Every morning, in every investment bank, a number lands in the inbox of the Head of Trading, the CFO, and the Chief Risk Officer: yesterday’s PnL. The number itself takes seconds to read. What takes skill is the short paragraph underneath it that answers the only question senior people really care about: why?

That paragraph is the daily PnL commentary. It is the most widely read piece of writing a Product Controller produces, and for many readers it is the only thing they ever see from Product Control. A good commentary tells the reader what the desk was positioned for, what the market did, and whether the result makes sense. A poor one says “PnL driven by market moves” and leaves everyone to guess.

In Chapter 6, Front Office Desks in an Investment Bank, we looked at what each desk does, how it makes money, and what risks it runs. This chapter builds directly on that foundation, because you cannot explain a desk’s PnL unless you understand how the desk earns it. A cash equities desk earns commissions and facilitation gains. A rates desk earns from curve positioning and risk warehousing. The commentary has to speak the language of the desk it describes.

We will cover:

  • what a daily PnL commentary is, and what it is not,
  • why it matters to Management, Risk, and Regulators,
  • the building blocks of PnL that every commentary draws on,
  • ten best practices that separate strong commentary from weak,
  • worked examples that climb from the simplest asset class to the most complex: Cash Equities, Prime Brokerage, FX, Credit, Equity Derivatives, Rates, and Macro,
  • common mistakes, a daily workflow, and a checklist you can use tomorrow morning.
Practitioner note

All desks, positions, prices, and PnL figures in this chapter are illustrative. They are constructed so that the arithmetic can be followed line by line. Follow the method, not the specific numbers.

◆
SECTION 01

What is a daily PnL commentary?

The daily PnL commentary is a short written explanation of the drivers of a desk’s profit or loss for one trading day, produced by Product Control and agreed with the desk. It sits alongside the official PnL report and the PnL attribution (also called PnL explain), which breaks the day’s result into its sources.

It helps to be clear about the three things that travel together each day:

DeliverableWhat it isWho owns it
Flash / estimate PnLThe trader’s own end-of-day estimate of the day’s result (T+0).Front Office
Official PnL and attributionThe independently produced and reconciled PnL, split by driver (T+1 morning).Product Control
PnL commentaryThe narrative that explains the attribution in plain business language.Product Control, agreed with the desk

The attribution gives the numbers. The commentary gives the story that connects the numbers to positions and to the market. A commentary is therefore not:

  • a restatement of the attribution table in sentence form (“delta was −1.2, vega was +2.0”),
  • a market wrap copied from a news service with no link to the desk’s positions,
  • a transcript of what the trader said, passed on without independent checking.

The simplest test of a good commentary is this: could a senior reader who has not seen the desk’s positions reconstruct, roughly, why the desk made or lost this amount? If yes, the commentary has done its job.

◆
SECTION 02

Why the daily PnL commentary matters

It is tempting to see commentary as a daily chore. In practice it is one of the bank’s core control evidences. Different audiences read the same paragraph for very different reasons.

The Management perspective

Senior management uses daily PnL commentary to run the business. They want to know:

  • Is the desk making money the way it is supposed to? A flow desk that earns most of its PnL from client spread is behaving as designed. A flow desk that earns most of its PnL from one large directional position is a different business from the one management approved.
  • How good is the revenue? As Chapter 6 stressed, not all revenue is of equal quality. Client franchise income, carry, and repeatable spread are high quality. Day-1 gains on complex trades, reserve releases, and one-off remarks are not.
  • Is anything building up? Repeated small losses on the same position, growing inventory, or rising dependence on one client all show up first in daily commentary.
  • What do I need to act on today? Capital allocation, limit changes, and conversations with desk heads start from these explanations.

Commentary also feeds upward. Daily comments roll into weekly and month-end business reviews, CFO packs, and eventually into the narrative that supports quarterly results. Weak daily commentary means a painful month-end.

The Risk perspective

Market Risk reads the commentary to check that PnL is consistent with the risk the desk reported. Risk managers know the desk’s sensitivities (delta, DV01, CS01, vega) and its VaR. If the risk report says the desk is flat and the PnL swings by several million, something is wrong with the risk capture, the PnL, or both.

  • Backtesting. Daily PnL is compared against the prior day’s VaR. When a loss exceeds VaR, it is a backtesting exception, and each exception must be explained and documented. The daily commentary is the starting point for that explanation.
  • Limit monitoring. PnL commentary confirms whether results came from positions within mandate and within limits, and it supports stop-loss and drawdown monitoring.
  • Risk-not-in-VaR and model weaknesses. A persistent unexplained residual in the attribution is often the first evidence that a risk factor is missing from the risk model.
  • Concentration and liquidity. Commentary that names the specific positions behind the PnL lets Risk see concentration that aggregated sensitivities hide.

