On Monday this newsletter explained why bitcoin had risen straight through a credit selloff, and it did it in four words: bitcoin is long liquidity. The week since has been a clean test of that sentence. It failed.

Start with what was supposed to hurt it. Last week the ten-year Treasury yield printed 5.29% on 30 September, its highest since May 2002. High yield credit spreads went through 3% and reached 3.24% on 1 October, the widest since March. Bitcoin rose 3.4% on Coinbase closes, from its 29 September close to its 4 October close. Everything on the list that is meant to be bad for a risk asset happened, and it went up.

By the time this week opened, the list had turned the other way.

The Treasury's cash account at the Fed, the TGA, fell $98.3 billion in the seven days to 7 October, to $885.8 billion, about $90 billion of it on 1 October as the quarter turned. Cash leaving that account goes back into the banking system, so that is liquidity going in, and it was already in before this week began. High yield spreads came in from 3.24% on 1 October to 3.09% on 7 October, most of it on 2 October. The one input that moved inside this week moved the helpful way: the ten-year yield came down from 5.31% on 5 October to 5.22% on 8 October, on Treasury's curve. Cheaper money, calmer credit, more cash in the system. If you had been handed the first two on Sunday night, told the third was coming, and asked for a direction, you would have said up.

Bitcoin opened Monday 5 October at $86,502.65 on Coinbase. The daily closes, on UTC days: 85,748.96. 85,539.77. 83,275.06. Then 81,692.79 on Thursday 8 October, after an intraday low of 80,314.70. That is down 5.6% in four sessions. It was trading at 81,902 at 01:45 UTC on Friday, which is a quarter to ten on Thursday night in New York.

The people who were supposed to be the steady bid sold. US spot bitcoin ETFs gave back $694.5 million across those four days on the funds that had reported, with IBIT's 8 October figure still missing, and the largest single day so far was Wednesday 7 October at minus $484.9 million. The full sequence is in the metrics below.

The obvious suspect is the Fed, because the minutes of the September meeting landed at two o'clock on Wednesday, the day of the biggest fall and the biggest outflow, and they were hawkish. We checked it before writing it down, and it does not hold. The December hike odds the minutes are supposed to have pushed up made their jump between 10 a.m. and 1 p.m. New York time on Monday, two days before the minutes were published. And a hawkish Fed shock raises yields. On 7 October the ten-year rose one basis point and the thirty-year three, and the ten-year ended Thursday nine basis points below Monday's 5.31%. If the minutes moved bitcoin, the bond market barely noticed.

So here is the honest version, and it is the same one in both directions. Last week bitcoin went up when the macro inputs said down. This week it went down with them saying up. Two weeks running, the asset has done the opposite of the textbook, and the inputs this newsletter tracks most closely are not what is driving the price right now. We are not going to invent a cause to fill that gap. A reason supplied after the fact is a comfort, and it is not the product.

Which also means Monday's four words were doing work they had not earned. The FILTER takes that apart properly.

Three metrics.

Metric 1, Fear and Greed: 59. That is the reading stamped 9 October, which the index posts the evening before. Monday's issue printed 70. Since then: 73, 71, 64, 59. Eleven points off in four days is a real move for this gauge. The detail worth keeping is the label. At 59 it is still classified as greed, after a week in which the price fell 5.6% and institutional holders sold. Sentiment came down. It did not come close to fear.

Metric 2, ETF flows: minus $694.5 million ex IBIT's Thursday, 5 to 8 October. Our sum of Farside's printed daily totals: minus 89.8 on Monday, plus 118.8 on Tuesday, minus 484.9 on Wednesday, of which BlackRock's IBIT was minus 207.7, and minus 238.6 on Thursday. That last figure is partial. IBIT had not reported when we wrote this, so the four-day total will change in either direction, and we are not calling it final.

Metric 3, Strategy's cushion: 8.6%. The company holds 848,000 BTC at an average cost of $75,440.70, per its 5 October filing. Against the 81,902 above, the stack sits 8.6% over cost, and at Thursday's intraday low it was 6.5%. The more interesting number in that filing is not the holding. It is what the company spent its money on instead, which is in the RADAR.

The STACK grades three dated tests this morning.

A quiet window, stated as one. We found no new exchange, bridge or protocol incident in the window, and no movement on CLARITY that we can find since 15 September.

📣 THIS SPOT IS OPEN

Baseline Crypto reaches nearly 7,000 Bitcoin-focused readers every Monday and Wednesday morning, with an open rate just under 50% over the last four weeks. Direct sponsor placements are open, premium position, top of the issue, one per send.

Reply to this email for rates.

"The liquidity is back. The Treasury has put $98 billion into the system, yields are coming down and credit is healing. Bitcoin is long liquidity, so it goes up from here."

No. All of it was in place this week and bitcoin fell 5.6%. And four of those words are ours, from Monday.

