On the last night of July, someone emptied roughly 500 bitcoin wallets in about 25 minutes. The confirmed haul was 594.5 BTC — about $38.3 million. The wallets were Coldcards: hardware devices, kept offline, bought by people who did everything they were told to do.

The cause was not a phishing link or a leaked seed phrase. It was a build error shipped in Coldcard 4.0.0 in March 2021. The firmware quietly stopped using the device's hardware random number generator and fell back to software key generation seeded from non-secret chip data. On the Mk3, that collapsed seed entropy from 128 bits to roughly 40 — brute-forceable. Mk4, Q and Mk5 land near 72. Five years of keys were weak the moment they were created, and no firmware update repairs a seed that already exists. Estimates of the full campaign keep climbing: Galaxy Research puts it near 1,367 BTC, roughly $89 million. Treat every number above the confirmed sweep as still moving.

Then the second pillar went. Strategy reported Q2 on Thursday and disclosed that it sold bitcoin — 3,620 BTC across the first seven months of the year — to meet preferred dividend obligations. It is a rounding error against 843,775 coins, and the reason is mundane plumbing. But the company built its entire identity on never selling, and now it has. Its stack is underwater by roughly $8.9 billion against a cost basis near $75,476, and it has not bought a coin in five weeks.

Self-custody has a software supply chain. The permanent bid is not permanent. Both of those became demonstrably true in the same seven days.

And bitcoin barely moved. It broke $62,500 intraday Friday — the low printed $62,344, roughly 48 hours after the Fed — then closed the week near $63,600 and sits around $63,400 this morning. It still finished July up roughly 7%, through a hawkish Fed, an AI-led equity selloff and a nine-figure exploit. The Fed itself held at 3.50–3.75% on a 9–3 vote, with all three dissenters wanting a hike now, and the 30-year yield closed Wednesday at 5.193% — its highest since 2007.

That is the tension in this issue. The two load-bearing beliefs of the bitcoin thesis took real damage, and the asset shrugged. Either the market has not finished pricing it, or the thesis was never what we said it was.

Three metrics. Here is the read.

Metric 1 — Fear & Greed: 28. Fear, and functionally unchanged. This index sat at 27 before a hawkish hold, a $38 million theft, a broken floor and the end of Strategy's never-sell doctrine. It sits at 28 after all of it. That is not calm. That is an index measuring a crowd that has already stopped reacting — and a market with nothing priced in is a market that moves on the next real surprise, in whichever direction it arrives.

Metric 2 — ETF Flows: net zero. Three sessions after the decision: $32.1 million in Wednesday, $233.1 million in Thursday, $265.4 million out Friday — the largest single-day withdrawal in weeks, led by IBIT. Net across the three days: negative $0.2 million. A hawkish Fed, a cool inflation print, a GDP miss, an $8.2 billion loss at the largest corporate holder and a hardware-wallet exploit, and the institutional complex finished exactly where it started.

Metric 3 — BTC Dominance: 56.3%. The corporate bid now has a trigger price, and it is not bitcoin's. Strategy's CEO said the objective is for STRC to trade at $99 to $100, and that when it returns to par the company issues more and buys more bitcoin. STRC closed Friday at $89.46. Until that gap closes, the largest single buyer in this market is a spectator — and the number that restarts it is a yield instrument's price, not a bitcoin level.

The full board — including two Fed positions we closed at a loss this week — is in the POLYMARKET STACK.

📣 THIS SPOT IS OPEN

Baseline Crypto reaches 7,000+ Bitcoin-focused readers every Monday and Wednesday morning. Direct sponsor placements are open — premium position, top of the issue, one per send.

Reply to this email for rates.

Concede the real part, because it is genuinely good news. June core PCE printed 3.3% year over year and just 0.1% on the month. Headline PCE actually fell 0.1%. The Fed held. If you had been handed those three facts in isolation on Tuesday, you would have called the top in this tightening scare and you would have had every right to.