The Regulatory perspective

Regulators do not usually read each day’s commentary, but they expect it to exist, to be independent, and to be retrievable. Several regulatory frameworks lean on the same discipline:

  • Basel market risk backtesting. Banks using internal models compare one-day 99% VaR against actual and hypothetical PnL. The number of exceptions over 250 days places the bank in a green, yellow, or red zone and can increase the capital multiplier. Supervisors expect a documented reason for each exception.
  • FRTB and the PnL Attribution (PLA) test. Under the Fundamental Review of the Trading Book, a desk can use internal models only if its risk-theoretical PnL (RTPL) tracks its hypothetical PnL (HPL) closely, measured by statistical tests at desk level. A team that already explains daily PnL by risk factor is far better placed to understand and fix PLA failures.
  • The Volcker Rule. US rules restrict proprietary trading and require banks with significant trading activity to report desk-level metrics, including a comprehensive profit and loss attribution. Commentary that separates client-driven revenue from position-driven PnL supports the bank’s evidence that a desk is genuinely market making.
  • Fair value and Day-1 PnL. Accounting standards restrict the upfront recognition of profit on trades valued with significant unobservable inputs. Commentary that identifies Day-1 PnL and reserve movements gives Finance and auditors a clear trail.
  • Senior management accountability and financial reporting controls. Regimes such as the UK Senior Managers and Certification Regime and Sarbanes-Oxley make named individuals responsible for the integrity of reported numbers. Daily PnL sign-off, with commentary, is a key control in that chain.
WHY THIS IS A CONTROL, NOT A CHORE

The lesson from rogue trading cases. In the large unauthorised trading losses seen at some of the world’s top investment banks over the past two decades, a recurring theme was PnL that was unusual for the desk’s stated mandate and explanations that were accepted without enough independent challenge.

A low-risk arbitrage or client facilitation desk that produces large directional gains is a red flag even when the number is positive. A profit you cannot explain is as dangerous as a loss you cannot explain. The daily commentary is where that challenge is supposed to happen.

The Product Control perspective

Finally, commentary protects Product Control itself. Writing it forces the controller to prove that the PnL is right. If you cannot explain a number, you have probably found a booking error, a bad mark, a missing trade, or a broken feed. Many trade breaks and valuation issues are first discovered by a controller who could not make the commentary add up.

◆
SECTION 03

The building blocks of PnL

Before writing a word, a controller needs a clean attribution. Whatever the asset class, daily PnL can be split into a small number of standard building blocks. The names vary by bank, but the logic is universal.

Building blockWhat it capturesTypical commentary language
New deal / Day-1 PnLMargin earned on trades executed today: the difference between the client price and mid-market value.“Client franchise +$0.4m on $1.2bn volume”
Market moves on existing positionsRevaluation of yesterday’s positions for today’s moves in prices, rates, spreads, and volatility.“Long 10y DV01 of $50k lost $0.2m as yields rose 4bp”
Carry and timeCoupons, dividends, accrued interest, funding cost, roll-down, and option theta.“Carry +$0.1m; theta −$0.3m”
FX revaluationTranslation of non-reporting-currency positions and accumulated PnL.“FX reval of EUR PnL −$0.02m”
Fees and commissionsExplicit execution, clearing, and service income; brokerage and exchange costs.“Commissions +$0.9m at 4.5bp”
Valuation adjustments and reservesBid-offer, model, liquidity, and XVA reserves taken or released.“Bid-offer reserve increased $0.2m on wider spreads”
Trade amendments and late bookingsCancels, corrections, and trades booked after their trade date.“Late-booked trade from T−1, +$0.1m”
Unexplained / residualThe part of PnL the attribution cannot assign to any driver.“Unexplained −$0.06m, within tolerance”

Two points about this table are worth remembering throughout the chapter.

First, the mix changes as products become more complex. For cash equities, almost all PnL sits in the first two rows and the fee row. For an options or rates desk, every row is active every day, and the “market moves” row itself splits into delta, gamma, vega, curve, and basis. That is why commentary gets harder as we move up the ladder.

