📊 Profits from the coin price; cash from selling coins
Miners who hoarded coins are starting to spend them like cash.
MARA’s Q3 report, filed on November 9, was blunt. Since quarter-end, it sold another 13,597 bitcoins, bringing in about $1.1 billion at an average price of $81,039. It held 34,012 coins as of September 30. Nearly 40% gone. About 20,400 remain.
Where did the money go? $750 million paid off all loans collateralized by bitcoin. The two loans were taken out on August 4, backed by 17,477 coins, and now those coins have all been returned. The company says less than 5% of its holdings are now pledged as collateral or lent out.
The books tell a more painful story. Third-quarter revenue was $164.2 million, down 35% year over year. Net income was $632.1 million, compared with $123.1 million in the same period last year. Of that difference, $885.3 million came from fair-value changes in its bitcoin holdings. The average price of coins mined was $71,688, versus $114,440 a year ago. Electricity costs per coin also rose from $39,235 to $39,674, while hashrate climbed 18% to 71.2 EH/s.
That $600 million-plus came from accounting. Cash has to come from selling coins. In the first nine months, it had already sold 26,984 coins, bringing in $1.9 billion, and cut total debt by $1.4 billion, or 37%.
The proceeds from selling coins have two other destinations. The board approved a $500 million buyback. The stock closed at $7.34 on November 10, 46% below its three-month high of $13.63. The other half is being bet on AI. A 485-megawatt gas-fired power plant in Ohio cleared FERC approval; in Texas, Matagorda secured site rights for up to 2 gigawatts; and subsidiary Exaion signed a four-year GPU lease worth $79 million. The company also hopes to sign two data-center leases before year-end, though it cautioned that they could slip into next year. It hasn’t stopped.
It’s not alone on this path. Riot sold 10,599 coins in the third quarter for $751.5 million. Hut 8 sold about $250 million worth. Keel liquidated its holdings. Core Scientific is exiting mining by year-end.
That’s the flip side of the corporate treasury narrative. When prices rise, coins are reserve assets; when they fall, they’re collateral. When collateral ratios tighten, debt gets paid down first. Then comes the buyback of the company’s own stock. Even the valuation metric changes—from per coin to per megawatt of power.
On-chain, all you can see are the transfers. The direction is in the 10-Q.
BTC: $73,680–$73,791, down about 1% in 24 hours. ETH: $2,061–$2,064, down 2.4%.
Today I put the three rankings side by side, and their intersection is empty.
The top 20 on the popularity ranking generated a total of 16.97 million discussions, down 6.4% from yesterday. BTC alone accounted for 8.32 million, nearly half. Ten of the 20 spots were U.S. stock tickers. QQQ ranked third and SPY fifth. Together, those ten drew 3.71 million discussions, backed by just 131 accounts, while BTC alone had 301.
Across the 63 assets on the fund-flow ranking, inflows totaled $37,239 and outflows $35,974, leaving a net of just $1,265. The top-ranked LYX alone brought in $13,865, eleven times the net total. Of the 63 assets, 44 had only one trader active.
The 25 addresses on the profit ranking made a combined $594,000 over seven days, all in profit, with a median of $18,000. Broken down by day, November 8 alone brought in $209,000, a third of the total. On the 10th, that fell to $35,000. So far today, it’s just $3,394, with only 13 addresses showing activity.
What made me pause was that empty intersection. Not one of the 20 assets on the popularity ranking appeared among the 63 on the fund-flow ranking. Nor did any appear among the top-earning assets linked to the 25 addresses on the profit ranking. The three rankings are telling three different stories.
The popularity ranking measures discussion volume: whoever has the catchiest name rises to the top. The fund-flow and profit rankings measure money. Today, money is sitting in small-cap assets nobody is talking about, with 6,100 sell orders against 4,184 buy orders. Looking to the trending list for direction means starting from the wrong place.
I make a habit of treating the popularity ranking like a thermometer, just to gauge sentiment. I look to the other two rankings for money and profits, and first discount the portion driven by a single address. What these three rankings taught me today is that having no direction is itself information.
There are 8 AI skills in this batch from Binance Wallet, and only 1 can touch your money.
On 11-10, the official Chinese-language account announced a batch of Agentic Wallet Skills, saying they let AI help you look up on-chain data, copy trades, and interact with DeFi protocols. The wording was “multiple skills released.” But how many is “multiple”? The official developer docs give an exact number: 8.
Seven are read-only. Token information, address profiles, contract audits, smart money flows, rankings, and wallet transactions—these skills only view and look things up. There’s only one with write access, called binance-agentic-wallet. It can transfer funds, swap at market price, place limit orders, and trade on prediction markets.
Write access isn’t kept in check by trusting the AI to behave. In the Binance app, you set a daily limit, restrict which tokens it can access, and decide how high-risk trades are handled. The rules are enforced at the API layer: transactions that exceed the limits are rejected, or sent back to you for a second confirmation. The wallet uses MPC, so the private key is never fully assembled on any single device, and the agent never gets access to the key. Its balance is separate from your main wallet, and you can freeze it at any time.
It supports four chains: BNB Chain, Ethereum, Base, and Solana. Installation takes just one command. It’s built in by default for AI Pro users, with no setup required.
What really determines how far AI can go is that daily limit. The amount you set determines how much it can do. Of the 8 skills, the only one that can take action is worth a closer look than the other seven.
