Bitcoin closed above $80K. Strategy bought zero.

Here is the sentence I have been waiting since May to write. Bitcoin closed a week above $80,000.

It printed on Sunday night: $80,339.13 on Coinbase, the first weekly close above the line in this entire uptrend, after four rejections. For three Fridays the only number in this letter was that one. I have typed the words “weekly close” so many times this month that my keyboard finishes the phrase for me. On Monday’s show I said what I had promised to say. By the ruler this show runs on, the regime flipped.

Here is the sentence I did not expect to write right after it. The price above the line lasted about a day.

I’ll give you the candle in a paragraph below, because you deserve the numbers and I promised them. But the price is the least interesting thing that happened this week, and I’m not going to spend your Friday on it. Three things happened underneath it that matter more, and none of them were on the chart.

The largest corporate holder of bitcoin on earth bought zero coins in the only breakout week of the uptrend, and bought back $176 million of its own preferred stock instead. The inflation number that sent hike odds to 90% this morning was hot by a tenth, and the categories behind it included wireless phone plans and airfares, not just the barrel of oil I told you to watch. And the $800 million HYPE unlock that two calendars had circled for Sunday produced an on-chain claim of about 433,000 tokens, roughly 4% of the modeled tranche, and the token made an all-time high on the day.

Those are the stories. The candle is the scoreboard.

The weekly that closed: open $77,672.10, high $82,283.00, low $76,219.18, close $80,339.13, up 3.4%. The five days since: Monday closed $79,092, Tuesday $78,447, Wednesday $78,284, Thursday $76,536.55, and Friday, after the CPI, printed $76,030 and $79,852 ninety minutes apart and was trading near $77,000 at 3pm New York. The regime flipped on Sunday’s close, and by my rule only another weekly close, under the line, can flip it back. That candle closes Sunday at 8pm New York. The daily price fell back under a day later, and each completed daily candle since has closed below it. Both things are true. Neither cancels the other. There. That’s the candle. Now the part worth your coffee.

Messari’s Bitcoin update for August, published last Friday, has a line I keep rereading. Spot Bitcoin ETFs absorbed about $3.9 billion in August, the strongest month since last October, and corporate treasuries bought a net $390 million, about 4,450 coins. The funds took nearly ten times what the treasury companies did. A year ago that ratio ran the other way. Messari’s flow model also shows the ETF regime flipped to accumulation on July 30, three weeks before the August 19 breakout, and has pulled in about 53,000 coins since. The funds led the price. The squeeze accelerated a bid that was already there.

That’s the setup. Here is the week.

Strategy, the reason the phrase “corporate bid” exists, bought zero bitcoin between August 31 and September 7. Its September 8 filing shows no coins acquired, no stock sold, and $176 million of existing cash spent buying back 1.81 million shares of its STRC preferred at about $97.36 a share, on paper with a $100 face. It expanded that buyback authorization to $2 billion the same day. Strategy spent the only breakout week of the year buying Strategy. Messari’s August report had noted that “Strategy has started to accumulate again,” and for August that was true. The next reporting week, the one with the breakout in it, it bought none.

The boring explanation is the one I find persuasive. Strategy buys bitcoin with money it raises by selling preferred stock at or near par. When the preferred trades under par, selling more of it is expensive and buying it back is cheap, so the machine runs in reverse: it buys its own paper instead of coins. That is not a scandal. It is arithmetic. But it is arithmetic that switches the corporate bid off exactly when its paper is out of favour, which has a way of being exactly when you’d want the bid on. Strive’s preferred trades at par, which is consistent with Strive buying 1,375 coins at $79,281 the same week, a price we are now below. Read it as a story about two preferred stocks if you like. It still means the corporate bid is conditional on the price of its own funding, and that price was under $100 during the only breakout week of the year.

Then the other bid went quiet. The funds took $986.7 million in the week the candle closed. Then Tuesday, minus $46.6 million. Wednesday, minus $120.2 million. Thursday, minus $282.7 million, ARKB alone shedding $164 million. Three red sessions, $449.5 million out, the first back-to-back outflow days since mid-August, about 12% of the strongest three-week run of the year handed back. Messari’s test for September was whether the accumulation regime “survives a Fed that cannot justify moving in either direction,” with confirmation requiring “inflows persisting through a hawkish hold.” The market is now pricing a hike, not a hold, and the regime is running its first test under harder conditions than the one Messari wrote.