Second, the unexplained line is a control in its own right. A small residual is normal, because sensitivities are approximations. A large or persistent residual means the attribution, the risk, or the valuation is incomplete, and it should be investigated and escalated against a defined tolerance, not quietly absorbed.

◆
SECTION 04

Ten best practices for writing PnL commentary

1. Lead with the headline

Start with the number, the direction, and the one or two biggest drivers. A senior reader may stop after the first sentence, so that sentence must carry the message: “Desk PnL +$1.28m, driven by client commissions and a block facilitation gain, partly offset by inventory losses.”

2. Use the formula: position × market move = PnL

This is the single most important habit. Every market-driven line should state what the desk held, what the market did, and what that produced. “Lost on rates” is not an explanation. “Long 2-year DV01 of $85k gained $0.77m as 2-year yields fell 9bp” is one, and the reader can check the arithmetic in their head.

3. Rank by materiality

Explain the largest drivers first and in the most detail. As a rule of thumb, the commentary should explain at least 80% to 90% of the absolute PnL by named drivers. Small items can be grouped as “other”. Most banks set explicit thresholds, for example comment on any item above a fixed amount or above a percentage of desk PnL.

4. Separate client franchise from risk PnL

Always distinguish what the desk earned from serving clients (spread, commissions, structuring margin) from what it earned from holding positions (inventory, directional and relative value risk). This one split tells management about revenue quality, tells Risk whether results match exposures, and supports the bank’s market-making evidence.

5. Anchor every market reference to independent data

Quote actual moves from independent sources: “S&P 500 −2.1%”, “EUR/USD −60 pips”, “IG spreads +6bp”. Never rely only on the trader’s description of the market. If the commentary says the desk gained because spreads tightened, check that spreads did tighten.

6. Explain by risk factor, and state the unexplained

Tie PnL to the desk’s reported risk sensitivities. If the risk report shows a net DV01 of $5k and the desk made $0.5m on a 5bp move, the commentary must explain the gap, usually a curve or basis position. Report the unexplained amount openly, with its tolerance.

7. Call out everything that is not “normal trading”

Day-1 PnL on new structured trades, reserve releases, model or curve changes, remarks, cancels and amendments, late bookings, and operational errors should each be named separately. These items distort the picture of the day’s trading and are exactly what auditors and regulators look for.

8. Stay independent

The trader’s view is an essential input, but the commentary is Product Control’s document. Verify before you write. Where the trader’s explanation and the data disagree, say so and follow up. Phrases like “per trader” are a warning that the controller has not checked.

9. Write for a reader outside the desk

Use plain language, consistent structure, consistent sign conventions, and defined abbreviations. The commentary should stand alone, without the reader needing the attribution spreadsheet. Keep the same template every day so that readers know where to look.

10. Be timely, and close the loop

A perfect commentary delivered at 4pm is worth less than a good one delivered at 9am. Where something is still under investigation, say so, state the amount, name the owner, and update the reader when it is resolved. An open item that silently disappears damages trust more than the original break.

Words that weaken a commentary

AvoidWhy it failsWrite instead
“PnL driven by market moves”Says nothing about position or direction.“Long $60m delta lost $1.26m as the index fell 2.1%”
“Various small items”Hides the composition; may conceal a real issue.“Other −$0.07m: funding −$0.04m, trade error −$0.03m”
“In line with expectations”Whose expectations, and based on what?“Consistent with reported vega of $0.8m per vol point”
“Per trader” / “trader advised”Signals the explanation was not independently verified.State the verified position and move; note any open query.
“Noise” / “timing”Used to wave away unexplained amounts.“Unexplained −$0.06m (tolerance $0.25m)”
“Desk was well positioned”A judgement, not an explanation.State the position that produced the gain.

A template that works for any desk

Most strong commentaries follow the same five-part structure. The worked examples below all use it.

  1. Headline. Total PnL, direction, and the top one or two drivers in one sentence.
  2. Client franchise / new business. Volumes, margins, notable trades, Day-1 PnL.
  3. Risk and inventory. Position × market move for each material risk factor, with hedges shown against the positions they hedge.
  4. Carry, reserves, and one-offs. Funding, theta, valuation adjustments, amendments, late bookings, errors.
  5. Unexplained and open items. Residual against tolerance, anything under investigation, and any limit, VaR, or mandate observation.
◆
SECTION 05

Worked examples: from simple to advanced

We now apply these practices desk by desk. The order is deliberate. Each step up the ladder adds a new layer that the commentary has to handle.