🔍 The court hasn’t even opened the hearing, but the money has already been frozen
Last Friday, Poland’s Sejm passed the Restrictive Measures Act by 240 votes to 202. The media calls it Act No. 3003. One member abstained, 17 didn’t vote, and on Monday it was sent to the Senate.
The key is Article 2. The Act expands the definition of “funds” to match MiCA, putting crypto-assets on the same footing as cash, account balances, stocks and bonds. The freeze covers not only assets held in someone’s name, but also assets they hold or control.
The Interior Minister is the one who can press the button. They can designate people directly, or do so based on nominations from 15 institutions, including the Financial Intelligence Unit, the Financial Supervision Authority, the central bank governor, the Internal Security Agency and the tax inspectorate. The decision takes effect the day after publication in the government gazette, with no “please reconsider” step. You can challenge it in an administrative court. But the Act removes the rule that allowed a court to suspend enforcement, so the freeze remains in place while the case is being decided.
The list now has 554 entries: 119 companies and 435 individuals, most of them Belarusian and Russian nationals.
The consequences come in several layers: assets are frozen, providing funds to those listed is prohibited, they are excluded from public procurement, and foreign nationals are also added to the list of undesirable persons. There are some exceptions: funds can be unfrozen for food, rent, mortgage payments, medicine, taxes and legal fees. The maximum administrative fine for violating sanctions is 20 million złoty, or about $5.11 million at an exchange rate of 3.91. Hiding or transferring funds that should be frozen carries a sentence of six months to eight years.
The freeze applies to people on the list. Ordinary crypto holders aren’t in its sights unless they are listed. It’s worth understanding exactly what can be frozen. The Act doesn’t say how to obtain the private keys to self-custody wallets—and it can’t obtain them. The targets are exchanges, custodians and conversion services; they have to comply just like banks. On-chain transfers don’t need anyone’s approval. Withdrawals do.
There are two more gates. The Senate has 30 days and can amend or reject the bill. The President has 21 days to sign it, veto it or refer it to the Constitutional Tribunal. This president vetoed the crypto market law on May 15. The vote to override the veto in September was 241 to 198, 25 votes short of the three-fifths majority. This time, 240 votes wouldn’t be enough either; 266 are needed.
The market isn’t reacting. BTC is at $73,570–$73,688, down 2% in 24 hours. ETH is at $2,045, down 4%. Whether the list can freeze your coins has nothing to do with the price, and everything to do with where you opened your account and whether you hold your own keys.
On November 9, the management committee for the U.S. college football playoffs unanimously approved expanding the field from 12 teams to 24, effective in the 2027–28 season. On the same day, ESPN secured exclusive global broadcast rights through the 2031–32 season.
First, let’s look at the shape of the deal. Before, it was $1.3 billion a year for 11 games. Now, ESPN is paying about $345 million more a year for 23 games.
I divided the two figures. The price per game falls from $118 million to $71.5 million, a drop of nearly 40%. Buying the whole package is cheaper. The number of games rises 109%, while the total price increases just 26%.
That’s wholesale pricing. More volume means a lower unit price. Each item is worth less, but the total is bigger.
The seller gave ground, too. The Power Four conference championship games are being canceled, and the conferences will have to pay ESPN back about $140 million a year. So ESPN’s actual net increase in spending is about $200 million. In effect, it’s getting 12 new games in exchange for an existing product line.
The buyer is still getting futures. None of the teams that will play in those 12 games has been determined yet. The schedule only runs through the first full week of December 2027, and rankings haven’t even begun. The product hasn’t been made, but the money has already been paid. Pay first.
The same trade happens on-chain every day. Mining companies sell future computing power upfront, project teams presell locked-up output at a discount, and large over-the-counter orders get volume discounts. The discount doesn’t mean the asset is worse; it reflects the amount the buyer is taking all at once.
One more detail. ESPN bought out the exclusive global rights while keeping the right to resell them, and existing sublicensing agreements remain in place. The TV network has become a wholesaler. An intermediary.
The rules are changing along with the contract. The format has changed four times in four years: four teams in 2023; 12 teams with byes in 2024; byes removed in 2025; a separate berth for independent teams in 2026; and 24 teams in 2027. The contract runs through 2031–32, so the format is locked in until then.
The games themselves have changed, too. Sixteen games used to be played on campus; now it’s four. The top eight seeds get two rounds of byes. Seeds 9 through 16 host first-round games against seeds 17 through 24. The championship game is set for January 24, 2028, in New Orleans.
BTC: $74,259 to $74,365, barely moving in 24 hours. ETH: $2,090 to $2,093, down 1% to 2%. Broadcast rights are getting more expensive in the U.S., while global risk assets are standing still.
Nobody even knows who’ll play in a given game, and it’s already sold out. That’s the standard formula for every presale, on the field and on-chain alike.
From Friday to Sunday, the North American box office grossed $121.67 million, up 39.6% from the same weekend last year and 74.4% from the previous weekend. The numbers look good. The underlying picture doesn’t.
The winner was Toho’s Godzilla-0.0, in Japanese with English subtitles. It earned $31.61 million on 3,254 screens, or $9,714 per screen. That’s a record North American opening for a live-action foreign-language film. The previous record holder was 2004’s Hero, with $17.8 million. 2023’s Godzilla-1.0 opened to $11.4 million; this one made 2.8 times as much.
