The Clock

Ask ten people how far along the monetary transition is and you'll get ten moods. One says it's basically over, the dollar is finished, look at the debt. Another says nothing has happened, the dollar is 57% of world reserves, call me when it's 20%. Both are reading the same decade. Neither is reading a number.

That's the problem this letter solves, and it's a smaller problem than it sounds. You already have the receipts — they've been accumulating in the terminal since Arc One, one per letter, each with a source. What you don't have is a way to collapse them into a single reading you can check twice a year and compare against last time. A pile of receipts is not a measurement. A measurement has a method, a scale, and a number at the end.

So: the Transition Clock. Midnight is the old system fully intact. Noon is the transition complete. Everything interesting happens in between, and tonight the clock reads 7:46.

Here's how it gets there, so you can rebuild it yourself and disagree with my arithmetic on purpose rather than by feel.

There's a structural reason "how far along are we" resists an answer, and it isn't that the data is hidden. All of it is public, most of it is free, and the sources are boring institutions that publish on a schedule.

The reason is that the transition has no single unit. Reserve share is a percentage of a pool. Debt-to-GDP is a ratio of a stock to a flow. Interest-to-defense is a ratio of two flows. Mined supply is a fraction of a fixed cap. Real yield is a difference of two rates, and it can go negative. You cannot average those. There is no arithmetic that turns a percentage, a ratio, and a rate spread into one honest figure — which is why almost everyone gives up and reaches for a mood instead.

The trick — and it is a trick, not a discovery — is to stop trying to average the numbers and average their positions instead. Every one of those five figures can be placed on a journey from somewhere to somewhere. Once each hand reports a percentage of its own journey, they share a unit, and a mean becomes legal.

That's the entire intellectual content of this letter. Everything below is bookkeeping.

The Clock takes five receipts, scores each one on the same 0–100 scale, averages them, and maps the average onto a twelve-hour face. That's the whole instrument. No weighting scheme, no proprietary index, no black box — five divisions and a mean, which you can do on the back of this email.

Each hand needs three inputs: where the old system sat (midnight), where you'd call the transition complete (noon), and where the number is now. Then:

Score = (now − midnight) ÷ (noon − midnight) × 100, clamped to 0–100.

The five hands, and why each one is on the face:

Hand one — Reserve. The dollar's share of allocated global FX reserves. Midnight is 71%, the 2000 peak. Noon is 40%, which is where I'd say the dollar has been genuinely displaced rather than merely crowded. This is the hand that measures whether the world still defaults to the ruler.

Hand two — Ledger. Federal debt as a share of GDP. Midnight is 60%, roughly the pre-2008 norm. Noon is 130%, a level at which the arithmetic stops being a policy choice and starts being a constraint. This hand measures the pressure on the issuer.

Hand three — Interest. Interest on the debt divided by defense outlays — the quiet default ratio. Midnight is 0.5×, noon is 1.5×. Crossing 1.0× was the headline; the hand measures how far past it we've gone.

Hand four — Issuance. Bitcoin mined as a share of the 21 million cap. Midnight is 0%, noon is 100%. This is the only hand with non-arbitrary endpoints, and the only one that can never run backwards.

Hand five — Yield. The silent tax — the three-month T-bill minus core PCE. Midnight is +2.0%, a world that pays savers. Noon is −2.0%, a world that quietly bills them. This hand runs in reverse: as the number falls, the score rises.

Every figure below was pulled this evening. Do the division yourself; it takes four minutes.

Average the five: 64.7. Multiply by twelve hours: 7.76. The Clock reads 7:46.

Two things worth noticing before you decide what that means.

The first is the spread. The hands are nowhere near each other — Issuance is at 95.6 and Yield is at 37.5. That spread is the actual finding, and it's more useful than the average. The supply side of the new system is nearly finished building itself; the cost side of the old one has barely started biting. A single number hides that. The five hands don't.

The second is the direction of travel on hand five. The silent tax is +0.5% tonight — cash is currently paying you about half a point above core inflation. That is a positive real yield, and it is the least transitional number on the board. If you've been carrying a story in your head where cash is being confiscated every month, tonight's receipt says otherwise. The Clock is only useful if it can tell you that.

It is not a countdown. Nothing happens at noon. There is no event scheduled for 12:00, and if the Clock reaches 9:00 next year that is not a signal to do anything differently than the floor, the slice, and the standing rules already tell you to do.

