One cent, then one close, then one speech.

On Monday, Bitcoin got to $79,999.99.

One cent. That is not a rounding and it is not a joke. Somebody's algorithm stopped a single penny short of the number this entire market has been staring at since May, and then the whole thing turned around and went the other way.

I have been laughing about it all week. Here is how the week ended.

Tuesday it went through properly, ran to $81,265, and closed back under. Wednesday, under. Then on Thursday it finally did it: a daily close at $80,275.34, the first close above eighty thousand since May. At 01:00 UTC on Friday it printed $81,479.50, the highest Bitcoin has traded in this entire move.

About thirteen hours later, Kevin Warsh stood up in Wyoming and talked for forty minutes.

By lunchtime in New York, Bitcoin was $76,845.71.

High to low, $4,634, in about fifteen hours. As I write this on Friday evening it is around $77,700, which is roughly $2,300 below the line it spent all week trying to hold, and slightly lower than where it closed last Friday.

A week that felt enormous, ending lower than it started, having touched the number twice and given it back both times.

And that arc is not even the interesting part. This week every number that mattered came back with two different answers depending on where you pulled it from. Opposite conclusions, from good sources, about the same market, in the same window. It happened six times, and one of them put roughly a billion dollars of tokens on the calendar that never moved this morning.

One: a daily close is not a weekly close

Eight days ago, with Bitcoin at $72,100, I put a number in writing on Thursday's show. The weekly regime is bearish and it does not flip until we close above the money line, which sits at about $80,000.

That sentence has one word in it doing all the work, and this week is the reason I wrote it that way.

On Wednesday, with Bitcoin having wicked to $81,265 and sold off, I said the compass does not act on wicks, it acts on closes, and we had closed below the line. Fair enough.

Then Thursday closed at $80,275.34, and I want to be straight with you about what that is and is not. It is a real daily close above the level, the first since May, and it was worth something. It is not the close I named. The money line is a weekly regime line. The weekly candle containing all of this does not settle until Sunday night, and Bitcoin is trading about $2,300 underneath it as I write.

So the scorecard as of Friday evening: one daily close above the line out of seven, immediately reversed. Zero weekly closes. No buy signal printed.

I am not going to pretend that is satisfying. What I will point out is what it cost the people on the other side. Anyone who bought Tuesday's wick at $81,265 is down about 4% tonight. Anyone who bought Thursday's close at $80,275 is down about 3%. Anyone waiting on the weekly is flat and still waiting.

The rule paid twice in five days. That is not skill. That is just what happens when you write the number down a week early and then do not move it.

Two: the leverage came off before the drop, not during it

Everybody is arguing about whether last week's squeeze left this market loaded or empty. There is a number for that, and the answer depends entirely on which version you pull.

During the squeeze, open interest went up. Bitcoin futures OI rose 7.18% on the day of the flush, and I read that last week as a market re-levering into strength. Here is what the week since did to it.

Bitcoin futures open interest, measured in coins, has fallen from 645,760 BTC on August 14 to 587,584 on August 25. Down 9%, a five month low. Crypto-margined open interest is at an all-time low near 52,000 coins, about 11% of the total. Funding sat below 10% annualised, which is barely awake.

That 7.18% dollar rise was real. What it did not prove was that traders had added 7.18% more risk. Over the fuller window, the coin series fell 9%. That is a market de-grossing into strength, and it means the simple fuel-reloaded read I gave you was too strong.

That lighter derivatives backdrop helps explain why today was a flush rather than something larger. Roughly $488 million of positions were liquidated in twenty four hours across about 97,000 traders, running three to one against the longs, with over $200 million of that inside the single hour after Warsh spoke. That is a real cascade, and it was overwhelmingly a long-side event: more than $360 million of the $488 million came from longs. The market took a hit without the kind of leverage build I described last Friday.

Now the part worth keeping, because it explains how both readings can sit on your screen at once. Over those eleven days Bitcoin repriced about 25%. Run the arithmetic: a 9% decline in BTC-denominated open interest can still show up as a rise of roughly 14% when translated into dollars. Same aggregate, same window, two units. One ruler can look like added risk. The other shows less BTC notional outstanding. Neither number has to be wrong.

There is no warning label on a dollar-denominated figure in a week the collateral moved 25%. It just sits there being accurate while the conclusion goes the wrong way. It did that on four other screens this week.

Three: nine straight sessions of buying, and a lower price

The spot Bitcoin ETFs have now taken money on nine consecutive trading sessions, August 17 through August 27, totalling $3.04 billion. Thursday alone added $242.3 million. Ether ETFs ran the same nine sessions for $1.41 billion, and Thursday's $225.8 million was their biggest day of the month.

At this Friday timestamp, Bitcoin is slightly below where it closed last Friday.