Runner-up was Warner’s The Cat in the Hat, on 4,003 screens—the widest release in the chart—which earned $20.5 million, or $5,121 per screen. It played on 23% more screens than the winner, but made only 53% as much per screen. Third place went to Angel Studios’ Drummer Boy, with $9.31 million on 2,850 screens, or $3,268 per screen. Warner had just merged into Paramount Skydance on October 6.
In Japan, it opened first on November 3, timed to Godzilla’s 72nd birthday, and made 2.7 billion yen, or about $8.7 million. Its worldwide total is around $63 million.
One more detail. It made $14.6 million on opening day, including $6.7 million from previews and early screenings: $3.7 million from early-bird screenings and $3 million from Thursday shows. That was 46% of its first-day total. The industry calls this strong demand. The same money had already been collected before the film opened.
Crypto works the same way. Trading volume on TGE day often rolls private sales, whitelist sales, and points redemptions all into day one. It looks like the token opened at its peak, but the trades were already done before launch. On-chain, only one thing counts: the timestamp of the transfer.
Screen count is distribution; per-screen revenue is demand. You can buy distribution by spending money. You can’t buy demand. Theaters and exchanges run on the same ledger.
BTC: $74,438 to $74,564, down 0.3% to 1% over 24 hours. ETH: $2,111 to $2,114, down 1.2%. North America set a foreign-language-film record this weekend, and crypto prices didn’t even bat an eyelid.
Today, five coins all called “牛逼” showed up across two leaderboards. Only one is rising.
The smart-money leaderboard ranks 57 assets by net inflows over 24 hours. 牛逼 is first among those with positive inflows: 23,825.88, up 83.1%, with 5 traders participating and a market cap of 4.71 million. The first and second in outflows are also both 牛逼: -32,644.26 and -29,905.93. Both are marked -100%; one has just 0.0000003 in market cap left, and the other 0.37.
The three names are identical, but their contract addresses are three completely different strings. The newest one was launched at 2:38 a.m. today. In less than ten hours, by midday, it had already gone through a full cycle of gains and losses.
There are more on the profit leaderboard. Of 24 addresses, 6 have 牛逼 listed as their most profitable asset. Open their address pages and they all show 0x9665…7777—the same one that tops the inflow leaderboard. Another address, 0x877af2, also made money on 牛逼. But its contract is 0x73c6…4444, and this one never appeared on the inflow leaderboard. Add the separate 0x9c91…7777 held by 0xa83b73, and in today’s data, the same name corresponds to five different contracts.
The leaderboards show names, not addresses. You can make one version today, another tomorrow, and replace it if it turns out badly—the name stays the same. Recognizing a coin by its name is treating a label as an identity.
The other numbers today are telling the same story. The top 20 in social buzz totaled 18,133,328, more than double yesterday’s 8,131,556, with BTC accounting for 40.2% on its own. Sentiment was split exactly down the middle: 10 Negative and 10 Positive. On the smart-money side, 30 assets had positive flows totaling 63,037.36, while 27 had negative flows totaling -93,621.78, for a net of -30,584.42, compared with -3,219.46 yesterday. Of the 57 assets, 35 had only one person trading them.
The 24 addresses on the profit leaderboard totaled 555,605.39, up 13.1% from yesterday. None were negative. The median win rate was 46.79%, with more than half below 50%. The only amount newly recorded by midday today was 3,653.44; yesterday’s 129,863.02 all came from positions closed the day before.
Every day, I look at these three leaderboards side by side. The first thing I check is the contract-address column. A name can help me find a coin, but it can’t tell me which coin it is.
Binance has turned rebates into a seven-tier system. But the total commission paid out across the entire team, from top to bottom, is capped at the rate the top affiliate would have earned on their own.
On 11-09, Binance published an announcement that almost nobody shared, launching Sub-Affiliate. Affiliates can invite sub-affiliates, who can invite others below them, for up to seven tiers. Top-level affiliates can invite a maximum of 50 people in total. It sounds like a plan to drive growth through referrals, but the announcement spelled out the ceiling first: the team’s total payout will never exceed the top-level affiliate’s rate, and Binance pays each person’s commission directly, without money passing through an upline’s hands.
Two numbers are what really change the distribution: 20% and 5%. For each sub-affiliate added, at least 20% of the spot rate and at least 20% of the futures rate must be shared, while the affiliate must retain at least 5% of the overall rate. That 5% only applies when the rate is set. The announcement adds that if the rate is adjusted downward because an affiliate’s tier rating falls, the affiliate’s own retained share is cut first. It may fall below 5% or even to zero; the next tier’s share is only affected after that.
Suppose an affiliate has a spot rate of 40%. To invite a sub-affiliate, they must share at least 8 percentage points and keep 32. If that sub-affiliate invites someone else, they in turn share 20% of those 8 points. The amount shrinks at each tier, while the overall pool of 40% remains unchanged.
The requirements aren’t especially low, either. Joining is free, but commissions come only from actual trades; simply bringing someone in earns nothing. To become a sub-affiliate, you must first pass KYC and open a futures account, must never have been an affiliate before, must not be associated with another affiliate, and must not be in a restricted region.
For people who build communities and create content, this design replaces “one person bringing in others” with “revenue sharing at every tier.” You don’t control the rate, but you do control the tiers.
On November 9, PNC Bank issued an announcement saying that business clients can now accept USDC payments.
Sounds unremarkable. The key detail is the direction: funds flow in.
