Which Risk DecidesStart here

Five ways an instrument can go wrong, and which of them actually drives the outcome — read across all 118 products at once, in one place.

9 min read · 1 602 words

The question worth asking about anything

• Every instrument on this site can go wrong in five ways, and on almost every one of them, one of the five does most of the work. Naming which is a two-minute exercise and it is most of what a risk seat does.
• It is also the fastest way to see through a name. A covered bond and a contingent convertible are issued by the same banks into the same market, and they belong at opposite ends of this page.
• The value is in the comparison, not in any single entry. Two products with the same profile behave alike under stress even when nothing about their documentation resembles the other's — which is why the lists below cut straight across the eleven markets.

The five, and what each looks like when it arrives

• Market — the price of the thing moves. The most talked about and the most survivable, because it is visible from the first day and it is what everybody prepared for.
• Credit — somebody who owes you does not pay. Slow, then sudden. The instrument keeps performing right up until the counterparty stops, and the warning signs are in a document rather than on a screen.
• Liquidity — you cannot get out at anything near the marked price. Nothing has defaulted and nothing has fallen; there is simply nobody there. The mark is not a lie, it is just not a bid.
• Funding — cash is needed before the position pays off. Margin, capital calls, a roll that does not happen. This is the one that ends firms that were right: Metallgesellschaft had the correct view on oil and ran out of money financing it, and the 2022 LDI episode was a hedge working exactly as designed and demanding cash faster than it could be found.
• Operational — the failure is in documents, systems, keys or people, not in prices. The wrong reference obligation, a missed exercise cut-off, a misspelt port on a bill of lading, an address wrong by one character. Nothing broke; the process simply did what it says.

What decides what

Every product page carries the same five bars in the same order. Here they are read the other way round — by risk rather than by instrument. The counts are derived at build time from the same table the product pages use, so the two cannot disagree.

How to use this

• Before you read a product page, guess which of the five decides it, then check. Being wrong is the useful outcome: it tells you which page you actually needed.
• In an interview, this is the frame that turns a memorised description into an answer. "What is the risk in a repo" is a weak question until you say that credit is small because it is collateralised and marked daily, and funding is the whole of it, because a balance sheet financed overnight is re-underwritten every morning by people free to decline.
• On a desk or in an operations seat, the fifth family is the one nobody prepares for and the one that generates most of the actual incidents. It is also the only one of the five that a careful reader of the documents can see coming.
• Across a portfolio, the point is not to avoid a family but to notice when the same one is deciding everything. Positions that look unrelated and share a funding profile are one position. Diversification covers what that costs when it goes wrong.

Where to go next

• How products fail — six patterns that between them cover every case study on this site, and the natural companion to this page: this one says which family, that one says what the failure looks like.
• Risk measures — what the numbers that claim to quantify all of this actually assume.
• Margin and collateral — the mechanism behind almost every entry in the funding column.
• Liquidity — why the third family is the one that turns a bad week into a solvency question.
• The case studies — every one of them is one of these five arriving.