Study build, not the P11 site — built Fri 16 Oct 2026, 16:47 from notes\fin\ with tools\render_note.py. KaTeX 0.16.11 and markdown-it 14.1.0 are vendored in vendor/ and the figures are copied into figures/, so this folder works on its own from file://, with no network.

Financial Economics — the meeting notes

All 17 meetings from 3 Sep to 15 Oct, in date order: 16 notes on pages (910,310 words) and 1 with no note. The midterm is Tue 20 Oct. Each note opens on its own page, with the previous and next note in its top bar. Its PDF prints every collapsed block expanded; a PDF with any text below 6.5 pt is held back rather than shipped, and marked PDF held.

The exam provides the formula sheet. Dávila, 10 Sep: “I'll give you formulas for the exam, for everything” (Lecture 3, 66:16); and on 15 Oct, with the review deck's p. 4 on screen: “these are the equations that you see in the exam” (Lecture 12, τ 17:20).

Leak check, 16 Oct. Every note, its parts and the drafts it includes were scanned against the exam papers kept back for practice; each content hit was traced and judged a coincidence with course material (beat numbers, link digits, the decks' and handouts' own data and wording). No note is held back. One thing to know before reading: 26 of the adapted self-checks in 7 notes restate the claim of the item they adapt, so reading them first would make those practice-paper items easy to recognise. Each idea is also in the course's own decks, workbook or recordings. They are named in the rows below and at the top of each page, so they can be skipped until the papers are sat. Tony's ruling, 16 Oct: leave them as they are, flagged.

Provenance: recorded = from the class recording · stitched = two recordings of the meeting joined on their overlapping words · reconstructed = no recording, rebuilt from the written materials and marked · part reconstructed = part of a recorded meeting is not on the recording and is rebuilt, marked. Midterm, Tue 20 Oct: as the 15 Oct review settled it — the review deck (21 p. 5) lists Blocks 1–5 and puts nothing out; the recording narrows inside them, and each row gives what was said about that note's material. Audit state: passing with the date of the note's last confirming P9 re-check, and what was edited after it (those edits were checked by their own math, quotation and leak runs, not by a new audit) · audit failed, fix not run · held back by §0 · approved = approved by Tony. STALE: a reconstructed note whose range a later record could still correct.

Known, and deliberately left as it is

• fin-2026-09-17-L4 is the approved exemplar and still carries 61 gendered pronouns for Dávila. It was not edited because approval protects it; its row reads approved — pronoun pass owed, Tony's call.
• Eight gendered pronouns remain in practice-problem drafts that two pages include — three in the 8 Sep note's (fin-lec-01b-p8-spread.md l. 74; fin-lec-01b-p13-short-sale-payoffs.md ll. 78, 98) and five in the 3 Sep note's. The pronoun scanner reads only a note's own parts, not the drafts it includes, and Tony said to leave the drafts, so these pages show them.

Found by this build, not fixed here

• The 15 Oct page's CML-and-SML figure does not display (its PDF shows the gap too): the drawing's file has an unescaped “<” in a label, so browsers refuse it. The fix is one character in the figure's drawing script, which this build may not edit. The figure's description and its reading line are on the page; the drawing is not.

Deck citations: plain 03a/03b/03c p. N are the copies Dávila posted (title pages “Updated: September 22”); 03x-v0826 p. N is the earlier copy the build first had. 21 p. N is the 15 Oct review deck. Lecture and Section numbers in the titles count meetings; the decks keep their own codes (01a, 04b, …).

Rebuild: .venv-whisper\Scripts\python.exe -B work\_batches\fin-site-tools\build.py, then pdf.py, build.py again and check.py in the same folder.