It is not a trading input. If you find yourself thinking "the Clock hit 8:00, time to move," you have converted a thermometer into a trigger, and you've reintroduced exactly the timing question this publication has refused since letter one. Your slice is sized by your capacity to hold it through a 70% drawdown. That sizing does not change because a ratio moved.

It is not precise. It reads to the minute because dividing by twelve produces minutes, not because the underlying data supports that resolution. Round it in your head to "somewhere between seven and eight" and you've lost nothing.

And it is not mine. The endpoints are arguable, which is the point of publishing them.

One — the endpoints are judgment calls dressed as arithmetic. Four of the five noon values are numbers I chose. Move Reserve's noon from 40% to 20% and that hand drops from 44.7 to 27.5, and the Clock falls to about 7:25. Move Ledger's noon to 150% and the whole reading drops further. The method is transparent, but transparency is not the same as being right, and anyone who wants a different time can get one honestly by arguing the endpoints. I'd rather that argument happen in the open than inside a proprietary index.

Two — equal weighting is almost certainly wrong. I average the five hands because I can't defend any particular weighting, not because equal weighting is correct. Issuance is arguably the least informative hand on the board — Bitcoin's mining schedule was fixed in 2009 and tells you nothing about adoption, yet it contributes a fifth of the reading and it's the highest score. A reasonable person would weight it near zero and get a Clock around 7:00.

Three — five hands is a small sample of a large thing. There's no hand for cross-border settlement, none for the share of trade invoiced in dollars, none for stablecoin float, none for domestic price stability. I picked receipts that already had sources in the terminal. That's a convenience criterion, not an analytic one.

Four — a mean can move for boring reasons. Debt/GDP falls when nominal GDP runs hot, not only when debt is repaid. The Clock would read that as the transition reversing. It isn't; it's a denominator.

The Blind don't check. Not out of stubbornness — out of the reasonable belief that this is all too complicated to have a number attached, so why look. The Clock is built for exactly this person: five divisions, twice a year, and the thing you were told was unmeasurable turns out to have a reading.

The Scared check daily and read the Clock as a countdown. They see 7:46, decide four hours fifteen minutes are left, and start making calendar decisions about their portfolio. This is the failure mode I'd most like to prevent. The Clock has no hour at which anything is required of you. If checking it more than twice a year makes you want to act, check it once a year.

The Prepared check it, note the spread, and change nothing. They log the reading next to the date, so that in July they can ask the only question the instrument is actually good for: which hand moved, and why? A number you record twice a year becomes a trend. A number you feel every day becomes noise.

Four tools now: the floor (months), the slice (size), the multiplier (buckets and reps), the clock (the reading). Defense, upside, engine — and now the instrument that tells you what the weather is doing to all three. The Position arc has its dashboard.

Next week: what to do when the dashboard says you've drifted. Your slice was sized once, at one price, in one mood — and then the market moved it for you without asking. The twice-a-year review that fixes that, without a single timing decision in it. Working title: The Ledger.

If this one was useful, three things:

One — build your clock tonight. Five rows, three columns, one average. Fifteen minutes with the terminal open, and you own the method rather than borrowing my number.

Two — write the reading down with tonight's date. The Clock is worthless as a snapshot and genuinely useful as a series. You need two readings before it tells you anything, so start the series.

Three — reply with the one hand you'd remove or replace, and what you'd put on the face instead. I'll publish the best objections. The endpoints get better when more people argue with them.

Understand — the transition has a number, and the number has a method. Position — five rows on one page, twice a year. And don't panic — nothing happens at noon.

Your portfolio has a dashboard now. Check it about as often as you check your smoke detector.

— Bill2Billion

P.S. Not financial advice, and the Clock least of all — it's a measuring instrument, not a recommendation, and it licenses no action whatsoever. Every figure was pulled the evening of 22 August 2026 and several will be stale before you finish reading: the three-month bill at 3.81%, core PCE at 3.3% (June print — July lands the week after this letter and will move hand five), the Fed's range held at 3.50–3.75%, debt/GDP at 122.6% for Q1, the dollar at 57.13% of allocated reserves for Q1, central banks having bought 863 tonnes of gold across 2025, Bitcoin near $77,000 — about 1,299 sats to the dollar — and 95.6% of the cap mined. Check every one of them; the method is only worth what the inputs are. The standing rules survive here as everywhere: no leverage, no timing, no all-in, and the 1–10% band is the band regardless of what time the Clock says. I hold positions in some of the asset categories discussed.