Payments go over Ethereum and are automatically converted to US dollars at a 1:1 rate upon arrival, then deposited into a PNC account designated by the client. The payment interface is provided by the largest publicly listed crypto exchange in the US, while PNC provides the accounts and treasury management. During the pilot phase, the service is available only to eligible clients.
So there’s only one thing businesses can do: receive. The service doesn’t send payments, and there’s no crypto to withdraw. The moment the money arrives, the coins are gone; all that remains is a dollar deposit.
Not even for a second. In, but never out.
The same two partners offered the opposite arrangement in their previous product. Eligible PNC private banking clients could trade digital assets: buy, hold, sell, with custody included. Now, for businesses, the rules have changed: no holding crypto.
One bank, two types of clients, two sets of books. Wealth clients can hold crypto; business clients can only let it pass through. The door opens in one direction.
On this route, the stablecoin acts like a remittance slip. A cross-border payment comes in, gets converted to dollars, and then proceeds through the usual treasury systems. The bank takes in a deposit; the other side collects an interface fee, whose rate hasn’t been disclosed. At the end of the third quarter, PNC had $466.4 billion in deposits and $627.8 billion in total assets. Deposits are its raw material, not crypto.
Exchange-rate risk disappears too. Whatever happens to the coin’s price, it’s converted upon arrival, leaving the business with only dollars on its books. Over the past few days, BTC has fallen from $80,000 to $75,000. Merchants holding coins while awaiting settlement would see their books fluctuate too.
The announcement leaves out three figures: fees, a launch timeline, and transaction volume. It doesn’t even compare settlement times to support claims of “faster” or “cheaper.” All we can confirm is the process. There’s not a word about the results.
As for scale, $70.36 billion worth of USDC is in circulation, of which $42.64 billion is on Ethereum—about 60%. All stablecoins combined total $293.8 billion. This new entry point is aimed at that pool.
Crypto doesn’t earn interest; deposits do. That alone explains why the bank only lets you receive it.
BTC: $75,122–$75,228, up 0.5%–0.7% in 24 hours. ETH: $2,131–$2,134.
⚡ Old GPUs aren’t being renewed; racks are waiting for power in the warehouse
The world’s largest compute tenant has started choosing machines by age.
Bloomberg reported on November 6 that Microsoft is no longer renewing its AI compute leases with Oracle and has been winding them down over the past few months. The servers in this batch from Oracle use older-generation Nvidia chips that can’t keep up with Microsoft’s current specs. Wrong vintage. Reuters echoed the report the same day; Microsoft declined to comment. An Oracle spokesperson said demand for OCI remains strong, including for earlier GPU generations. In September, Oracle also disclosed that it was renewing GPU leases deployed four years ago at prices 20% above the original contracts.
The market moved faster than the words. Oracle shares fell as much as nearly 5.2% intraday, setting a new low of $109.21, then closed up 0.8% at $116.10. That’s half the June 1 price of $248.15.
The Bloomberg report the day before was more pointed. Microsoft had committed to more than $80 billion in leases with newer cloud providers like CoreWeave, Nscale, and Nebius, and is now reviewing them one by one. It won’t exercise the option to expand Nscale’s Texas project by about 700 megawatts; the existing 240 megawatts remain unchanged. It has already pulled out of its flagship project in Loughton, Essex, UK, worth about $3 billion. Local power won’t be available until the 2030s. Waiting for power. Amazon moved earlier: internal emails show AWS is taking delivery ahead of schedule under a new process. About $5 billion in GB300 racks arrived in August, followed by roughly $3 billion more each month. Some data centers have room for the racks but not enough power, so the machines are sitting in intermediate warehouses. Power is the bottleneck.
Ending leases doesn’t mean Microsoft is pulling back. Last quarter, Azure revenue was $33.165 billion, up 48% year over year; commercial remaining performance obligations were $677 billion, up 73%; quarterly capital expenditure was $45.2 billion, up 30%; and free cash flow fell 27% year over year. The money is shifting toward building its own infrastructure.
The contrast with crypto is clearest here. IREN was originally a Bitcoin miner. In its November 5 quarterly report, AI cloud revenue was $200 million, up 182% quarter over quarter, with an 86% gross margin. Bitcoin mining revenue fell from $67 million to $28 million, and mining rigs are being dismantled. Its GPU hardware on the books rose from $2.04 billion to $5.82 billion, with $4.61 billion in construction in progress. This is also the company that, after Microsoft requested a network design change, pushed its $4 billion annualized revenue target from year-end to January next year, with a possible delay to the end of February.
One company, two depreciation schedules. Two clocks. GPUs are replaced every three or four years; data centers are depreciated over 20. The difference comes down to who owns the land and controls the power. That’s why it can sign five-year contracts at $220,000 per megawatt, with customers prepaying 90% of GPU capital expenditure. Short-term contracts are quoted at $40 million per megawatt. When old-card leases expire, they’re renewed at double the price, and buyers still have to queue for deliveries. The mining industry has been through this before. A new ASIC generation comes every three years, and resale value moves with coin prices and electricity costs. Platforms that lent money against mining rigs in 2021 were later crushed by falling used-rig prices. Nobody wanted to take them.
A third party offers a yardstick. On October 14, Gartner estimated that, at the current rate of spending, more than $3.5 trillion in AI infrastructure could be impaired before 2030. It would take $8.5 trillion in revenue to cover that spending plus a 12% return on investment; Gartner forecasts related revenue of at most $3.6 trillion. The gap is there.