Read those two paragraphs together, because that is the most useful thing on this page. Last week I made a big deal of the flows leading price by two sessions. They did. This week the same flows, bigger and steadier and running nine deep, led nothing at all. The flows were never the trigger. They were the floor, and today we found out roughly where the floor is.

And then there is the thing I found on Sunday night that I have not been able to put down.

Morningstar runs a study called Mind the Gap. Across roughly 23,000 funds it measures one narrow thing: the difference between what a fund earns and what the people in the fund earn. Across everything, that gap is about 1.2 points a year.

For the spot Bitcoin ETFs, from launch in January 2024 through the end of June, the funds returned about 8.5% a year and the average dollar invested in them lost about 5.8% a year.

Same funds. Same period. Fourteen points apart.

Honest note, and it is load-bearing. In May, four academics published a rebuttal in the Financial Analysts Journal arguing Morningstar's gap should not simply be read as bad timing, and putting the actual timing cost nearer a tenth of a percent a year. The criticism bites hardest exactly here: a fund that launched in January 2024 and grew enormously into a rising market prints a huge gap from the shape of its asset growth, which is arithmetic rather than a character flaw. The divergence is not in dispute. The explanation is.

For whatever it is worth, three billion dollars walked into these funds over nine sessions, most of it after a 24% move rather than before one, and the Friday-to-Friday comparison is red. I am not going to tell you what that proves. Draw your own line and I will give you mine at the end.

The billion dollar unlock that did not show up, and the trackers that called it. Last night four sources disagreed about a HYPE unlock scheduled for today. Messari's dashboard said 9.9 million tokens. Messari's own AI said 9,916,666 and put it tomorrow. CoinGecko, echoed by Bloomberg, said 14,175,778 tomorrow, worth about $1.2 billion. Tokenomist said there was no August event at all. DefiLlama said "no events."

No August 28 unlock shows up in the supply data. Hyperliquid's circulating supply tonight is 222,445,714, unchanged, to the token, for a fifth straight month. DefiLlama's last recorded unlock is 433,024 tokens on August 6.

And here is the part I enjoyed. CoinGecko's page tonight says the next unlock is 9.92 million tokens on September 6. Yesterday the same page said 14,175,778 tokens on August 29. The date moved by eight days and the size dropped by 30%, with no notice, after Bloomberg had already run the old figure. Messari, for its part, had the size roughly right all along and the date wrong by nine days.

The lesson is not that anyone is bad at their job. It is that roughly a billion dollars of disagreement sat between four reputable sources for a week, and the cleanest check was whether the supply number moved. It did not. If you traded that event, you traded a number nobody could verify.

We said we would tell you which one revised. Last Friday Farside printed $10.3 million for Thursday's ether ETF flow and SoSoValue printed about $221 million. Farside's table now reads $219.5 million. SoSoValue was right, and the likely mechanism is the satisfying part: that day's breakdown lists two separate BlackRock ether products, ETHA at $173.3 million and ETHB at $35.9 million. ETHB is the iShares Staked Ethereum Trust, launched in March, and an easy column to miss. One likely missing column, one twentyfold gap, and no revision log we could find.

We named Monday as the risk and Monday decided it. Last week's radar said that if Bessent reloaded the energy base, the whole disinflation schedule moved. He went Monday: roughly sixty designations, five sectoral determinations, and the thing everyone was watching for did not happen. No major foreign bank and no country was itself designated. Bessent said the quiet part on the record: "Why would I want to blow up the global financial system?" Brent's expiring October contract settled at $89.30, down 5.4% on the week, with the new November front month at $88.29. Honest note: month over month our Truflation feed still rose, and gasoline on it has not moved at the pump. One week down does not undo a month up.

One update on last week's leverage read. I called the open interest rise re-levering. The coin series over the fuller window says de-grossing, and the section above has the numbers. The direction of the tape was right. The word was wrong.

The number has not moved. $80,000 on the weekly, and it flips on a close. Support $70,000, resistance $74,000 now well beneath us, four hour invalidation below $68,146. No buy signal has printed.

One number to keep everyone honest while the calls get louder in both directions. Bitcoin's realized price, the on-chain cost basis of every coin in circulation, sits near $53,000 and has barely moved in six weeks. At $77,700 this asset trades roughly 47% above that aggregate on-chain cost basis. Every bear market low in Bitcoin's history has traded at a discount to that line.

And now the part I have to put in, having spent this whole letter explaining the difference between a wick and a close.

On Wednesday's show, about two minutes after refusing to call a wick a close on Bitcoin, I flagged the Coinbase premium as a bottom checklist dial that had "just blinked." It had. It had been negative for 97 straight days, the longest run on record, and on Monday it went positive.