Start with what is right in it, because there is plenty, and because we argued it ourselves. Over a long enough horizon the amount of cash in the system matters to an asset with no earnings and no coupon. When the Treasury runs down its account at the Fed, bank reserves go up, and some of that money ends up in things with more risk in them. That is not a crypto superstition. It is plumbing, and across months it has a respectable record.

Now look at when the money actually arrived. Of the $98.3 billion that left the Treasury's account in the seven days to 7 October, about $90 billion went in one day, 1 October, as the quarter turned. Monday's issue reported that day. After it the account barely moved: $893.7 billion on 1 October, $885.8 billion on 7 October. Bitcoin closed 4 October at $86,507.11, up 3.5% from its 30 September close, with the money already in. Over the next four sessions it gave all of that back and more.

If liquidity is the cause, it is a cause that pushed the price up and then down with nothing changing in between. That is not a mechanism. It is a story that fits any chart you lay it over.

The incentive question gets further, as it usually does. Who actually sold this week, and what are they paid to do? ETF holders took out $694.5 million on the figures reported so far. Strategy, the largest corporate holder there is, spent $176.3 million on its own preferred and $28.7 million on bitcoin. Neither of those decisions needs the Treasury's cash account to explain it. We do not know why the ETF money left. The filings that would show who sold are 13Fs, which report quarter-end holdings and are due 45 days after the quarter closes, so the October sellers will not be visible until the middle of February. Anyone naming them before then is guessing.

So here is exactly where Monday went wrong. Bitcoin is long liquidity may well be true over a year. We used it to explain a week. A slow mechanism stretched over a fast move will explain anything, which is the same as explaining nothing, and the four sessions since are the proof. We keep the claim. We withdraw the use we made of it.

The other half of Monday's FILTER needs the same treatment. We wrote that a market stripping out hikes without adding cuts was saying the tightening is probably over. Since then the minutes said most participants see another increase as likely by year end, Waller said he anticipates additional hikes, and the December leg went from 71.5% to 75.5%. That reading was early at best. We would not write it this morning.

What we would defend is narrower and less satisfying. Over a horizon of days, bitcoin's price is set by who is buying and who is selling, and this week the two largest buyers anybody can identify both stepped back. That is a description, not a cause. It is as far as the evidence goes, and a newsletter that offers you more than the evidence is a newsletter to stop trusting.

Three numbers that define the next three weeks.

Minus $484.9 million, Wednesday 7 October, the largest daily ETF outflow of the window. IBIT was $207.7 million of it, less than half, so this was not one fund having a bad day. It came in a week when every rate input we follow was pointing bitcoin's way, which is exactly why it is worth keeping. The thing to watch is Thursday's total once IBIT reports. If 8 October finishes below minus $484.9 million, the largest day of the window is also the latest, and that is a direction rather than a day. If IBIT reports something small, Wednesday stands on its own.

$98 billion, the TGA's fall in a week, and the symmetry test it sets for us. If $98 billion going back into the system did not hold bitcoin up, then the Treasury rebuilding its account is not, on its own, a reason to fear for bitcoin either. We are writing that down now, while the account is low, so that nobody here reaches for it as an explanation the next time it rises and bitcoin happens to fall. Evidence has to count the same in both directions or it is not evidence.

28 October, the decision that settles the October leg. As of Thursday evening, the October hike is at 15.5% and no change at 84.5%. December carries 75.5%. Waller's point that hikes need not come at consecutive meetings is most of the reason those two numbers sit where they do. The decision settles the October contract that afternoon. It will not settle the question this issue is actually about, which is whether any of it is moving bitcoin. Two weeks of evidence say no. A third would be a pattern.

Forget analyst predictions. Polymarket is a real-money prediction market, where 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, so read direction over ticks.

Scorecard. Three dated tests grade this morning, two days late.

These three were due to grade in Wednesday's issue. There was no Wednesday issue, so they grade now. Each rule named a print for 30 September, and the grades below use exactly those prints. Rule, number, verdict, consequence, in that order.

Test one, high yield spreads, set 9 September. The rule: the ICE BofA US High Yield option-adjusted spread, FRED series BAMLH0A0HYM2, as printed for 30 September. Branch A is 3.00% or above. Branch B is 2.70% up to but not including 3.00%. Branch C is 2.40% up to but not including 2.70%. Branch D is below 2.40%.

The number: 3.12%. The verdict: Branch A.

The consequence. When we set this on 9 September the latest value was 2.68%, which was Branch C, and we wrote that Branch A was what our own argument implied and 32 basis points away. It resolved there. It then fell back to 3.03% on 6 October, which the grade does not care about, because a branch is a reading on a date and the date was 30 September.

Test two, the ten-year real yield. The rule: FRED series DFII10 as printed for 30 September. Branch A is 2.80% or above.