In the markets, BTC was at $74,285–$74,406, down 2.1% over 24 hours. ETH was at $2,111, down 5.6%. AI-themed tokens didn’t all move down in unison. TAO at 218.6 was down 2.9%; FET at 0.1423 was down 1.1%; RENDER at 1.602 was up 1.8%; NEAR at 3.53 was down 5%.
Compute leasing now sells two credentials: delivery date and chip vintage. Old GPUs won’t rot in a warehouse; they’re waiting for a buyer who doesn’t care about the year they were made, and a power line that’s live.
📊 Factory-gate prices rose 5.7%, while consumer prices rose just 0.7%
At 9:30 this morning, the National Bureau of Statistics released two sets of figures moving in opposite directions.
Industrial producer prices rose 5.7% year over year in October, up 0.7 percentage points from September’s 5.0% and the highest since July 2022. They accelerated for the third month in a row. Month over month, they rose 0.7%. All the price increases were upstream: coal mining rose 36.6%, oil and gas extraction 35.0%, and petroleum and coal processing 29.9%.
The consumer price index, released the same day, rose just 0.7% year over year, below the market forecast of 0.8% and last month’s 0.9%. Month over month, it was unchanged. Food prices fell 2.9%, fresh vegetables fell 16.9%, and pork fell 8.6%. Core CPI was 0.8%, a nine-month low.
Put the two together and the gap is 5.0 percentage points. It was 4.1 in September. The gap is still widening. Prices are rising upstream, but there’s no one to pass them on to downstream.
A closer look reveals two more telling figures. The biggest support for CPI came from gasoline, up 22.4% year over year and contributing 0.63 percentage points on its own. The portion of oil-price increases driven by the war has already made its way from wholesale to retail. On the other hand, data storage equipment rose 45.4%, tablets 29.3%, computers 23.0%, and mobile phones 16.7%. Together, these four categories added 0.26 percentage points to CPI. The memory chips snapped up by AI are already showing up on consumers’ shopping lists.
Gold jewelry is moving in the opposite direction. Its retail price fell 5.9% month over month in October, with the decline widening by a further 3.8 percentage points from September. Its year-over-year increase narrowed from 22.5% the previous month to 4.5%. International gold prices briefly fell below 4,000 during the month. Store prices followed. Today, COMEX is back at 4,129.9. The quoted gold price and business at the counter are no longer moving in lockstep.
Markets barely reacted. The Shanghai Composite rose 0.48% to 3,665.94. BTC is at 74,807, down 40.7% from 126,199 on October 6 last year; ETH is at 2,139, down 4.3%. The S&P 500 fell 2.68% last week, its fourth straight weekly decline, while BTC dropped from 80,504 to 74,807 over the same period. Last Friday, the 10-year Treasury yield was 4.86%, and the dollar index was 102.17.
Read the two sets of figures side by side: pricing power in energy and chips lies abroad, while pricing power in food and rent lies at home. We can’t control the half where prices are rising, and demand hasn’t yet arrived to support the half where prices are falling.
For crypto, only one part is a hard constraint. Mining rigs, storage, and electricity are all tied to upstream costs, which are rising. Risk appetite is tied to the stock market, and it’s shrinking. Today, Bitcoin chose the latter.
U.S. October CPI comes out tomorrow. The Fed’s rate decision is on December 9, and markets are pricing in a 77% chance of a rate hike.
The profit leaderboard made 491,000 today, while smart money recorded a net outflow of 3,219. Both numbers are real, but they refer to different time periods.
The top 20 assets on the social leaderboard had a combined heat score of 8.13 million, up from 5.87 million yesterday—a 38% increase in one day. BTC alone accounted for 49%, and ETH for 24.7%; together, they made up 73.7%. The number of KOLs for BTC climbed back from 161 to 204. ETH’s 24-hour curve rose from 760,000 to 2.01 million, the steepest climb of the day, yet its sentiment column reads Negative. U.S. stock tickers were back up to nine spots. QCOM, in third place, had a heat score of 450,000, but its KOL count was just one.
The flow of money has reversed. Of 64 assets, 37 had positive inflows totaling 76,827, while 27 had negative inflows totaling -80,046, for a net of -3,219. Yesterday, that figure was +147,052. ZIG, the top asset for inflows, took in 41,406—53.9% of all positive inflows. Its trader count was listed as 1. Across the entire leaderboard, 49 assets had just one trader each, accounting for 76%. ANT had the largest outflow at -22,543, with nine traders retreating together.
The profit side looks steadier. Across 23 addresses, the 7-day total was 491,310, with a median of 17,346 and not a single negative result. But breaking it down by asset reveals the catch: the largest contribution for 10 addresses came from the same GM token, totaling 151,000, or 30.9% of the entire leaderboard. For eight of them, GM accounted for more than 90% of their own profits. The records for 0xd40df4 show just six assets traded and 19 trades, with all 16,119 coming from a single token, GM.
The two leaderboards use different time windows. The profit leaderboard counts positions closed within the past 7 days, while the inflow leaderboard measures net flows over the past 24 hours. When one is positive and the other negative on the same day, it means most of the money on the leaderboard was realized in previous days. Today, no new money came in; instead, money is flowing out.
I lay out all three leaderboards side by side every day to keep an eye on this time lag. Whether the numbers look good or not can wait—the first thing to check is which day they’re talking about.
The person who spent five years as Binance’s CFO announced last night that he was leaving. Five hours later, CZ replied, saying right away that he was willing to invest in him.
