Read the essentials
What was done and what it shows, in one PDF.
Open essentials.pdfWorkspace · 2 October 2026
The framework, the 2 October answers, and the data as PDFs. Zip this folder and upload it. Raw downloads, spreadsheets and Python stay in the workspace, not on the server.
On 1 October 2026 the 30-year Treasury was at 5.67%, its highest since 2002. The journal’s three-year probabilities were B (higher real rates) 50%, C (fiscal or inflation) 35% and A (disinflation) 15% by gut, and 55/30/15 from the data. The data room’s headline vote that day was A1 B2 C0, which is not a call. The 2002 back-test shows the vote lags, so the probabilities are held loosely.
What was done and what it shows, in one PDF.
Open essentials.pdf126 pages, 88 tables, one verification pack per series.
Open audit.pdfThe path from the source notes to the framework, the data room and the journal.
Open the noteTen questions raised while reading the essentials. Questions 1–5 are about using the data room. Questions 6–10 use data fetched on 2 October 2026. Charts and tables are in the PDF.
Open the answers PDF · Markdown master
| # | Question | Short answer |
|---|---|---|
| 1 | Audit PDF with a list of tables | Done. 126 pages, 88 tables. Open the audit PDF. |
| 2 | See the data wide, and print it | Wide means one column per series. That workflow lives in the workspace data room, not on this site. |
| 3 | Update the series | make refresh on the machine that holds the data room. |
| 4 | What changed between updates | make changes in the workspace compares the register with the last git commit. |
| 5 | Is it automatic? | It was meant to run at 07:00 and had never run, because a missed launch is skipped while the Mac is off. It now also catches up at login, at most once a day. |
| 6 | UK 2022 and Japan’s yield cap | Japan capped yields; the UK did not. The Bank of England’s purchases were a two-week backstop. In Japan the cap held, the yen fell 31% against the dollar and prices rose 8%. |
| 7 | Why a nominal cap does not protect wealth | A cap fixes the bond’s price in yen or pounds, not what that money buys. In three cases, holders lost 10–24% in real terms. |
| 8 | Basis for the 4–5.5% range | Rebuilt from the yield’s three parts, the range is 3.9–5.4%. The past range of the 10-year in regime B is 3.75–5.4%. |
| 9 | 3–4% against 4–5.5% | A 3–4% yield is as common in regime B as in A. The regime is read from the parts of the yield. All three parts say B. |
| 10 | Each driver against each part of the yield | Short rates explain most of the rise since 2019. Saving minus investment explains real yields. Debt adds about 3bp per point of GDP. |
For UK 2022, Japan YCC, building blocks and the seven-economy charts, open answers.pdf.
Every file downloaded on 2 October 2026, as charts and wide tables, one section per publisher. This site carries the report as PDF only.
Open the research data report (PDF) · Audit PDF (all series)
The CSV files and fetch scripts that produced this report are not published here.
How to think about the question: the parts of the long yield, the three regimes, and the slow forces scored −1, 0 or +1. Written from the three source notes and the yield chart of 1 October 2026.
The reading copy, with the chart embedded.
Open the frameworkThe file the HTML was made from.
Open the markdown5y5y, ACM, TIPS and the other terms.
Open the glossaryThe forecast written on 1 October 2026. Past entries are not edited; a revision is a new dated page. The one-page copy is the sheet that prints on a single A4. The full entry keeps the appendix for each watch metric.
The call, the probabilities and the levels to watch.
Open the one-page entryThe same call with the appendix.
Open the full entrySource text for the HTML entries.
2026-10-01.mdWould the method have worked in May 2002? The vote, read with what was knowable then, pointed the wrong way on the level bands because those bands carry the calibration of a later era. The structural drivers were the part that got the direction right. Charts are in the HTML report.
Votes, inputs and what happened next, with the charts embedded.
Open the reportHold the probabilities loosely. The vote lags.
Open the noteVintages, the splice of the 5y5y, and the three ways the vote was scored.
Open the noteMarkdown masters and source notes for the published pages. No Python, CSV, XLSX or live data-room exports on the server.
Daily, monthly, quarterly and the annual review (workspace only).
Open the noteWhat changed, by date.
Open the changelogWhat the engine does in the workspace.
Open dataroom-README.md