StepDeskNature of PnLNew skill the commentary needs
1Cash EquitiesLinear. Price × quantity.Splitting commissions, facilitation, and inventory.
2Prime BrokerageAccrual. Balances × spread × days.Explaining carry, day count, and balance changes.
3FX (spot and forwards)Linear, multi-currency.Currency pair conventions; gross drivers hidden by a small net.
4Credit TradingSpread risk plus rates plus carry.CS01, hedges and basis, issuer events, reserves.
5Equity DerivativesNon-linear.Greeks: delta, gamma, vega, theta; Day-1 and reserves.
6RatesCurve and basis.Bucketed DV01, curve shape, why net DV01 misleads.
7Macro / FICCCross-asset.Explaining by theme across rates, FX, and commodities.

For each desk we show what drives the PnL, a sample day with its attribution, a weak commentary, a strong commentary, and the checks a controller should run.

5.1 Cash Equities: the starting point

Cash equities is the right place to learn commentary because the PnL is linear and transparent. Shares have observable closing prices, and PnL on a position is simply quantity × price change. As Chapter 6 explained, the desk earns from commissions, bid-ask spread, block and facilitation gains, and short-term inventory.

What drives the PnL

  • Commissions and execution fees on agency flow: volume × commission rate.
  • Facilitation and block trades: the discount or spread earned for providing immediacy, less the cost of unwinding.
  • Inventory: price moves on residual positions, and the performance of index or sector hedges against them.
  • Financing, stock borrow, and dividends, plus occasional trade errors.

The day

Equity markets were modestly weaker: the broad index fell 0.4% and the bank sector fell 0.6%. Client volumes were elevated on index rebalance flow. The desk bought a block of 3.0 million shares in ABC Industries from a client at $19.70, a 1.5% discount to the prior close of $20.00.

PnL driverPosition / riskMarket movePnL ($m)
Commissions$1.9bn executed notionalAverage 4.5bp+0.86
Block: realised2.2m shares sold at $19.92 vs $19.70 cost—+0.48
Block: unrealised0.8m shares held, cost $19.70Close $19.85+0.12
Bank-sector inventoryLong $45m basketSector −0.6%−0.27
Index hedgeShort $40m index futuresIndex −0.4%+0.16
Funding and stock borrow——−0.04
Trade error25k shares, wrong side—−0.03
Total+1.28
Weak commentary

Cash Equities +$1.28m. Good day on client flow and a block trade. Small losses on inventory as markets were weaker.

Strong commentary

Cash Equities +$1.28m, driven by client commissions (+$0.86m) and the ABC Industries block (+$0.60m), partly offset by residual bank-sector inventory (−$0.11m net of hedge).

Client franchise +$0.86m: $1.9bn executed notional at an average 4.5bp. Volumes were above recent averages on index rebalance flow.

Block facilitation +$0.60m: bought 3.0m ABC Industries shares from a client at $19.70, a 1.5% discount to the prior close of $20.00. 2.2m shares were sold intraday at an average $19.92 (+$0.48m realised). The remaining 0.8m shares are marked at the $19.85 close (+$0.12m unrealised). The residual $15.9m long is within the single-stock limit.

Inventory −$0.11m: the $45m long bank-sector basket lost $0.27m as the sector fell 0.6%. The $40m short index futures hedge gained $0.16m as the index fell 0.4%. The hedge is imperfect: the sector underperformed the index by 0.2%.

Other −$0.07m: funding and stock borrow −$0.04m; one trade error −$0.03m (25k shares executed on the wrong side, corrected same day and logged as an operational risk event).

No reserve movements or late bookings. Unexplained: nil.

What the controller should check

  • Commission PnL against volume. Does volume × average rate reproduce the commission line? A mismatch often points to missing or duplicated trades.
  • Block economics. Verify the discount, the quantity sold, and that the residual is marked at the official close, not at the trader’s view of where it can be sold.
  • Unrealised versus realised. $0.12m of the block gain is still at risk. Saying so tells the reader the story is not finished.
  • Hedge effectiveness. Showing the position and its hedge side by side reveals the basis risk that a net number would hide.
  • Errors are named, not netted. Even $0.03m is disclosed because it is a control event, not trading.

5.2 Prime Brokerage: explaining an accrual business

Prime brokerage PnL is mostly carry, not price movement. Chapter 6 described it as a financing and service business: margin lending, securities lending, and synthetic financing. Daily PnL is close to balances × spread × days, so the commentary is about balances, rates, and anything that breaks the run-rate.