Glassnode drew the ceiling those two bids are standing under: about 1.07 million coins bought by long-term holders between $83,000 and $86,000, a shelf of short liquidations at $82,000 to $86,000 that has thickened 21% since the squeeze, and the ETF complex’s own break-even near $86,000, which it has closed below for 228 straight sessions. Corporate treasuries, by Glassnode’s model, break even near $80,500. Spot stopped short of the band on September 3 and has not been back.

So the week’s price story has a mechanism after all, or at least a mechanism I’m watching. The two bids that carried August were a treasury company whose funding traded under par and a fund complex about to be tested by a hike instead of a hold. The candle showed the retreat. The filing and the flow table show two sources of demand weakening beneath it. I can’t prove the second caused the first. I can tell you it’s where I’m looking.

Last Thursday Governor Waller handed the market the sentence this morning was built for. From the Fed’s published text: “If there is evidence that progress toward 2 percent inflation reversed in August, a small adjustment in our stance would help ensure that it resumes.” The August inflation report he meant is the one that printed at 8:30 this morning. He wrote the question a week early and told everyone where to sit.

The week between the question and the answer ran one way. The ECB raised its deposit rate to 2.50% on Thursday, effective the 16th, its second hike since the Iran war began, exactly as all 65 economists in Reuters’ survey expected. Producer prices rose 0.4% in August, in line, and 5.4% on the year, a tenth hot; excluding food and energy the core rose 0.2% against 0.3% expected, with more than a third of the goods increase from diesel, up 24.1% in a month. The Treasury said it would buy up to $6 billion of its own long bonds, triple the usual size, then bought $5.19 billion, about half of what dealers offered it, and the ten-year rose anyway, to 4.95%, its highest since 2023, which is the bond market’s way of saying thanks, we’ve got it from here. Brent touched $108. Saudi Arabia told OPEC its output had fallen to 6.24 million barrels a day, the lowest since 1990. By Thursday afternoon CME priced a September hike near 70%. A week ago it was 58%.

Now the ruler, because this is the week it got graded. Truflation’s call for this morning, published Thursday: 0.3% on the month against a Street 0.4%, 3.4% on the year, core 0.2% and 2.4%. Inside it was the argument I was ready to make: gasoline up 27.6% on the year, housing down 2.32% on the year, much of the rise is oil and, in their words, “may be seen as temporary.” I was going to tell you a tenth was the whole fight and that a rate hike does not pump crude.

At 8:30 the tenth went the other way on both monthly lines. Headline 0.4% and 3.4%, exactly the Street. Core 0.3% against 0.2%, a number only nine of the 73 economists in Bloomberg’s survey had, and 2.4% on the year, the smallest annual core reading since the spring of 2021. Truflation’s annual calls landed on the number. Both of its monthly calls were a tenth light. The barrel was not wrong either; gasoline was still more than a third of the monthly headline. It was incomplete. Core pressures in the month included wireless phone plans, airfares, used cars, education, and a 25.4% annual rise in computer software, the largest on record, which Bloomberg pinned on the AI boom. The pressure I said lived in a barrel of oil turned out to have a second address: your phone bill. I have been outrun by a lot of things in this business. This is the first time it was a wireless plan. Unrounded, the core was 0.29%. Five hundredths less and every headline this morning reads 0.2, and every chyron you saw today says hold instead of hike. Hike odds went from 72% to 90% inside the hour. That is the market’s judgment. The committee’s comes Wednesday, and a rounded number does not cast its vote.

Then the market did the thing that should stay with you into Wednesday. With the hike all but priced, stocks rallied more than 1% with breadth, oil fell four dollars, the ten-year backed off from a hair under 5% to 4.91%, and Bitcoin dropped to $76,030, ran to $79,852 just after ten, and by three o’clock had given most of it back. A hike everyone can see turned out to be cheaper than one nobody could price, for everything except Bitcoin, which took the relief and could not keep it.