Now look at what the market did with them. Polymarket priced the July hike tail near 20% going into Wednesday. As of this morning it prices a 25 basis point increase in September at 57.5%, against 40.5% for no change, on more than $11 million of volume. The odds of a hike went up, and they went up in four days without a single new inflation number arriving. Whatever repriced that market, it was not data.

It was structure. Warsh has abolished forward guidance. The statement is now three short paragraphs with no bias language of any kind — no "additional firming," no tilt, nothing. So the thing everyone was waiting to read did not exist, and the absence of hawkish language is not dovishness. It is silence by design.

With that channel closed, the signal moved to two places. The first is the vote: 9–3, with Hammack, Kashkari and Logan all dissenting in the same direction, all wanting a hike immediately. The second is the reaction function, which Warsh simply read aloud — that when a policymaker sees underlying inflation moving higher, he or she is more inclined to tighten. That is an asymmetric rule stated in the open. Asked directly whether June's cool inflation drove the hold, he answered: "In two words, not much." Asked whether "pause" was fair, he said he would not characterize it as anything like a pause.

Here is the non-obvious part. If the September repricing came from learning the reaction function rather than from learning about inflation, then the next repricing will not come from CPI either. It comes from labor, because a Fed that discounts a cool inflation print has told you inflation prints are not its binding constraint. That is why this week — ISM Monday, JOLTS Tuesday, ADP and ISM Services Wednesday, and July payrolls Friday at 8:30 — matters more than any inflation release between now and September.

So, dated and gradeable. If Friday's payrolls come in soft and September hike odds are still above 50% at the close, this is a Fed tightening into a visible slowdown, and $61,400 does not hold. If those odds fall back under 50%, then the last four days were a press conference echo and the hard part really was behind us. We will grade it Wednesday either way, and we will not move the goalposts.

Three numbers that define the week ahead.

$100. Not a bitcoin level — STRC's par value, and the switch that restarts the largest corporate bid in this market. It closed Friday at $89.46 against a 12% coupon, an effective yield near 12.77%, with a ratchet that adds half a point if it trades under $95. Management has now said plainly that par is the condition for issuing more and buying more bitcoin, and Saylor's own arithmetic points at roughly September 8. Watch this ticker the way you would watch an ETF flow print: it is the cleanest public read on when 843,775 coins stop being a static pile and start growing again.

$61,400. The first real support under Friday's break. Bitcoin is below its 20-day moving average inside a descending channel, with $59,070 behind that and the bear-market low of $58,000 from July 1 behind that. Overhead, $64,567 is the immediate ceiling and $65,000–65,500 is the supply zone where every rally this month has died. The constructive read is that Friday's break at $62,344 was intraday only and the week still closed above it. The honest read is that the level we called the last step standing is now the level we call resistance-adjacent, and the burden of proof has flipped.

Friday, 8:30 AM. July non-farm payrolls, and the single print most likely to set September. The whole week builds to it — ISM manufacturing Monday, JOLTS Tuesday, ADP and ISM services Wednesday, jobless claims and unit labor costs Thursday. Watch unit labor costs in particular: Q2 GDP came in at 1.5% annualized, down from 2.1%, with the gross domestic purchases price index at 5.7%. Slowing growth against a deflator like that is the stagflationary combination that drove the 30-year to its highest yield since 2007, and payrolls is the number that confirms or kills it. Capital Economics is looking for about 130,000; treat that as one house's estimate rather than the street.

Forget analyst predictions. Polymarket is a real-money prediction market — traders put actual dollars on outcomes. Scorecard first, then the board. Not financial advice. Our read. Disclosure: we hold personal positions in Polymarket itself and may earn a commission from Polymarket referrals. These markets are thin and the odds below move — direction over ticks.

Scorecard, and it is a bad one. Start with the test we dated on Wednesday: soft energy language and $62,500 holds the week; explicit hike bias retained and the floor gets tested within 48 hours. Here is what actually happened. The energy and supply-shock language stayed in the statement — it did not soften. There was no hike bias to retain or drop, because Warsh has removed forward guidance entirely, so that half of the test was unanswerable as written. And the market did what the hawkish branch described: $62,500 broke intraday Friday at $62,344, roughly 48 hours after the decision. We got the direction right and the mechanism wrong. We wrote a conditional test against a statement structure that no longer exists, which is our error and not the market's. No credit claimed.