On the evening of November 8, Kaiser Ng wrote on X that after five years as CFO at Binance, he was now going to go out and do something of his own. He said the past ten years in the digital-asset exchange industry had gone by remarkably fast because he loved the work.
CZ’s reply left no room for empty pleasantries. He said Kaiser had done a great job and was a good friend, and that Kaiser had mentioned his plans to start a company a few months ago. YZiLabs has always been willing to invest in companies founded by current or former colleagues.
The key is the last part. In places where people leave and are soon forgotten, executive turnover is a loss. Here, it becomes an opening for capital. People move on, but their projects can still receive funding from Binance-affiliated capital, their experience remains in the ecosystem, and their status shifts from drawing a salary to holding equity.
What does this mean for $BNB holders? Binance no longer defines itself by the output of any one department. Today at 12:00 UTC, He Yi and Richard Teng are also set to appear on X Spaces with Chief Security Officer Jimmy Su. The CFO’s departure hasn’t disrupted the schedule.
Being able to let people go—and still have the money follow them. That says more about what Binance has become than any announcement could.
🔍 A $60 billion war bill, in exchange for a shipping lane that requires a nod
Iraqi Foreign Minister Fuad Hussein gave Al Jazeera a figure on Sunday: the war with Iran has cost Iraq $60 billion.
At Brent crude’s price of $98.02 a barrel, that works out to 612 million barrels of oil. Iraq exports 3.35 million barrels a day, so that’s equivalent to wiping out half a year of the country’s export revenue.
The oil still has to be sold. In the same interview, he said Iraq had reached an understanding with the United States allowing crude to leave through a shipping lane near Oman in the Strait of Hormuz. On the same day, the U.S. Treasury Department called on Iraq to close all but two of its border crossings with Iran, one of them in the Kurdish region.
The northern route is also picking up steam. On November 4, Iraq’s exports through the Turkish port of Ceyhan rose to 500,000 barrels a day. When the strait is blocked, the pipeline takes over.
The land routes are under pressure too. That same day, a newly established U.S. command task force was tasked with choking off Iran’s overland supply lines, while Iranian officials publicly played it down. Choking off supply lines and closing border crossings are two halves of the same move.
The shipping lane needs approval; the border crossings need to close. There’s no toll to use this corridor—the price is picking a side.
The market moved first. Brent stood at $98.02, up 53% from a year ago. WTI fell 13% in a week to around $90, its steepest weekly drop since mid-April. Trump said that day that if negotiations made progress, another strike on Iran was “extremely unlikely.” Iran’s response was that it would make its adversary “rain fire.”
Bitcoin didn’t take the deal. Quotes across six sources ranged from 74,553 to 74,734, down 1.3% to 1.5% over 24 hours. Ethereum ranged from 2,131 to 2,135, down 4.4% to 4.6%—a sharper drop than oil.
Easing tensions in a war is usually considered good news for risk assets. Not this time. Oil is shedding its geopolitical premium; crypto is tracking positioning and liquidity.
The shipping lane sends a pretty clear message: the right to pass belongs to whoever issues the permit.
⚡ The IPO hasn’t been priced yet, but the dilution math is already on the table
Anthropic is set to list on Nasdaq under the ticker ANT, sometime in the next few weeks. CNBC said on November 7 that the target is mid-November.
The pricing outlook has already softened. Last month, the market was talking about $2 trillion. On November 6, CNBC quoted several investors saying that even $1.5 trillion was now in doubt. One called that figure “a hard sell”; $1 trillion to $1.25 trillion sounded more realistic. The company’s May private funding round valued it at $965 billion.
The crypto world is doing a different kind of math. On November 5, BitMEX founder Arthur Hayes published “D is for Dilution,” laying out the vesting schedules for SpaceX, Anthropic, and OpenAI.
SpaceX listed on Nasdaq in June under the ticker SPCX, priced its shares at $135, raised about $75 billion, and was valued at $1.77 trillion. By June 2027, about 8.6 billion shares are set to unlock—equivalent to adding half the current share count. For Anthropic, Hayes estimates from the amended prospectus that early public investors could face dilution of up to 83% by June 2027. At a $1.5 trillion valuation, for the share price to stay where it is, someone would have to come up with another roughly $1.245 trillion to absorb those shares. Who’s going to pay?
That figure is not the company’s funding requirement. crypto.news made a point of noting that unlocking old shares is not the same as issuing new shares. The 83% figure is Hayes’s estimate based on his reading of the filing.
The real question is where that money would come from. Hayes’s chain of reasoning is short: share prices fall, borrowing gets more expensive for AI companies, banks pull back both lending and stock financing, bank balance sheets stop expanding, and there are fewer dollars. Fewer dollars means less fuel for Bitcoin. That’s the whole chain. He acknowledged that bank balance sheets had declined week over week last week, and also that one month of data isn’t enough to establish a trend.
D.A. Davidson’s Gil Luria put it more bluntly to CNBC: the market needs this IPO to succeed, the underwriters understand that, and the best thing to do is price it where it can sell.
What is Hayes doing himself? Maelstrom has been sitting on the sidelines since November 5, after previously closing about $22.28 million in directional positions. Waiting for a signal. On October 15, he was still bullish.
There was another backdrop that same week. In the midterm elections, most of the ballot measures to slow data center construction passed. AI lost its first round at the ballot box.
Raoul Pal disagrees with this line of reasoning, saying that debt at hyperscalers amounts to only about 3.5% of their market value. The three labs Hayes is watching aren’t the same companies as these cloud providers in the first place. That’s where the disagreement lies.