For four Fridays this letter has been watching a HYPE unlock that keeps moving. Two trackers said Sunday, September 6. Messari’s calendar said the 28th. Sunday came, and HYPE printed a new all-time high of $89.60 that day. So what was actually claimed?

About 433,000 tokens, worth about $37 million, by Tokenomist’s count of on-chain claims that day. The two calendars had 9.92 million circled. That is about 4.4% of the modeled tranche, on a day the token made a high. Vesting, claiming and entering circulation are three different steps, and the calendars only model the first.

The reason is on Messari’s own unlocks page, and I want to quote it exactly because it is the whole story. Under the core contributor allocation of 238 million HYPE, the vesting description reads “12-month cliff followed by an unknown vesting schedule,” and under Assumptions it reads “2 year monthly linear rate.” Two hundred and thirty-eight million divided by twenty-four months is 9.92 million. The Hyper Foundation published a one-year lock and a broad completion window; it did not publish a monthly schedule. So the unlock the market trades around every month is a spreadsheet’s assumption about a schedule that was never published at that resolution, and contributors claim what they claim when they claim it. As displayed on September 11, Messari’s overview page lists the next one as September 29, $813 million, and Messari’s unlocks page, one click away, lists September 28, $792 million. Tokenomist says October 6. Four dates, three trackers, one assumption. I have group chats that settle on a dinner time faster.

And the circulating count is not settled either. Messari and CoinGecko show 222 million circulating; Hyperliquid’s own interface showed about 299 million on August 27. Those two numbers measure the same thing and sit 76 million apart, and every “supply overhang” you read about HYPE is built on one of them.

Here is the number that matters more than any of that, from the same Messari page. Hyperliquid’s gross revenue fell about 43% from the third quarter of last year to the second quarter of this one, from about $357 million to about $202 million, because builder-run markets grew from about 2% of volume to about half and builders keep up to 50% of their fees. Ninety-nine percent of the protocol’s share goes into the buyback fund, so quarterly buybacks fell about 49% over the same span, from about $290 million to $149 million. That is the real supply story. The unlock is an assumption. The buyback is a receipt, and through June it was shrinking.

Arthur Hayes walked through exiting HYPE on my show last week. I hold none. I have stopped predicting the unlock and started reading the fine print, which is where the story was the whole time.

The close. For three Fridays the only number I gave you was $80,000 on a weekly close. It printed $80,339.13. Cashed. The same rule says one weekly close back under it decides the other direction, and I will print that too if it comes.

The note. The research report I promised last Friday went out to members on Tuesday. Promise kept.

The flows and the price. Two weeks of disagreement, one week of agreement. Three weeks in, the flows tell you the direction of the bid and nothing about the direction of the price. I’ll keep reporting the shape, not a rule.

The number is $80,000 on the weekly and it works both ways. It flipped on Sunday’s close and it flips back only on a close under, and that candle closes Sunday at 8pm New York, after this goes out. The week’s low is Friday’s $76,030, printed in the CPI minute. On the shorter frames I’m watching my orange weekly trend line, which Thursday’s four-hour closed under and Friday’s spike could not close back above. Glassnode’s ceiling is $83,000 to $86,000; its old floor at $62,000 to $65,000 has thinned while a newer band of buyers formed between $76,000 and $82,000, which is where we are standing. Fear and greed printed 69 Thursday and 56 Friday. Ben Cowen’s odds from last week stand, 65% the low is ahead, 35% it is behind. I am not calling either.

Now my trade, because I showed it to you. The leveraged Bitcoin long I opened on BloFin on August 31, on the first Compass buy of this uptrend, was a small position with its entry, its stop and its size written down that day. Its exit rule is the four-hour close, which is a different instrument from the weekly regime, and Thursday’s four-hour closed red. On Thursday’s show I told you what I was doing about it: raising dry powder, with the Compass’s shorter signals turning bearish under me. I said it there first because that is where I showed you the trade, and the next change gets said there too. By my own count the Compass loses about two trades in three. A small position and a written stop are what that number is for.

Two disclosures belong here and not at the bottom. The trade is on BloFin, an affiliate partner of the channel, and I read their deposit promotion on Monday’s show. Weigh the second when you hear the first.

