenAI was still pitching this as the “last chance to invest before next year’s IPO,” trying to raise $30 billion to lock in an unassailable moat in computing power. But buyers voted with their feet—the first tranche didn’t even reach its full allocation. Investors no longer blindly believe Altman’s compute story. They’re starting to do some very real math:

Trimming the fat from output ratios: Anthropic is moving toward an IPO, and open-source models are dramatically outperforming closed-source models on compute cost-effectiveness. The cost of iterating GPT-level infrastructure is rising exponentially, while the growth curve for commercial monetization is beginning to flatten.

The crisis of sunk costs in compute: Under an asset-heavy model, the cash burn rate for compute rentals and self-built data centers remains high. If subsequent funding rounds fall out of sync, a liquidity collapse could follow.

Valuation pressure: The $1.25 trillion valuation offered by private markets is still a heavy burden. If public markets can’t absorb an exit of this scale at the IPO, every dollar invested now carries a high risk of being left holding the bag.

The narrative bubble in private markets has been squeezed to the bone, and VC and institutional capital is accelerating its shift toward more cost-effective inference and deployment ecosystems.

Valuations in compute infrastructure and traditional large-model businesses will inevitably come under pressure in the second half of the year. The rules have changed: from here on, it’s not about who can tout the grandest AGI vision, but who can survive until their cash flow becomes self-sustaining.

#openai下调融资目标100亿美元

2026 Jiangsu City Football League Final Preview: Wuxi vs. Changzhou

Tonight, the final of the 2026 Jiangsu Provincial City Football League will set the Nanjing Olympic Sports Center Stadium alight. Which team will emerge victorious in the showdown between Wuxi and Changzhou? Traffic police remind fans to travel safely. #2026JiangsuFootballLeague#

My points finally passed 10,000, but unfortunately I still can't make the top 1,000. Is everyone this good? #币安神算子

Uniswap and a16z have finally had enough.

Illinois previously proposed a digital asset transaction tax, planning to take a 0.2% cut of on-chain transactions. It even dared to include the outrageous requirement to record private keys in the bill.

Fortunately, industry heavyweights didn’t let it slide this time. Uniswap Labs, a16z, and the DeFi Education Fund joined forces and submitted a hard-hitting comment letter: DeFi protocols, front ends, self-custody withdrawals, wallet transfers, and cross-chain bridges must all be excluded from the tax.

The state has now backed down, announcing a delay until July 2027.

But this shows one thing: the battle over compliance has moved out into the open. If local tax laws like this are allowed to gain a foothold, every on-chain interaction could become a compliance nightmare.

Keep an eye on what happens next. This isn’t just about taxes—it’s a battle over the boundaries of on-chain sovereignty.

#uniswap与a16z寻求伊利诺伊州defi税收豁免

Sun Xinran advances to the Guangzhou Open women's singles semifinals, becoming the fourth Chinese player to reach the last four

Sun Xinran has become the fourth Chinese player to reach the semifinals this season, where she awaits a challenge from Britain's Harriet Dart. Technical analysts are focusing on her service games and baseline attack; her ability to seize key moments will be crucial to advancing. #Tennis #SunXinran

Sun Yingsha narrowly defeats Shin Yu-bin to reach the quarterfinals, speaking candidly after the match!

In the women's singles round of 16 at the WTT Montpellier Champions tournament, Sun Yingsha edged Shin Yu-bin 3–2 to advance to the quarterfinals. After the match, she spoke frankly about the ups and downs in her form and praised her opponent's progress. A cool-headed analysis, with every word ringing true!