Credit Derivatives

This marketWhat it is, what trades, and the ideas it runs on.

The market at a glance

Credit derivatives isolate one question — will the borrower pay? — and make it tradable on its own. Born in the 1990s, the market grew explosively into the 2008 crisis and then consolidated under reform into a cleaner core: CDS, dominated by the index products and largely centrally cleared, plus a structured layer (CLOs) that finances the leveraged-loan world.

Users split by need: banks shedding concentrated loan exposure (and lately, whole-portfolio capital relief via SRT deals), credit funds expressing long/short views a cash bond can't (shorting credit means buying protection), and macro traders using indices as the fastest recession-risk dial available.

The credit triangle — this market's E=mc²

One approximation organises everything here: spread ≈ default intensity × loss severity.

What the symbols mean

• lambdaan intensity, usually of defaults per year
• Ra return

A 200bp spread with 40% recovery implies a ~3.3% annual default intensity. Every quote you see — single-name CDS, index level, CLO tranche margin — is a statement about \(\lambda\) and \(R\). The calculator below inverts it live.

Interactive: spread ⇄ default probabilityMedium

Turn a CDS spread into the market's implied default probabilities — the translation every credit analyst does in their head.

CDS spread (bp) Recovery rate %

Implied hazard rate

—

P(default ≤ 1y)

—

P(default ≤ 5y)

—

Expected loss p.a.

—

Flat-hazard credit triangle — the market's own quoting shortcut (ISDA standard model uses exactly this skeleton). Risk-neutral probabilities include risk premia; real-world default rates run lower.

How the products fit together

The single-name CDS is the atom: insurance on one borrower. CDS indices (CDX, iTraxx) bundle 100+ names into macro credit instruments with options and tranches on top. Credit-linked notes fund the same risk into a bond wrapper for investors who can't trade derivatives. CLOs apply the tranching idea to portfolios of leveraged loans — an actively managed securitisation that has become the buyout industry's banker.

Concepts to master

• Spread duration vs. jump-to-default — two different risks: mark-to-market pain from spread widening, and the binary loss when default actually hits. Books are managed on both.
• The basis — CDS and bonds price the same credit; their gap (the CDS-bond basis) trades on funding, deliverability and documentation, and blows out precisely in crises.
• Correlation — tranches turn the portfolio loss distribution into products: equity tranches fear many small defaults, seniors fear the correlated catastrophe. Correlation is the price of "together".
• Credit events are legal facts — determinations committees, auction protocols and restructuring clauses decide payouts; documentation literacy is alpha here.

Why this market matters beyond itself

Credit spreads lead. The high-yield market and CDS indices typically reprice weeks before equities accept bad news — "credit leads equity" is one of the most durable cross-asset regularities. Watching iTraxx Crossover is watching the economy's overdraft warning light.

Interactive: spread-duration P&LMedium

Credit portfolios live and die by one product: spread duration × spread change. This is the mental arithmetic of every credit trader's day.

Spread duration (years) Spread change (bp) Position ($m) Spread level %

Price impact

—

P&L on position

—

Breakeven widening (1y carry)

—

First-order only — convexity, defaults and rating migration are extra. The breakeven answers: how much can spreads widen before this year's carry is gone?

Interactive: tranche loss calculatorMedium

Where does a portfolio loss land in the capital structure? Set the attachment and detachment points and find out which layer absorbs it.

Portfolio loss % Attachment % Detachment %

Tranche loss

—

Subordination left

—

Tranche width

—

Status

—

Loss = min(max(L−A, 0), D−A) / (D−A). Note the leverage: a 3–7% tranche is untouched at 3% losses and destroyed at 7% — a four-point move in the pool wipes out the whole layer. That convexity is the whole story of CDO tranches and CLO equity.

The same questions, asked of all elevenThese sections answer the same thing on every asset class, so the answers can be read next to each other.

The units this market speaks in

• Everything is a spread, and there are several. Over the government curve, over the swap curve, as a constant spread over the zero curve, or as the spread on an asset-swap package. They answer different questions and they are different numbers. Asset swap.
• The unit of risk is CS01: what one basis point of spread widening costs, per million of notional. Its cousin, jump-to-default, asks the other question — what if it does not widen but simply stops paying.
• CDS is quoted in basis points running, but traded with a standardised fixed coupon and an upfront payment that squares the difference. The quote and the cash flow are deliberately not the same thing. CDS.
• Recovery is an assumption stated in percent, and it is an input, not an observation. Two desks with the same spread and different recovery assumptions hold different views.
• Ratings are letters, and the boundary is a cliff. The line between investment grade and high yield moves who is permitted to hold the bond, which is why the price gaps there rather than sloping. What a rating tells you.
• Loans are quoted in points of par and settle over weeks rather than days — a different clock from the bond market next door. Leveraged loan.