What drives the PnL

  • Financing spread on client debit balances and synthetic (swap) financing.
  • Securities lending: fees on general collateral and on hard-to-borrow “specials”.
  • Day count: weekends and holidays accrue on the preceding business day, so Friday PnL is typically three days of carry.
  • Repricing and corrections: client rate changes, backdated adjustments, and balance movements.

The day

It is a Friday, so the PnL contains three days of accrual. One client’s financing spread was found to have been set up at 70bp instead of the agreed 50bp, and the over-accrual of the past 30 days was reversed.

PnL driverPosition / riskMarket movePnL ($m)
Margin lending$9.0bn debits at 65bp spread3 days’ accrual+0.49
Synthetic financing$6.0bn at 50bp spread3 days’ accrual+0.25
Securities lending: GC$4.0bn at 45bp net fee3 days’ accrual+0.15
Securities lending: specials$150m at 12% fee3 days’ accrual+0.15
Rate correction$500m balance, 20bp over-accrued30 days reversed−0.08
Total+0.96
Weak commentary

Prime Brokerage +$0.96m. Higher than usual due to the weekend. Financing and stock loan revenues in line.

Strong commentary

Prime Brokerage +$0.96m, which is three days of accrual (Friday to Sunday) at a run-rate of about $0.35m per day, less a one-off rate correction of −$0.08m.

Financing +$0.74m: margin lending +$0.49m on $9.0bn of client debits at an average 65bp spread; synthetic financing +$0.25m on $6.0bn at 50bp. Balances are broadly unchanged on the week.

Securities lending +$0.30m: general collateral +$0.15m on $4.0bn of shorts at 45bp. Specials +$0.15m on only $150m of balances at an average 12% fee. Specials are under 4% of lending balances but half of lending revenue, concentrated in a small number of names.

One-off −$0.08m: reversal of 30 days of over-accrued spread on one client ($500m balance booked at 70bp against a contractual 50bp). Static data has been corrected; the root cause is with Client Onboarding and Operations.

Underlying daily run-rate is unchanged. Unexplained: nil.

What the controller should check

  • Normalise for day count. Always translate a multi-day accrual back into a daily run-rate, otherwise every Friday looks like a record day.
  • Reconcile balances × rate × days to the booked accrual. This simple reasonableness check catches static data errors like the one above.
  • Watch revenue concentration. A few hard-to-borrow names can dominate lending income, and that income disappears when the special ends.
  • Separate run-rate from one-offs. Corrections and backdated repricing belong on their own line with a root cause.

5.3 FX spot and forwards: a small number can hide a big story

FX is still a linear product, but it introduces currency pair conventions and a new lesson: the size of the net PnL says nothing about how busy or risky the day was. FX desks earn from high-volume spread capture and from positions carried in major and emerging market pairs.

What drives the PnL

  • Client spread on spot, forwards, and swaps: volume × margin.
  • Spot positions: position in base currency × move in the exchange rate. Note which currency the PnL arises in.
  • Forward points and carry: interest rate differentials on forward and swap positions.
  • Amendments and revaluation of PnL held in non-reporting currencies.

The day

The dollar strengthened against the euro and weakened against the yen. EUR/USD fell from 1.0850 to 1.0790 (−60 pips). USD/JPY fell from 151.20 to 150.40 (−80 pips). The desk carried a long EUR/USD position of €150m and a short USD/JPY position of $80m.

PnL driverPosition / riskMarket movePnL ($m)
Client franchise$4.2bn client volumeAverage 1.0bp margin+0.42
EUR/USD positionLong €150m1.0850 to 1.0790 (−60 pips)−0.90
USD/JPY positionShort $80m151.20 to 150.40 (−80 pips)+0.43
Forward points / carryForward book—+0.05
Trade amendment€120m forward, rate corrected by 10 pips—−0.12
Total−0.12
Weak commentary

FX −$0.12m. Quiet day, broadly flat. Client flow offset by small position losses.

Strong commentary

G10 FX −$0.12m. The small net result masks offsetting drivers: client franchise +$0.42m, a −$0.90m loss on long EUR/USD, a +$0.43m gain on short USD/JPY, and a −$0.12m trade amendment.

Client franchise +$0.42m: $4.2bn of client volume at an average margin of 1.0bp, weighted towards corporate forward hedging.

Positions −$0.47m: long €150m EUR/USD lost $0.90m as the pair fell 60 pips from 1.0850 to 1.0790 (€150m × 0.0060). Short $80m USD/JPY gained ¥64m, or $0.43m, as the pair fell 80 pips from 151.20 to 150.40. The desk was long EUR and long JPY against the dollar; the two legs moved in opposite directions.