Crypto prices barely moved today. BTC ranged from $75,874 to $75,981, with six price sources tracking closely. ETH was at $2,238 to $2,241. AI-related tokens TAO at 224.8, FET at 0.1437, and RENDER at 1.57 were all within 1% of their previous prices.
The market is quiet because the price hasn’t been set yet. Once that opening price comes out in mid-November, AI stocks and AI tokens will both know where they stand.
📊 What was deemed in violation this time was the billboard
October 14, Lujiazui, Shanghai. An ad for a public blockchain lit up the exterior wall of the Citigroup Tower. The ad went up. Two days later, its first in-person gathering in Shanghai, originally scheduled to take place there, was stopped at the door and never got underway.
On November 6, a relevant department official at the People's Bank of China gave the official explanation through the Financial Times. The reason had nothing to do with venue capacity or the approval process. The problem was the ad.
The official explanation was brief. Solana is a major virtual-currency issuance platform; SOL is, in essence, a virtual currency; and the chain also provides technical services for the issuance and trading of major virtual currencies such as USDT. All three fall within the scope of activities prohibited by the notice issued by eight government agencies this February. Promoting the brand amounts to promoting virtual currency.
The organizers offered a different account at the time. On October 16, Solana Foundation President Lily Liu explained on X that registration had surged overnight to four times the limit set by local authorities, and they couldn't turn people away, so they had to postpone the event.
Both accounts are publicly on record. Only one will make it into the investigation's findings.
The key is what comes next. The central bank said it would keep a close eye on “domestic affiliates of domestic virtual-currency traders and overseas virtual-currency issuance and trading platforms.” Overseas public blockchains are beyond the reach of inspections; the people who can be investigated are those taking orders on the mainland. Whoever signed the contracts on the mainland—for ad space, venues, event production, or sponsorship—is on the list. No one gets a pass.
What this news is really about is something else. For years, public blockchains entering the Chinese market have talked about technology but not coins. This new definition closes that loophole. Issuing tokens on-chain, providing settlement for stablecoins, and putting the brand on display in a financial district: all three are now linked together, leaving no separate space for “just technology.”
There was no reaction in prices. SOL is now at $89.3, down 0.7% over 24 hours, with an intraday range of $88.73 to $90.39. BTC is basically flat at $75,925, and ETH is at $2,238. In the two trading days since the news broke, the market hasn't made a single meaningful move.
What has changed is the cost structure. Compliance used to cost as much as a legal opinion; now it comes down to whether domestic partners are willing to take the job. The path of brand exposure is narrowing, leaving only code and developers as entry points. Neither needs a billboard.
When you look at all three rankings together today, the buzz doesn’t hold up.
The top 20 assets on the social leaderboard had a combined buzz score of 5.87 million, down from 14.83 million yesterday—a 60% evaporation in a single day. KOLs fell from 267 to 161. BTC’s share, on the other hand, rose from 36.9% to 60.4%, even though its own buzz also dropped by 35%. The rise in its share has nothing to do with people piling in; it’s because the denominator collapsed first. The U.S. stock tickers that took up 14 spots yesterday are down to just four today.
Now look at the money. Smart money saw net inflows of 147,000 over 24 hours, up fourfold from yesterday’s 36,000, which makes it look like money is coming back. But 66 of the 73 assets had just one trader throughout. The top-ranked contract took in 108,000, or 55% of all positive inflows. Its market cap is listed as $1, its liquidity as 0, and its 24-hour change as -100%. This kind of record looks more like an empty shell being tallied after a pool was drained than a sign of anyone favoring a particular coin.
The 25 addresses made a combined profit of 473,000 over 7 days, all in the black, with a median of 16,500. Break it down by day and the cracks show: 39% came on November 6; yesterday’s total was just 8,743, and today’s is 121.61 so far. Nine addresses haven’t moved at all in the past two days.
All three rankings tell the same story: the numbers are still there, but fewer people are taking part. The buzz is concentrated in one coin, the money in one pool, and the profits in one day. On the day prices fell below 75,000, open interest hit a new high for the year. That line in the official summary says it more plainly than any leaderboard.
This round, Binance opened 58 “coin + network” pairs, not 58 coins.
Recovery Update #9, published at 11:06 on November 6, said withdrawals would open in batches starting at 12:00. Of the nine networks, BSC accounts for 28 pairs, BASE 7, ETH 6, ARBITRUM 5, OPTIMISM 4, MATIC 4, AVAXC 2, and TRX and SUI 1 each. After removing duplicates, that’s only 43 coins.
Where did the difference go? ZRO and AXL each appear on six chains, so those two coins account for 12 of the 58 slots. The same coin on different chains counts as a different asset, and each one has to be unlocked separately.
Compare that with #8 on November 5: that round covered 6 chains, 69 pairs, and 62 coins. Today’s count is down by 11 pairs, but there are three more chains—OPTIMISM, ARBITRUM, and AVAXC. Broader coverage came at the cost of fewer pairs.
Two sentences in the announcement are worth calling out. One says, “Asset recovery will be prioritized based on user demand over the past week.” The other says recovery will happen only “after all security checks are complete and the conditions for reopening are confirmed.” This schedule is based on demand and inspection results—not market cap or trending topics.
INJ is the clearest example. Its version on BSC made the reopening list on November 5, while the official timeline for INJ’s own chain is “before next weekend.” One coin, two lives—and the one that opens first isn’t the one you check every day.