Who is choosing, and who is forced

The most reliable thing in credit is that a rating change moves who is allowed to hold a bond, and that this happens on a date rather than gradually.

• Forced: index funds when a bond leaves the index. A downgrade out of investment grade removes the bond from every index that excludes high yield, and every fund tracking one sells — regardless of whether the downgrade told them anything new.
• Forced: insurers by rating bucket. Capital charges step at the boundaries, so a bond that crosses one becomes costly to hold for reasons unrelated to its cash flows.
• Forced: CLOs against their tests. Coverage and quality tests can require action at exactly the moment the manager would rather do nothing. CLO.
• Choosing, and paid for it: crossover funds. Mandates written to hold both sides of the boundary exist precisely because the forced selling at it is predictable.
• Choosing: the issuer. Refinancing is a decision, and the maturity wall — when a lot of it falls due at once — is a calendar everybody can read in advance.

What a bad day looks like here

• The shape of it: liquidity leaves before the fundamentals do. Nothing has defaulted, and nothing can be sold in size at anything near the marked price.
• The first tell: the cash-CDS basis. When protection moves and the bonds do not, the derivative is telling you what the cash market is too slow to.
• The second tell: new issues pulled. An issuer that decides not to come to market has more information about demand than any spread quote.
• The third tell: the hedge moving more than the position it protects. Index protection is the liquid instrument, so it is where the fear is expressed first — and it can overshoot the thing it is standing in for.
• Where it has happened: 2008, and in a narrower form at the 2023 AT1 write-down, where the surprise was the order of the queue rather than the size of the loss.

How a trade actually happens here

Credit is three settlement systems wearing one name, and confusing them is the most common operational mistake in the asset class. A bond, a loan and a default swap on the same company move on three different clocks.

• A bond behaves like any other bond — quotes from dealers, delivery against payment at a depository a day or two later. High-yield bond.
• A loan is transferred, not delivered. The buyer takes an assignment of the lender's position, which needs the agent bank's paperwork and often the borrower's consent. It takes weeks, and the compensation for those weeks is calculated and paid separately rather than accruing by itself. Leveraged loan.
• A default swap starts as a document. It runs under a master agreement signed long before this trade; the trade fixes the reference entity, the maturity and a standardised coupon, and an upfront payment squares the difference between that coupon and the market spread. CDS.
• Index products are cleared; single names often are not. A cleared position faces a central counterparty and posts margin daily. An uncleared one faces the dealer, collateralised under an annex to the master agreement — the same economics, a different exposure. Margin and collateral.
• When the credit event arrives, a committee decides whether one occurred and an auction of the defaulted debt sets a single recovery price. Every contract then settles in cash at that number, which is how thousands of separate bilateral trades all close at the same price on the same day.
• When it fails, it usually fails on documentation rather than delivery: the wrong reference obligation, a successor entity after a merger, or a restructuring that counts as a credit event under one region's standard terms and not under another's.

Where the spread is, and who earns it

This is the only class on the site where the word spread means two different things in the same sentence — the price of the protection, and the dealer's margin on top of it. Keeping them apart is most of reading a quote here.

• The spread is the product. What the protection buyer pays per year, in basis points, is the market's price for the risk. The dealer's bid-offer sits around that number and is a small fraction of it.
• Upfront against running. Contracts trade with a fixed coupon and a cash payment at the start that reconciles it to the market spread. The same economics quoted two ways, and the conversion between them is a model rather than arithmetic. Credit spreads.
• The basis is where the money is made and lost. The same credit has a price in the bond market and a price in the contract market, and the two differ. A dealer hedging one with the other earns or pays that difference; it widens under stress, which is when the hedge is being relied on. CDS.
• Index against single name. An index trades far tighter than the sum of its parts trades individually, because it is one liquid instrument rather than a hundred illiquid ones. Hedging a specific name with an index is cheaper and is not the same hedge.
• The auction, which is not a spread at all. When a credit event happens, the recovery is discovered in a formal auction rather than negotiated. The cost of the mechanism falls on whoever has to trade in it.

How a position here ends

Every other class ends when a price is agreed. This one can end because a committee reads a document and decides what a word meant.

