The Fed votes Wednesday. It is looking at the wrong month.

Here is the fact almost nobody is saying out loud this week, and it is the whole letter.

Next Wednesday the Federal Reserve decides what to do with interest rates, and it will do it without the one number it claims to care about most. It has the markets, the surveys, the dashboards, all of it. What it does not have is a finished July, because July is not finished. The most recent full month of inflation sitting on its desk is June, the month oil fell out of bed and pulled the whole index down with it.

So the people who set the price of money vote Wednesday looking at June, while the tape in front of them trades a July that already turned.

Because the thing that cooled inflation in June spent this week climbing back over a hundred dollars a barrel, and it did not come alone. A second oil chokepoint came under attack. On top of the Strait of Hormuz, already throttled for months, the Houthis declared a blockade on Saudi shipping and put the Bab al-Mandab, the Red Sea's narrow gate, in play, claiming hits on two Saudi tankers and turning several more around. It is not a sealed door yet. It is a second lock being picked. Brent settled Thursday at $100.69, up 7% on the day and up from $88 a week ago. Then Friday it gave part of the handle back, easing toward $98 on reports that Pakistan, with Chinese backing, was pushing to restart US-Iran talks. Even after that fade, crude finished the week up about 11%.

And Bitcoin, after two months of knocking, finally cleared $65,000 on Tuesday, on the eighth run at a wall that had turned it away seven straight times. It pushed toward $67,000, and then the retest failed. By Friday it had closed back under the wall at about $64,100, the level it fought so hard to take now sitting just overhead again. That is the tell. The breakout was real, but the tape has not confirmed it, and the market is holding its breath, because next week hands it three verdicts in three days.

If you walked over from the channel, welcome. The show gives you the daily chart, the three dates and the two levels. This letter builds the weekly case. Let's go.

The referee is watching June

Start with the official scoreboard, because it is about to get tested and it is already stale.

Official CPI printed 3.5% and falling, and that number measures June. PCE, the gauge the Fed likes best, lands Thursday, and it measures June too. Both are staring at the month oil collapsed. Neither can see July, because nobody has counted it yet.

Now set the live tape beside it, with one honest caveat up front. Truflation's real-time read is not the same instrument as CPI. Different basket, different method, different clock. You do not line them up to match decimals. You line them up to read direction and timing. And the directions have split hard. The official gauge stepped down on last month's collapsing energy. The live read tagged a fresh 2026 high of 2.12% this week and sits near 2.11% now, climbing right alongside the oil it tracks. One line points backward and down. The other points forward and up. They have not crossed, but they are closing on each other, the official gauge sliding down toward the live read as the live read climbs to meet it. That is the setup: the people deciding on Wednesday are reading the line that points down, while the one that points up is the one describing today.

Oil, with two doors under attack

The reason the live read is drifting up is not a mystery, and it is sitting in the shipping lanes.

Hormuz has been choked for months. This week the Houthis added the Bab al-Mandab, the strait at the southern mouth of the Red Sea that Asia-bound crude has to thread. The workaround everyone counted on, the Saudi pipeline that carries oil east to west to a Red Sea terminal and skips Hormuz entirely, empties right at the water now being blockaded. You cannot bomb either strait back open. Brent crossed $100 for the first time since the spring on Thursday, settling at $100.69. Then Friday it slipped back toward $98 on the talks headline, and US crude eased to about $90. The handle cracked and then uncracked inside two days.

One thing I will not do is pretend I can already see this in the inflation data. The live read is rising and oil is rising, but I do not yet have the category-level proof that the spike has landed in the components, and I am not going to reverse-engineer a hundred-dollar barrel into a number that closed before it happened. Oil is the input I am watching next, not the receipt for what already drove this week. That distinction is the whole job.

And this tape lies to you all day long. Oil faked the market out twice inside one 24-hour stretch midweek, ripping on a bombing headline, dumping on a "Tehran wants to talk" headline, ripping again when nobody believed the second one. Then Friday ran the trick a third time, pulling Brent back under $100 on the talks report. Trade every one of those and you get whipsawed into confetti. The move that matters is never the candle right after a headline. It is the level that holds once the noise clears, and right now $100 is acting like a ceiling again, not a floor.

The first domino is blind

Which brings us back to Wednesday.

A hold is still the base case, priced near 62%. But the tail this letter has flagged all summer, a hike and not a cut, got a lot less theoretical this week. CME's implied odds of a July hike sit near 38%, up from about 12% a week ago. That is more than a tripling in seven days, and oil paid for all of it. The range is 3.50 to 3.75%, the committee is in its quiet period, and Chair Warsh keeps telling anyone listening he has no patience for inflation that will not quit.