Liquid. On Sunday somebody drained about 4,000 of the roughly 4,200 bitcoin backing Blockstream’s Liquid sidechain, about $320 million, through a bug that let them mint L-BTC that did not exist and peg it out for the real thing. They returned about 3,400 coins on Monday after the patch, kept about 598 as a self-declared bounty, a phrase doing a great deal of work for $47 million, and Bitcoin’s base layer was never touched. By Thursday block production had restarted and SideSwap’s markets had reopened, with peg-ins and peg-outs still closed. Ninety-five percent of a federation’s reserves in one transaction, on a sidechain that had run quietly since 2018.

The register. On September 1 the SEC proposed letting a blockchain serve as the official record of who owns a security, or part of it. It would not by itself make a token the share, and it does not explain last week’s two prices for one stock. It does mean the ledger the transfer agent keeps and the ledger the chain keeps could be the same ledger. Comments close November 3. On the 17th the SEC hosts a roundtable on 24-hour stock trading with BlackRock, Nasdaq, NYSE, Robinhood, Citadel and Jane Street at the table.

CLARITY, from someone who helped write it. Tyler Williams helped draft GENIUS, helped shape CLARITY and worked for Bessent at Treasury, and he sat down with me for Friday’s show. His read: Tuesday’s cloture vote is a door, not a destination, and it needs 60 votes just to open debate; if it opens, the bill can move faster than people expect, with three areas still likely to change on the floor; and Senator Lummis has warned that a failure now could sink it for the rest of the decade. The detail that stuck with me: 21 of the biggest banks in the world, Goldman, Bank of America and Citi among them, are now building a joint stablecoin after years of fighting the ones that already exist. Pushback became imitation. The House leaves Thursday until after the midterms. Polymarket has law-this-year near 15%, from 82% in February. I’ll cover what happens; I won’t pretend to know.

The calendar. Sunday, the weekly close. Monday, the House returns; Solana Summit in DC. Tuesday, the CLARITY cloture vote and day one of the FOMC. Wednesday, the Fed decides, with a dot plot. Thursday, the SEC’s 24-hour trading roundtable; the House leaves; the Bank of Japan meets Thursday and Friday, hike expected. Sept 28 or 29, the next HYPE unlock on Messari’s two pages; Oct 6 on Tokenomist’s. Sept 30, PCE, the MSCI deadline and the Evernorth vote. Oct 3, the first AQAv2 reserve-yield payment into Hyperliquid’s Assistance Fund, for buybacks, not a distribution. Oct 5, Ethena’s final investor release and the start of my bottom window. Oct 14, September CPI. Oct 16, MSCI. Oct 20, comments close on the SEC’s crypto rule. Nov 3, comments close on the register, and the midterms.

Here is what I think this week was actually about.

A rule I wrote three weeks ago fired on Sunday. The price was back under the line a day later, and by Thursday’s close it was more than 4% under. Those are two different things and I intend to keep them different. By my rule, only Sunday’s close can flip the weekly back. If I graded it early because the tape hurt, it would not be a rule. It would be a mood with a chart attached.

What I got wrong this week was more useful than what I got right. I told you to watch the barrel, and the tenth that mattered came from a phone bill. I have been wrong in more expensive ways, but rarely in a funnier one. I told you Messari’s calendar had the unlock on the 28th, and it turns out the unlock is an assumption on a page that says so in plain text. I told you the flows were the floor, and the floor and the price went down together. None of that changes the plan, because the plan was never “be right about September.” The plan is a written rule for the regime, a written stop for the trade, and a quality list I keep adding to while other people argue about candles.

And the bid. The two buyers that carried August were a treasury company whose funding traded under par and a fund complex that is about to be tested by a hike instead of the hold Messari wrote its test for. STRC’s quote will tell you whether the machine can run forward again, and the next filing will tell you whether it did. The Farside table will tell you whether the regime survives Wednesday. Those two things will tell you more about the next weekly close than any golden cross.

Stay liquid. Keep the quality list growing. Let the candle close.

See you next Friday.

Kyle

This letter is free, weekly, and complete on its own, because the read has to earn your time before anything underneath it earns a dollar.

Out here is the read. In there is the research note that went out Tuesday, and what I do about it.

No rush from me. Read a few more Fridays, then decide.