• It matures with nothing having happened. The protection buyer paid every quarter for a payout that never came, which is the intended outcome for insurance and is still a loss on the position.
• A credit event is determined. A determinations committee decides whether what happened meets the definition — bankruptcy, failure to pay, or a restructuring. The definition is the instrument. The 2023 AT1 write-down is what happens when a document says something nobody had read that way.
• The auction settles it. Once an event is determined, an auction sets one recovery price for everybody, and contracts cash-settle against it. Nobody has to find a bond to deliver.
• It is unwound or assigned. Closed at a price with the original counterparty, or novated to a new one with consent. Novation is the ordinary way a position changes hands.
• The reference entity ceases to exist. A merger, a demerger, a transfer of the debt — and succession rules decide what the contract now protects. This is an ending that changes what you own rather than removing it.
• The index rolls. A new series starts every six months. The old one stays alive, stays tradable, and becomes steadily thinner — an ending by attrition rather than by date.

Which risk decides across this class

Every product page here carries the same five bars. Read down the class instead of across one instrument, the question changes: is this a shelf of things that fail the same way, or a shelf of things that only share a department?

Which of the five decides what, across these 9

Counted from the same table each product page prints, so the two cannot disagree. Not a rating and not a ranking: it says which failure mode drives the outcome, not how dangerous anything is.

• Market3 of 9The price of the thing moves.Decides: CDS Index, CDS Option, Credit-Linked Note. Matters on 6 more.
• Credit8 of 9Somebody who owes you does not pay.Decides: Asset Swap, CDO & Synthetic Tranches, CDS Index, CLO, Credit Default Swap, Credit-Linked Note, Factoring & Receivables Finance, Leveraged Loan. Matters on 1 more.
• Liquidity3 of 9You cannot get out at anything near the marked price.Decides: CDO & Synthetic Tranches, CLO, Leveraged Loan. Matters on 3 more.
• Funding0 of 9Cash is needed before the position pays off — margin, calls, rolls.Decides: none of them. Matters on 5 more.
• Operational2 of 9The failure is in documents, systems, keys or people, not in prices.Decides: Asset Swap, Credit Default Swap. Matters on 4 more.

Credit decides 8 of the 9 — which is what makes this a class rather than a list: the instruments differ in shape and fail the same way. Funding decides nothing here — which is not the same as being absent.

The same five read across all 129 instruments →

Go deeper, and practiseLonger pieces, what an interview asks here, and a page to print.

Go deeper

Interactive: CDS–bond basisHard

The same default risk trades in two markets: the bond's spread and the CDS premium. The gap between them — the basis — is credit's classic relative-value signal.

Bond spread (bp) CDS spread (bp)

Basis (CDS − bond)

—

Reading

—

Negative-basis package earns

—

Negative basis: buy the bond, buy CDS protection, earn the difference with default risk (theoretically) hedged — the trade that "can't lose" until funding vanishes, as 2008's basis blow-out taught a generation of prop desks.

How this market works

Analysis

Loss given default: the other half of credit risk

Default probability gets the attention; the loss when it happens decides the answer. Expected loss is the product, and a spread that does not cover it is not compensation:

What the symbols mean

• Pa price, or a present value
• Dduration: how far a bond's cash flows sit in the future
• Lleverage, or a loss given default

Interactive: expected loss and the breakeven spreadMedium

Exposure at default $ Annual default probability % Unsecured recovery % Extra recovery from collateral %

Loss given default

—

Expected loss

—

Expected loss in bp

—

Loss if it actually defaults

—

Breakeven spread

—

Reading

—

Senior unsecured recoveries have historically clustered near 40% and secured loans far higher — but recoveries fall exactly when defaults rise, because the assets are being sold into the same downturn. The breakeven spread here is the floor before any compensation for that correlation, for illiquidity, or for being wrong.

Distance to distress: the Altman Z-score

A 1968 discriminant model that has outlived far more sophisticated successors, largely because it is transparent and hard to game:

Interactive: Altman Z-scoreMedium

Working capital $m Retained earnings $m EBIT $m Market value of equity $m Sales $m Total assets $m Total liabilities $m

Z-score

—

Zone

—

Profitability contribution

—

Equity cushion contribution

—

Health warning

—

The weights come from US manufacturers in the 1960s and transfer poorly to banks, insurers and asset-light businesses — separate variants exist for those. Its real value is that the largest term is profitability and the second is the market's own view of the equity cushion, which is a defensible summary of what distress actually is.

Market mapWho is on the other side of your trade

Prices are made by participants with different jobs, different constraints and different reasons to trade. Knowing whose need you are meeting explains more about a market than any indicator.