The venue gap resolved in our favor and we did not own it. We called no-trade on CME futures pricing the hike tail near 35% against Polymarket near 20%. The Fed held, so Polymarket was the better-priced venue and the read was correct — but a correct read with no position is an opinion, not a win, and we are not scoring it as one.

Two Fed positions closed at a loss. Our stated exit on the zero-cuts short was explicit: cover above 88%. It prints 88.75% this morning, so we are out — shorted at 78%, trimmed at 85%, covered the rest here. The December-cut long goes with it, bought near 18% and sitting at 12%. The post-mortem is worth more than the loss: the thesis was that a cool inflation print would crack the no-cut consensus. Core PCE did come in cool, at 3.3%. The consensus got stronger anyway. We were not wrong about the data. We were wrong about the reaction function — and Warsh described that reaction function out loud on Wednesday, which means it was knowable and we did not act on it fast enough. The book is now close to flat by design. After a week like this, that is the position.

Zero Fed rate cuts in all of 2026 | Market: 88.75% Yes ⚫ EXITED — the rule fired, so we followed it
Shorted at 78%, trimmed at 85%, and the stated cover level was 88%. It trades at 88.75%, which means the trade is closed at a realized loss rather than held on hope. The temptation here is obvious: three cuts still price at barely 1%, the whole ladder looks like a crowd that has stopped imagining any other world, and that is exactly the kind of consensus we like to fade. But we published an exit and the exit hit. A rule you renegotiate at the moment it costs you is not a rule.

Fed decision in September — 25 bps increase | Market: 57.5% ⚪ NO TRADE — until Friday at 8:30
A hike is now the market's single most likely September outcome, ahead of no change at 40.5%, on over $11 million of volume. Four days ago the hike tail was near 20%. We are not fading a move that large on the same week we just misread this Fed twice, and we are not chasing it either. The entire repricing happened without new inflation data, which means payrolls is the first genuine test of whether it is real. Ask us again Friday afternoon.

Bitcoin dips to $55,000 before 2027 | Market: 55.5% 🟡 WATCHING — the cleanest way to hold this view without timing the Fed
Roughly a coin flip on a 13% drawdown, with nearly $5 million of volume behind it, in a market that already touched $58,000 on July 1. What makes it interesting is that it expresses the September hike risk without requiring you to call the meeting. No position yet — we have taken enough Fed-shaped losses this week to earn some patience. The trigger we will act on: September hike odds above 65% with this contract still under 60%.

CLARITY Act signed into law in 2026 | Market: 29% ⚪ STAY OUT — recess starts Friday
We exited this near 33–37% on Wednesday under a stated rule, and it has drifted to 29% since, so the exit was correct even though the trade was a loss. Nothing has improved: no floor vote, no cloture motion, no calendar date, and the Majority Leader has said he does not expect it to reach the floor before the recess that begins around Friday. It still needs reconciliation with the Senate Agriculture text, sixty votes, reconciliation with the House bill, and a signature. Industry endorsements from BlackRock, Fidelity and Goldman are real but they are not floor time. Re-entry only reopens in September, and only if the merged text survives recess intact.

Track the whole board live at polymarket.com — free, no account required.

A five-year-old firmware bug emptied 500 wallets in 25 minutes. The company that said it would never sell bitcoin sold bitcoin. The Fed held rates and the market immediately made a September hike its base case. And bitcoin ended the week almost exactly where it started, still up roughly 7% across July.

One question: after this week, is a September hike more likely or less likely than the 57% the market is quoting this morning?

Hit reply. We read every response, and the best calls run Wednesday.

See you Wednesday — with the ISM and JOLTS read, the STRC clock ticking toward par, and whether $61,400 held.

— The Baseline Crypto Team