The Fed Printed Two Reserve Numbers for One Week: One Fell $88B, One Rose $17.9B

Two Fed numbers hit the tape on Oct. 1. They cover the same seven days. They point in opposite directions. That's the entire liquidity debate in a nutshell.

The chain doesn't lie, but reserve charts absolutely can if you don't know what you're reading.

Sept. 23 to Oct. 1: The Sequence

Start with the setup. The Fed's Wednesday reserve series, that's the one measuring balances at a single point each week, sat at $2.969922 trillion on Sept. 23. By Sept. 30 it was down to $2.881686 trillion. That's a drop of $88.236 billion in one week.

Now the other series. The weekly-average reserve gauge, which averages daily figures across each week, went from $2.930193 trillion to $2.948090 trillion over those same dates. Up $17.897 billion.

Same institution. Same release. Two opposite headlines. Anon, let me explain why this matters more than usual right now.

The Treasury General Account tells the same confusing story. Its Wednesday balance jumped $36.729 billion to $984.046 billion. Its weekly average fell $28.410 billion to $948.674 billion. So the government's checking account at the Fed also grew and shrank in the same week depending on how you look at it.

Here's the thing. Both numbers are real. They just measure different windows. One compares two Wednesday snapshots. The other compares two seven-day averages. You can't blend them without breaking the math.

Then Oct. 1 arrived.

The H.4.1 release dropped and gave us the accounting bridge. Factors supplying reserves fell $4.650 billion. Factors absorbing funds rose $83.586 billion. Toss those together and you get the $88.236 billion drop, fully reconciled.

Where'd the absorption come from? Reverse repos rose $41.158 billion to $361.883 billion. But here's the detail everyone skips. Most of that, $350.344 billion, belonged to foreign official and international accounts. Only $11.539 billion was in the domestic "others" bucket.

And on Oct. 1, that domestic overnight reverse repo line collapsed from $11.539 billion to $350 million. Basically gone in a day.

Let's talk funding. The New York Fed's SOFR reading for Sept. 30 came in at 3.90%. That matched interest on reserve balances exactly. The five latest SOFR medians sat between 3.88% and 3.90%. The 99th percentile was 3.99% on $3.230 trillion of underlying volume.

SOFR did tick up from 3.88% the day before. But the quarter-end median also matched Sept. 25 and Sept. 28. So no outright panic in the middle of the market.

Real talk: that's a calm tape, not a squeeze.

What Actually Changed

Nothing broke. That's the story.

Every crypto account I saw this week instinctively read "$88 billion in reserves vanished" as dry powder leaving the system. Fewer dollars sloshing around. Fewer bids under Bitcoin. That narrative writes itself fast and travels faster.

It's also wrong, or at least wildly incomplete.

That $88.236 billion didn't evaporate into the void. It moved into Treasury cash and reverse repo balances. And most of the reverse repo side was foreign official money, not domestic banks parking cash overnight because they were scared.

If you'd built a trade around the Wednesday drop and then subtracted a domestic reverse repo series from total Fed assets, you'd have mis-sized the whole thing. You'd be counting foreign liabilities as if they were domestic drains. That's a real error with a real cost.

Who wins here? Anyone who waited for the second data point. Anyone who checked whether their "weekly" chart was actually a weekly average or just a once-a-week balance.

Who loses? Traders who aped into a short on Sept. 30 headline fear and woke up Oct. 2 with SOFR sitting right at IORB and domestic reverse repo usage down to $350 million.

The standing repo facility was live too. $1.2 billion of outstanding repo agreements on Sept. 30, with a 4.00% rate, two daily operation windows, and a $40 billion proposition limit per counterparty per security type per operation. A backstop existed. It wasn't screaming.

This is bigger than people realize, and it's not because liquidity is tight. It's because the measurement layer is doing the work that the market layer used to do. The number you pick decides the narrative before any price forecast enters the room.

So what does an $88 billion reserve decline prove about Bitcoin? Almost nothing on its own. Reserve balances show where cash sat on a Wednesday. They don't show anyone getting margin called, any desk pulling bids, or any exchange funding rate blowing out.

For that you need a financing channel. You need dated funding rates on crypto venues. You need futures basis, order book depth, credit conditions. You need something that connects the Fed's balance sheet to actual coins changing hands.

We don't have that link yet.

What to Watch Now

Oct. 1 SOFR wasn't even published at the early Oct. 2 snapshot. That's your next real data point. Watch whether the median holds at 3.90% or drifts.

Then watch the reverse repo line. If the foreign official pool drops the way the domestic bucket just did, the $361.883 billion figure unwinds and the reserve picture softens. If it stays sticky, you've got a real drain worth tracking.

And watch the standing repo facility. Any meaningful pickup in usage above that $1.2 billion balance tells you access to cash got harder for somebody. Zero or near-zero usage tells you the backstop stayed decorative.

The threshold I care about is simple. Keep an eye on whether SOFR stays at or above IORB across the next several observations. If it does, and if the 99th percentile climbs past 4%, the quarter-end adjustment theory starts breaking down. That's when a Bitcoin transmission story gets legs.

But even then, simultaneous Bitcoin weakness and elevated repo rates prove correlation, not causation. Two things moving together isn't one thing causing the other.

Look, the honest read is this. The Fed's own accounts show the $88.236 billion drop leaving reserves and landing in Treasury cash and foreign reverse repos. The weekly average went the other way. The latest secured funding median sat right at IORB. That adds up to a quarter-end reshuffle, not a liquidity shock.

I've been saying this for weeks. Stop trading the headline number. Trade the measurement you can actually defend.

Next real test lands with the next funding print and the next H.4.1 drop. Until then, the reserve "crash" is a chart artifact, and the chain, as always, doesn't lie.