• Insurers are the anchor buyers of investment-grade credit, matching long liabilities. They are rating-sensitive by regulation, which is why a downgrade to sub-investment-grade forces mechanical selling.
• CLO warehouses and funds are the dominant buyers of leveraged loans. Their own funding structure decides how much they can hold, so loan demand is a function of CLO issuance rather than of credit views.
• Credit ETFs converted a dealer-intermediated market into one with a visible daily price. That improved transparency and created a new mechanism: fund flows that must be met from a market with far less dealer inventory.
• Banks increasingly distribute rather than hold, using significant risk transfer to move exposure while keeping the relationship.
• Distressed funds are the natural buyers at the bottom — and they will not bid until the price reflects the workout, which is why falling knives fall a long way in credit.

Costly mistakesThe five mistakes that cost the most here

Not a list of ways to be clever — a list of the errors that recur, why each one is structural rather than careless, and which tool on this site settles it.

• Reading spread as free income. A spread is compensation for expected loss, illiquidity and being wrong. The expected-loss tool shows how much of it is not income at all.
• Assuming recoveries are stable. They fall exactly when defaults rise, because the assets are sold into the same downturn. Modelling a fixed 40% is modelling the good case.
• Trusting the rating over the documents. Covenants, ranking and security decide the outcome; the rating is an opinion about the average case.
• Buying a credit fund for the yield. In credit, high yield is the market telling you the distribution has a long left tail — not that you found value others missed.
• Ignoring liquidity in a downgrade. Forced sellers appear at rating boundaries and everyone knows where those boundaries are.

What an interview asks here

Credit questions test whether you can separate the probability of an event from the price of that probability, and whether you know what recovery actually depends on.

Try each one out loud before you open it. What is underneath is the shape of a complete answer, not a script — somebody who can produce those parts in their own words can also answer the four variations that follow, and somebody who has memorised a paragraph can answer one.

Q1What does a credit default swap actually do?

What it is checking. The instrument definition, and the answer has to reach the protection seller's position.

A complete answer contains:

• The buyer pays a periodic premium; the seller pays out if a defined credit event occurs on the reference entity.
• So the seller is economically long the credit — similar to owning the bond funded at the reference rate — without holding the bond.
• What counts as a credit event is defined in the documentation, and that definition has been litigated.
• Settlement is normally through an auction that establishes a single recovery price for everybody.
• It is the cleanest way to take a view on credit without taking a view on rates, which is why it exists.

Read it properly: Credit default swap · Credit

Q2A five-year spread is 200 basis points. What default probability does that imply?

What it is checking. A mental-arithmetic question with a standard approximation and an important caveat.

A complete answer contains:

• The rough relationship is spread divided by loss given default, giving an annual hazard rate.
• At 200 basis points with a 40% recovery, loss given default is 60%, so the hazard rate is about 3.3% a year.
• Over five years the cumulative probability is roughly one minus the survival probability, so around 15%.
• The caveat matters: this is a risk-neutral probability, which is higher than the real-world one because it contains a risk premium.
• So quoting it as a forecast of default is wrong, and saying why is the point of the question.

Read it properly: The CDS calculator · Credit spreads

Q3What actually determines recovery?

What it is checking. Whether the candidate gives a table or a mechanism.

A complete answer contains:

• Seniority and security, first — but security is only worth what the collateral is worth and whether it was perfected.
• The enterprise value at the time, which is a going-concern number if a plan is funded and a liquidation number otherwise.
• The cost of the process, which comes out before anybody recovers.
• The jurisdiction: priority rules, employee and tax preferences, and whether cram-down is available.
• And increasingly the document: a drop-down can move the collateral out of the security package before anybody defaults.

Read it properly: The recovery calculator · The uptier transactions

Q4Why might a bond and its CDS imply different spreads?

What it is checking. The basis, and it is a good test of whether somebody has thought about funding.

A complete answer contains:

• A bond position has to be funded; a CDS position largely does not, so funding costs drive a wedge.
• Deliverability and the cheapest-to-deliver option affect the CDS side.
• Different documentation: what counts as a credit event in the CDS may not match what the bond's holders experience.
• Counterparty risk on the CDS, and the collateral terms behind it.
• And plain supply and demand — a large hedging flow in one instrument moves it relative to the other.

Read it properly: The basis calculator · Credit default swap

Q5What is a rating actually telling you?

What it is checking. Whether the candidate knows what the ordinal scale measures and what it deliberately does not.

A complete answer contains:

• An opinion about relative creditworthiness — for most agencies, an ordinal ranking of default probability or expected loss.
• It is not a price, not a recommendation, and not a statement about volatility or liquidity.
• It moves slowly by design, which is why the market's spread moves before the rating does.
• It matters mechanically because mandates, index rules and capital requirements reference it — so a downgrade can force selling regardless of anybody's view.
• That mechanical channel is the reason a rating action is a price event even when it contains no new information.