The number: 2.93%. The verdict: Branch A. Unrevised.

The consequence. It was the highest print of 2026 at the time, and has since been passed, 2.95% on 5 October. We had the direction of the cost of money right through September.

Test three, US spot bitcoin ETF flows for September. The rule: net positive for the month, on our own sum of Farside's printed daily totals. Branch A is net positive.

The number: plus $2,647.7 million. The verdict: Branch A. The same figure we printed on Monday.

The consequence, and the honest context that goes with it. October has opened net negative, with $694.5 million out from 5 to 8 October before IBIT's Thursday figure. That is not the test and does not change the grade. It does mean the month we were right about is not the month you are in.

Three Branch A results, and what they are worth. All three were the branch our argument implied. On the inputs, September went the way we said. Then the SIGNAL above happened: we were right about the cost of money and the flows, and the asset those inputs are supposed to explain did the opposite of what they implied, two weeks running. A scorecard that grades the inputs and not the asset would call this a good week. It was a good week for our reading of rates. It says nothing about our reading of bitcoin.

The board has no live rule. The last one, the October entry, was retired on Monday after it fired with nobody at the desk. Nothing replaces it today.

Fed decision in October, 25 bps increase | Market: 15.5% ⚪ NO TRADE, rule retired Monday
No change is 84.5%, as of Thursday evening. It was 16.5% on Monday, so a point has come out in a week that produced hawkish minutes and a governor saying he anticipates more hikes. That fits Waller's line about consecutive meetings, though a one-point move in a thin market proves very little. What would move it: a voter saying October out loud before the 28th. Nobody has.

Fed decision in December, 25 bps increase | Market: 75.5% 🟡 NO TRADE, and Monday's read on it was wrong
No change is 22.5%, as of Thursday evening. On Monday we wrote, at 71.5%, that the hike was losing weight across the year. It has put four points back since, and the jump came between 10 a.m. and 1 p.m. New York time on Monday, two days before the minutes were out, so not even the minutes can take the credit. The hike is the clear base case for December. What would move it: the 28 October statement, which either sets December up or does not.

Zero Fed rate cuts in 2026 | Market: 95.55% ⚫ EXITED at 88.75%, a loss
As of Thursday evening. Short from 78%, trimmed at 85%, covered at 88.75% on a published stop on 3 August, and recorded as a loss in every issue since. It was 95.95% on Monday. With the committee talking about more hikes, nothing on the calendar looks likely to pull this down before December. What would move it: a dot below the current midpoint in the December projections.

CLARITY Act signed into law in 2026 | Market: 5.05% ⚪ NO TRADE, nothing to trade on
As of Thursday evening, against 5.0% on Monday. We can find no movement on the bill since 15 September, and the re-entry rule here was retired in print on 16 September. What would move it: a markup of the substitute text, and a new rule written before the fact.

The book. Flat for the nineteenth consecutive issue, recomputed from the archive: eighteen issues published after the 3 August exits, so this is the nineteenth. On Monday we said eighteen issues of no position was either discipline or paralysis and that we were no longer confident it was the first. A week in which our inputs were right and the asset ignored them is not an argument for taking a position on the inputs. It is the reason we have not.

Track everything live at polymarket.com, free, no account required.

On Monday we asked one question with a one word answer: was the credit widening growth or plumbing.

Nothing came back. Our file of reader replies has nothing new since 3 August, which makes it seven issues running. Our own answer so far is in the RADAR, and it is that it is too early to say.

This issue's argument is short. For two weeks bitcoin has done the opposite of what rates, credit and liquidity said it should, and we would rather tell you we do not know why than sell you a reason.

One question: is that break between bitcoin and the macro noise that fades, or a shift that lasts? Reply with one word, noise or shift.

A reason is welcome. The word on its own is enough.

Hit reply. We read every response, and the best calls run in the next issue.

The Baseline Crypto Team

Two tools. Free, no account, nothing to install.

The hardware wallet checker takes your device and firmware version and tells you whether you sit inside a population with a known problem, and what to do if you do. The custody audit is nine questions about what happens to your coins when something goes wrong, and it is deliberately unkind about backups and inheritance.

This week we took back an explanation we had given you four days earlier. That comes with the job of explaining prices, and the fix is to lean on explanations less. Holders have their own version of the habit. A good story about why the price moved feels like control, and it is easy to spend more time on the story than on the setup that decides whether you actually keep your coins.

None of that setup cares why bitcoin fell 5.6%. The firmware on your device, where the seed phrase lives, whether anybody could reach your coins if you could not. Those were the same on Thursday as they were on Monday. They are what the two tools check, and nothing in either depends on being right about the Fed.

Send the custody audit to the friend who texts you the reason bitcoin moved within ten minutes of it moving. They always have the reason. The nine questions are the ones nobody has asked them.