If you have to get it right in writingEasy

Journalists, lawyers, auditors, supervisors, teachers: the mechanism checked before it goes out, and the four errors that survive every edit.

5 min read · 829 words

Somebody has to describe a financial mechanism correctly and cannot afford to be wrong in public: a reporter on a deadline, a lawyer drafting around an instrument, an auditor writing a memo, a supervisor asking a question, a lecturer setting a problem. This page is a checking route rather than a reading one — and it names the four errors that survive every edit, because they are errors of mechanism rather than of grammar.

The four that get through

1. Notional read as exposure. "A $100m swap" moves cash flows worth a fraction of that. The number in the headline is a reference amount, not money at risk — and confusing them is a whole genre of finance writing. See interest rate swap.
2. The wrapper mistaken for the thing. An ETF, a note, a certificate and a fund can hold identical assets and fail in entirely different ways, because what changes is who owes you. What the wrapper changes is the page on exactly this, and it is the single most common structural error.
3. Liquidity failure described as insolvency. A firm that cannot pay today and a firm that is worth less than it owes are two different events with two different remedies, and from outside on the day they look identical. What happens when a bank fails separates them.
4. A price move attributed to an opinion. A price is set by whoever has to trade, not by whoever has the best argument. Every asset-class page carries a section sorting participants by whether anybody had a choice — index funds at a rebalance, insurers at a rating boundary, liquidation engines with no telephone.

Checking a mechanism in four minutes

1. Find the instrument in the atlas or through search. Read level one; it assumes nothing.
2. Look at the who-pays-whom diagram, not only the payoff chart. It names every party and every payment, which is what a description usually gets wrong.
3. Read the risk profile. Five failure modes, one marked as deciding. If your sentence implies a different one, the sentence is probably wrong.
4. Check the vocabulary in the glossary. Several words on this subject travel under one name — "yield" is four different numbers — and picking the wrong one is invisible to an editor.

The pages that settle the recurring arguments

• What yield actually means — four numbers, one word.
• What an index really is — a rulebook with a number attached, and the rules decide the return.
• Volatility and risk — not the same thing, and the gap is where the ruinous events live.
• What a credit rating says — one opinion about one narrow question.
• Reading a market number — what a quoted figure is and is not.
• Who actually holds a share — the chain, and why "shareholders voted" is a longer sentence than it looks.

For a case, the mechanism rather than the moral

The 40 case studies here are written to describe what broke and why, mechanically. None of them characterises anybody's conduct, assigns blame or restates an allegation, which is deliberate: a description of a mechanism is checkable, and a characterisation of a person is a different kind of claim with a different standard of proof. If you need the second, this is not the source — but the first is exactly what these pages are for, and each one links to the instruments and concepts it used.

For a lecturer or a course lead

• Set a level, not a page. Every instrument is written at five levels; level one assumes nothing and level five assumes the desk. A single page serves a first-year and a postgraduate differently.
• 275 questions with a written explanation on every one, pickable by area at the knowledge test.
• Printable sheets per asset class at crib sheets — generated from the same tables the pages use, so a handout cannot drift from the site.
• The interview bank on every desk, asset class and industry group — 217 questions with the parts a complete answer contains, which is a seminar prompt as much as an interview one.
• Seats for a cohort: teams.

What this site will not do for you

• It will not give you a number to quote as today's market size. There is not one on the site, deliberately: none could be verified from inside this repository and none would carry a date you could check. Figures here are conventions, statutory ceilings or dated historical events.
• It will not tell you whether something is good. No instrument here is called cheap, attractive or dangerous, because that is a verdict and this is a description.
• It is not a source you should cite instead of the document. If a prospectus, a filing or a rulebook is the authority for your sentence, read that. This site is for understanding the mechanism well enough to read it quickly.

Everything here is original to this publication, and the licences page says what the three third-party components are.