The six that haven’t opened yet are HYPE, INJ, APT, PLASMA, NEAR, and LTC. The same update also says that all tokens on networks that have already been restored will be opened by next weekend.
So don’t use coin prices or hype to guess whether you can withdraw. Open the announcement, find your coin, then check which chain it’s on and whether that chain is listed.
🔍 3 million people can’t withdraw their money, and the president is taken into custody
Ueno, a bank in Paraguay, has the most customers in the country—around 3 million, or roughly 60% of the adult population.
In five days, it brought both itself and the president down.
On November 2, the central bank’s board voted to take it over. The next day, it formally entered the “resolution” process (a takeover before liquidation). What investigators found wasn’t complicated. The bank had lent $146 million to related companies, exceeding the legal limit, without enough collateral, and had booked some expenses as assets. Once the books were corrected, its capital adequacy ratio fell to less than half the legal minimum.
The fallout spread faster than the collapse itself. Payments for projects by the national power company ANDE and its employees’ pension funds were trapped in the bank; a criminal complaint was filed on the 7th. The Itaipu Binational Hydroelectric Company confirmed that $163.8 million and 115.2 billion guaraníes were also tied up there. The national pension and health insurance funds had about $380 million at stake. Total public funds: $800 million to $870 million.
How much will depositors get first? The initial cap is 10 million guaraníes, or about $1,600. Small deposits get paid first.
The political books were settled too. On the 6th, Peña resigned, and the Senate completed his impeachment within minutes. Vice President Alliana was sworn in that same day. On the 7th, a judge ordered Peña’s preventive detention on charges of breach of trust for personal gain. He had held shares in the bank’s parent company, and while in office had appointed Carvalho, head of Ueno’s predecessor company, as central bank governor. Carvalho has since resigned.
On the markets, the guaraní fell 8.2% over five trading days. According to Bloomberg, it was the world’s worst-performing currency over that period.
Why should the crypto world pay attention? Paraguay’s biggest crypto business is hydro-powered mining. HIVE’s mining site in Yguazú reached 200 MW in April 2025, 300 MW in September, and is set to add another 100 MW in its third phase, bringing it to 400 MW—all powered by Itaipu. At 300 MW, it produces around 8.5 bitcoins a day, with total investment exceeding $400 million. ANDE, which supplies power to these mines, has its own funds trapped in the failed bank.
So don’t treat this as mere political gossip. Digital banks change the front-end interface. Deposits are a bank’s liabilities; the moment they’re frozen, all you have is a place in line. Someone else holds the books. Self-custody has no “resolution” process—but the trade-off is that nobody else is there to backstop your private keys.
BTC: $75,650 to $75,774, down 0.2% in 24 hours. ETH: $2,231 to $2,235, basically unchanged. This week, Paraguay shut down the bank with the most customers in the country and got a new president, while crypto prices didn’t even flinch.
Five years, $5 billion, and not a single team sold. On November 5, LIV Golf spent thirteen minutes in bankruptcy court in New Jersey, announcing it would skip restructuring and go straight to liquidation.
First, the numbers. When it filed for bankruptcy on September 8, it reported assets of $100 million to $500 million and liabilities of $500 million to $1 billion. Revenue for all of 2025 was $208 million. Each event paid out $25 million to $30 million in prize money—more than a tenth of annual revenue at a single event. The PIF owns 98.48%. Since 2021, it has invested about $5 billion in equity.
In August, BC Partners Credit proposed a $300 million restructuring plan. It would get priority repayment and control, with players sharing whatever was left. A restructuring support agreement was signed on October 5, and the first $4 million came in. Not even a rounding error. The condition was plain: secure enough player signatures by October 25. More than 30 players signed for the 2027 season. Ten stars did not. On November 4, the FT reported that BC Partners had withdrawn. The next day came liquidation.
Why wouldn’t the players sign? The new plan offered no cash contracts, only equity in the league. For players, equity in a league still in bankruptcy proceedings is a liability.
Rahm would rather give up a roughly $100 million claim in exchange for his freedom; his contract was worth 40 times what an ordinary player’s was. Garcia, Hatton, Poulter, and Oosthuizen all left. In October, the DP World Tour also shut the door on dual membership.
The coldest detail came on August 24. Eleven of the 13 teams had player equity stakes, as high as 40%. The merger documents said this was to preserve net operating losses. The first-day bankruptcy statement listed this tax asset, which could offset future profits, as one of the most valuable assets. Two weeks before bankruptcy, the equity held by players and sponsors was wiped out, leaving behind a tax ledger. That was it.
This arithmetic plays out on-chain every day; nobody just calls it bankruptcy. A sports league’s assets are contracts, and contracts need people to sign. On-chain assets are positions, and positions need people to stay. They left. Once subsidies stop, locked value, validators, and market-maker inventory drain away minute by minute. LIV took seven months. The difference is speed.
In January, it was still hiring Citi to sell minority stakes at up to $300 million per team. Starting in August, team ownership began moving back out of the hands of the stars.
A price propped up by continual injections of capital cannot be used as a valuation. From the day the PIF stopped funding it in April, the league was selling assets. By the end, all that remained was a tax-loss ledger.
Money was never the shortfall. Signatures were. The most expensive item on the balance sheet was never in its own hands.
BTC 75,887-75,985, essentially flat over 24 hours. ETH 2,242-2,245.