It was back below zero the next session.

That is a wick. I called it a signal, on camera, minutes after telling you not to do exactly that. I would rather hand you that than have you find it yourself, because it is the most useful thing in the issue: the rule is easy to write down and genuinely hard to run, and it is hardest to run on the screen you most want to be right about.

The dog that did not bark in Wyoming. The symposium was themed "Financial Innovation: Implications for Payments and Policy." The Kansas City Fed's own announcement named cryptocurrencies and stablecoins among the innovations in scope. The program included a paper called "Innovation in Tokenized Finance."

In about forty minutes, Warsh did not say stablecoin, blockchain, digital asset or tokenization once. Every use of "token" or "tokens" in the body of his remarks referred to AI inference, including a line about the two leading labs annualising more than $100 billion in token sales. The only explicit cryptocurrency reference in the document is a footnote citing an essay he wrote in 2022. He took a conference about payments innovation and spent his keynote on AI, forward guidance and the conduct of Fed policy. Everyone in this industry spent the week waiting to be mentioned. We were not.

What he actually said, since it moved your portfolio. Two lines carried it: "I would be hard pressed to describe broad financial conditions as restrictive," and "We must be confident that underlying inflation is moving to our objective, clearly and at sufficient speed. Otherwise, we have work to do." He never named September and never said he intended to raise rates. He mentioned the FOMC only while describing the July minutes, not as guidance for the next meeting. The hike story is inference, not guidance, and it is worth holding that distinction because the market did not: September hike odds went from about 35% to roughly 60% on CME, and from 31% to 50.5% on Polymarket, inside a few hours.

The honest counterweight on this morning. $6.4 billion of Bitcoin options expired at 08:00 UTC, six hours before Warsh spoke, with max pain around $68,000 to $70,000 against spot near $80,000. That expiry removed a large block of options positioning and the gamma hedging around the $75,000 and $80,000 strikes before the speech. So: Warsh was the obvious accelerant, and the timestamps support that reading. He is not the only thing that happened this morning, and after is still not because.

Solana voted, and the close one passed. SGP-0001 and SGP-0002 carried, SGP-0003 failed. The one that matters doubles the disinflation rate and pulls the terminal issuance floor forward from 2032 to 2029. It cleared the two thirds threshold by about a third of a percentage point.

The calendar. Sunday, the weekly close, and it is the only thing on this list I will be watching. Sept 4, payrolls. Sept 6, the HYPE unlock that is actually scheduled. Sept 9 and 10, the first enlarged Treasury buyback. Sept 11, CPI. Sept 15, the CLARITY cloture vote. Sept 15 to 16, FOMC with a dot plot, now genuinely live. Oct 16, MSCI. Oct 20, comments close on the SEC rule.

Warsh opened the conclusion of his speech with a sentence I would have paid for.

I have spent this whole letter, and both shows this week, making that argument at a much smaller scale. A wick is a decision. A close is a discipline. Whether you feel like the bottom is in is a decision. Whether the weekly candle closed above a number you wrote down eight days ago is a discipline.

And then go back to the Morningstar number, because it is the only thing here that is actually about you.

The funds made about eight and a half percent a year. The average dollar inside them lost about five point eight. Morningstar says that gap is timing. Four academics say most of it is arithmetic. I do not have to settle that, because the undisputed part is the part that should keep you up: the return the asset produced and the return the people holding it captured were not the same number, and they were not close.

Every one of the six is a version of that. Coins or dollars. A daily close or a weekly one. Farside or SoSoValue. Messari's dashboard against Messari's own AI, arguing over a billion dollars of tokens that did not move. A 99.93% accuracy claim on the same website as a forecast that missed by twenty basis points.

You can outsource almost all of those. Pay someone to pull the right series, read the right filing, catch the missing column. That is most of what this letter is for.

The one you cannot outsource is the ruler you use on your own account. Nobody publishes your personal investor return. No dashboard, no second provider to disagree with, no revision log. It is the one number in your life with exactly one source, and the reason Morningstar can run that study at all is that almost nobody checks it.

So here is where I am tonight. The macro case did not get worse this week. It got harder, which is different. Oil fell more than 5% on the week, the disinflation feed turned back down on the exact variable we said would decide it, and the money arrived for nine straight sessions. Against that, a Fed Chair just told the market financial conditions are not restrictive, and the front end believed him inside an hour.

The trigger has not fired. We got there twice, closed above it once on a timeframe I did not name, and gave it all back by Friday evening.

The weekly closes on Sunday. Between now and then I am not doing anything, which is the whole point of having written the number down on the 20th.

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 I can show you that the line is $80,000 and that it flips on a close. In there I can show you the size, the entry, and what I actually do at both ends of Sunday night.

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