So picture the room. No finished July. A hundred-dollar oil print fresh in the memory, even after Friday's dip. A chair who does not want to look soft. That is the blind spot, sitting exactly where the Fed says it looks hardest, and it is not a setup you fade for easy money. It is one you respect, keep some powder dry for, and let resolve.

The scoreboard this week, kept honest across every lane. The numbers are the point, so here they are in one place.

Crypto: buyers showed up, the verdict has not

Bitcoin's win this week was real, and it was bigger than my first read gave it credit for. On Tuesday it finally broke $65,000, a level that had turned it away seven straight times, on the eighth attempt, and it ran toward $67,000. That is a genuine breakout, and it came in a week the macro handed it every excuse to fail. Then came the retest, and the retest is where the story turned. The $65,704 line I drew on the show as the level to reclaim held as resistance instead, and by Friday Bitcoin had closed back under the wall at about $64,100. The break happened. The follow-through has not, at least not yet.

The buying under it was real too, and then it wobbled. Farside shows about a billion dollars of net ETF inflows across seven straight sessions into Wednesday. The $727 million I flagged on the show earlier in the week was not a one-off, it was the start of a genuine streak. Then Thursday snapped it with a $225 million outflow, leaving the eight-session net near $774 million. Still demand, not a rounding error, but the clean story got a little less clean overnight.

It is also not the whole case, so I will not sell it to you as one. A billion-dollar streak does not fill the far larger hole that May and June dug, roughly three quarters of ETF money is retail and retail tends to show up late, and the spot tape underneath the breakout stayed thin all week. So the honest read is not "the bottom is in." It is "buyers showed up, the wall cracked, and Friday handed the level back." A break is a headline. A close, held through next week, is the verdict. We are not there yet.

Cash-flow names did not get a pass either. HYPE slipped on the week with the rest of the high-beta board, even though its own business did nothing wrong. Real revenue is a seat belt, not a force field. In a week like this, everything trades like crypto first.

The space trade and the crowded room

One corner is squarely our lane, and it flashed a warning this week. The AI trade, the single story holding the entire stock market up, wobbled. Wednesday's mega-cap earnings landed and by Thursday the market was punishing them, Tesla down almost 15%, Alphabet down 7%, the Nasdaq off more than 2%, the VIX jumping toward 19. It landed just a week after a new Chinese model, Kimi K3, showed near-frontier work at a fraction of the cost, and the familiar bears started asking out loud again whether the American AI names are priced for a perfection they may not get. Friday the selling eased and the tape steadied, which is the reminder that a wobble is not yet a fall.

SpaceX sits inside that same trade. The stock closed Friday at $115.07, off 2.7% on the day, giving back a Thursday bounce right into its own big night. Weather scrubbed Thursday evening's Starship attempt, and the next window opens tonight at 6:45 p.m. Eastern, so by the time you read this it may already have flown. Disclosure stands: I hold SpaceX and xAI exposure through WLTH, and the receipt there is the flight, not the daily candle. The broader point is the one to keep: if the AI story ever truly cracks, do not assume crypto is the padded room. Correlations show up right when you need them not to.

The source check: read the announcement, not the ticker

This week's viral number was a token in freefall. BitMEX's token crashed more than 90%, from six cents to half a cent, and the timelines lit up with "it's over."

Read the actual announcement and the story gets duller and more useful. BitMEX, the derivatives exchange Arthur Hayes co-founded eleven years ago, is winding down and closing its doors on September 23. A token falling more than 90% on that news is not a mystery to solve or a knife to catch. It is a utility token losing the exchange that gave it utility, behaving exactly the way it should. And for scale: BitMEX, once one of the biggest names in the business, now clears only a thin sliver of derivatives volume, so this is a piece of crypto history closing, not the derivatives market breaking.

The lesson is the one we keep hammering. The screenshot is the drama. The filing is the fact. Trade the fact.

Scheduled is not sold: Hyperliquid, July 28

The one crypto event that matters most next week is one this letter has tracked for two months, and it lands the day before the Fed.

The receipt so far is almost comically boring, which is the whole point. Hyperliquid's circulating supply has sat at 222,445,714 tokens through every check since the start of June, unchanged to the digit. Two scheduled unlocks came and went and put essentially nothing new on the chain. The calendar kept promising a flood. The chain kept showing a drip, or nothing.

Tuesday is the biggest test of that pattern yet, at least on Messari's calendar: roughly 9.9 million HYPE, about $570 million at Friday's price, a little under 5% of supply, the core-contributor tranche, projected to land one day before the Fed decides. Messari is our primary source here and still shows July 28; some public trackers put the date in August, which is exactly the kind of disagreement Tuesday settles. The same monthly cliff is projected again in August and September.