Read it properly: What a rating tells you · Credit

Q6Why does credit behave differently from equity in a downturn?

What it is checking. A shape question. The answer is about the payoff, not about sentiment.

A complete answer contains:

• Credit's upside is capped at par plus coupons — it can only be repaid, never more.
• Its downside is a default, so the distribution is a small chance of a large loss against a high chance of a small gain.
• That is a short-option shape, which is why credit returns are negatively skewed and look calm until they do not.
• Equity has the opposite shape: bounded loss and unbounded gain, so it reprices earlier and more continuously.
• Which is why the credit market frequently moves later and then all at once.

Read it properly: Bond vs. share · Credit spreads

Q7A company's spread widens sharply and its equity barely moves. What might be going on?

What it is checking. A cross-asset question, and there are several right answers.

A complete answer contains:

• A capital structure event: a debt-funded buyback or a dividend recap transfers value from lenders to shareholders.
• A leveraged acquisition announcement does the same thing, which is why change-of-control puts exist.
• Technical flows: an index exclusion or a large hedging trade in the credit with no equity counterpart.
• Or a documentation event — a liability management transaction that subordinates existing lenders without changing the enterprise.
• The general point: they are claims on the same business with different priorities, so a transfer between them moves one and not the other.

Read it properly: Dividend recap · Bond vs. share

Do these against a clock → — one at a time, ninety seconds each, answer before you look.

Whose questions these are. Every question on this page was written for this publication. None is taken from anybody else’s question bank, and none is a claim about what any named firm asks — that is neither verifiable from here nor ours to assert. They are our own reading of which mechanism a question of this kind is testing. Information and education only.

Test yourself: five questions

Five quick questions on this asset class — checked entirely on your device, nothing stored or sent. Wrong answers come with explanations; the deep dives above hold every answer. For education only.

Interactive: downgrade risk, not just default riskMedium

Most credit loss in a diversified portfolio comes from downgrades rather than defaults, because downgrades are far more common and the spread widening is immediate. This puts both on the same scale.

Current spread (bp) Spread if downgraded (bp) Probability of downgrade % Probability of default % Recovery % Spread duration

Spread income

—

Expected downgrade loss

—

Expected default loss

—

Net expected return

—

Breakeven spread

—

Reading

—

Why it matters

—

Raise the downgrade probability to 20% — a realistic figure for the bottom of the investment-grade band in a recession — and watch the net return. This is the arithmetic behind the fallen-angel effect: the forced selling that follows a downgrade across the investment-grade boundary is what makes that particular notch worth more than all the others combined. See how to read a credit rating.

Interactive: distance to a contingent-convertible triggerMedium

An AT1 bond converts or writes down when the issuer's capital ratio falls to a stated level. This measures how far away that is, both in percentage points and in the losses it would take to get there.

CET1 ratio % Trigger level % Risk-weighted assets $bn Coupon % Price

Common equity

—

Buffer above the trigger

—

Headroom

—

Losses to reach it

—

As a share of RWA

—

Running yield

—

Reading

—

The important caveat

—

The caveat line is the whole lesson. The mechanical trigger is almost never what fires: coupon cancellation is discretionary and comes earlier, supervisory intervention comes earlier still, and in 2023 an entire AT1 stack was written off while shareholders below it received stock — because the documentation said so. See the AT1 write-down and the bail-in waterfall.

⬇ The whole class on one printable page

Who pays whom, drawn

The payoff charts on this site say what an instrument is worth at the end. These say who the parties are and what each one hands over — every payment between the investor and the bank on the other side, in the order it happens. Each diagram sits on its product's own page. 1 of them carry a second diagram, folded shut, for what happens after the bargain is struck — clearing, delivery, custody. That half is deliberately separate: a clearing house is not a party to the bargain.

• Credit Default Swap — Insurance on a company you need not own
• Leveraged Loan — What the borrower pays, and what the lender is exposed to
• Factoring & Receivables Finance — Selling the invoice rather than waiting for it
• Credit-Linked Note — A bond wrapped around a default swap
• CLO — A manager, a waterfall and a test that forces its hand
• Asset Swap — Separating the credit from the interest rate

Who does this: Credit Derivatives is quoted from five sell-side seats — Sales, Trading, Structuring, Research, Prime Services — and held from the buy-side by Asset Management, Private Markets, Hedge Funds & Alternatives, Wealth Management, Insurance & Pensions. See the industry map.