We are not going to guess what it does. We wrote the checklist while things were calm, and on the day we just run it. Did the supply actually reach the chain, or stay locked? Where did it go, staked and held, or straight to an exchange? Did real buying show up to absorb it? Did the fee-funded buyback keep pace? A third straight month of nothing reaching the chain turns "watch the chain, not the calendar" into a house law with evidence behind it. A real flood arriving into a hawkish Fed finally prices the overhang, in the worst possible tape. Either way, we score it next Friday.

I still do not own HYPE. The entry I want is the one where the chain, the buyback, and the price finally agree on the same story, and a discount handed to me by a macro scare is not that agreement.

The oil call is scored, with a boundary on it. Issue 5 said watch crude before you watch anything else, and said June's one cool print bought the Fed time, not easy money. A week later crude went from $88 to $100.69, a second chokepoint came under attack, and the odds of a Fed hike went from just under one in eight to better than one in three. Friday pulled Brent back toward $98, but it still finished the week up about 11%. The method was early again. Being early does not mean the next move has to be up. It buys exactly one thing: the duty to watch oil just as hard on the way down, whenever the way down comes.

RTB closed Friday at $15.54, up on the day and about four times the $3.93 we covered it at on May 28, and Roundtable came up on the show again this week. The daily quote bounces around and always will. The receipt is not the price. It is that the product the company promised shipped in June and is live. Disclosure: I own RTB.

And the deliberately boring one, extended. Hyperliquid supply, flat to the digit, five checks running, into the biggest unlock we have watched. Nothing to celebrate yet, everything to score Tuesday.

Two threads stay open and unscored, and I will say so plainly. The inflation divergence gets its next official look at June PCE on Thursday, which is still a June number. The HYPE streak gets graded when Tuesday's tokens either move or do not. No victory laps until the tape actually prints.

Same rule as always. This letter is free, weekly, and complete on its own, because the read has to be worth your time before anything underneath it earns a dollar.

The layer beneath, the full reports with the names, the levels, and the positions I am actually holding, lives in the Data Room, and the Hyperliquid unlock-versus-buyback work that Tuesday will grade is in the Inner Circle now.

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

Here is what a week like this is really testing, and it is not your market view.

There is a study I keep coming back to, of judges ruling on parole. Early in the morning they granted it about two thirds of the time. Right before lunch, almost never. Now, the people who dug into it later poked some holes, the lunch story got oversold and the effect was probably smaller than the headline made it. Fine. The bones of it still stand, and they hold up across plenty of other work: a tired brain making its thousandth call of the day reaches for the easy answer instead of the right one. We make thousands of decisions a day, and the quality drains out of them as the day wears on.

Now drop that brain into this week. A Fed deciding on stale data. Oil faking you out three times in four days. A breakout that cracked a wall and then slid back under it. Three verdicts landing inside seventy-two hours. Try to decide all of it live, in the noise, and you will trade like a hungry judge at 11:55.

So we do the opposite. We make the decisions now, while it is quiet, and let the loud days just carry them out. The unlock checklist is already written: supply, destination, absorption, offset. The oil rule is already written: trade the level, not the tweet. And for the Fed, the rule is to separate the decision from the tantrum the market throws at the statement. The plan is not a prediction. It is a list your calm, rested self hands to your panicked, tired self, so that when Tuesday and Wednesday and Thursday show up swinging, all you have to do is read it back.

Stay liquid. Keep the powder dry. Let the three prints do the talking. I will be watching every one of them with you next week.

See you next Friday.

Kyle

Research and education, not personal financial advice. I may hold positions in what I cover, and I disclose them: I own RTB, I hold SpaceX and xAI exposure through WLTH, and I do not hold HYPE as of this writing. Figures come from Messari Enterprise on-chain and unlock data, Truflation Enterprise live inflation data, Farside Investors ETF-flow data, the Bureau of Labor Statistics and Bureau of Economic Analysis calendars, the Federal Reserve, CME FedWatch, Reuters, Al Jazeera, and CNBC reporting on oil and the shipping lanes, official BitMEX notices, SpaceX and FAA launch updates, and Yahoo Finance and Investing.com market data, as of Friday afternoon, July 24, 2026. Messari and Truflation are treated as primary sources for crypto supply and real-time inflation data, and their methods differ from other trackers. The parole-decision study is Danziger and colleagues (2011); later researchers argued the effect was smaller than first reported, and the point here rests on the broader decision-fatigue literature, not that single paper. Scheduled unlocks are not the same as distributed or sold supply. Fast-moving prices, flows, policy expectations, and launch schedules can change quickly. Do your own work. Past performance is not indicative of future results.