Carry +$0.05m on the forward book.

Amendment −$0.12m: a €120m client forward booked on T−1 at 1.0825 was corrected to the confirmed rate of 1.0835. The original booking overstated T−1 PnL by the same amount. Booking error raised with Middle Office.

Unexplained: nil. The EUR/USD position was within the desk’s overnight limit.

What the controller should check

  • Gross before net. A −$0.12m day contained a $0.90m single-position loss. Management and Risk need to see it.
  • Convention and currency. USD/JPY PnL arises in yen and must be converted. State the position in the base currency and the move in the quoted terms.
  • Amendments reverse earlier PnL. Link the correction to the day it relates to. Repeated amendments from the same desk or client are a pattern worth escalating.
  • Client margin reasonableness. An implied margin that is unusually wide may mean position PnL has been misclassified as franchise.

5.4 Credit Trading: spreads, hedges, and events

Credit adds several layers. Bond prices move with both interest rates and credit spreads, positions are often hedged with indices that do not move one-for-one, individual issuers can jump on news, and liquidity is thinner, so reserves start to matter. The key risk measure is CS01: the PnL for a 1bp move in credit spread.

What drives the PnL

  • Spread moves on bond and CDS inventory: CS01 × spread change.
  • Hedges and basis: index CDS hedges against cash bonds, and the bond-CDS basis between them.
  • Issuer-specific events: downgrades, earnings, restructuring headlines.
  • Carry: coupon accrual less funding. Client flow: bid-offer on trades.
  • Reserves: bid-offer and liquidity adjustments that move with market conditions.

The day

A risk-off session. Investment grade (IG) cash spreads widened 6bp and high yield (HY) widened 18bp. The IG and HY CDS indices widened by less: 5bp and 15bp. Government yields fell 3bp. One HY issuer held by the desk was downgraded and its bonds fell 4 points.

PnL driverPosition / riskMarket movePnL ($m)
IG bond inventoryCS01 −$95k/bp (long credit)IG spreads +6bp−0.57
HY bond inventoryCS01 −$40k/bp (long credit)HY spreads +18bp−0.72
IG index hedgeCS01 +$70k/bp (short risk)Index +5bp+0.35
HY index hedgeCS01 +$25k/bp (short risk)Index +15bp+0.38
Issuer eventLong $15m face, HY issuerBonds −4 points−0.60
Interest rate residualDV01 +$5k/bpYields −3bp+0.02
Carry——+0.11
Client flow$310m traded—+0.24
Bid-offer reserveWider market bid-offer—−0.18
Total−0.97
Weak commentary

Credit Trading −$0.97m. Losses as credit spreads widened in a risk-off market. Hedges helped. One name was downgraded.

Strong commentary

Credit Trading −$0.97m, driven by spread widening on bond inventory net of index hedges (−$0.56m) and a downgrade in one HY issuer (−$0.60m), partly offset by client flow and carry (+$0.35m).

Spread risk −$0.56m net: IG inventory (CS01 $95k/bp) lost $0.57m as IG spreads widened 6bp; HY inventory (CS01 $40k/bp) lost $0.72m on 18bp of widening. Index hedges gained $0.73m (IG index +5bp, HY index +15bp). Hedges covered 57% of the inventory loss. The shortfall reflects two things: the desk runs a net long credit position by design, and cash bonds underperformed the indices by 1bp in IG and 3bp in HY.

Issuer event −$0.60m: the desk’s $15m face holding in one HY issuer fell 4 points following a rating downgrade. This is idiosyncratic risk that index hedges do not cover. The position is within the single-issuer limit; the closing price was verified against two independent quotes.

Client flow +$0.24m on $310m traded. Carry +$0.11m. Interest rate residual +$0.02m (yields −3bp on a DV01 of $5k/bp).

Reserves −$0.18m: the bid-offer reserve increased as observed market bid-offer widened across HY. This is a methodology-driven move, not a trading loss.

Unexplained: nil after rounding.

What the controller should check

  • State sign conventions. “Long credit” loses when spreads widen. Writing CS01 with its direction avoids the most common misreading in credit commentary.
  • Show the hedge ratio and the basis. “Hedges helped” becomes useful only when the reader knows they covered 57% and why.
  • Separate systematic from idiosyncratic. A market-wide spread move and a single-issuer downgrade are different risks with different owners.
  • Price verification on event names. After a downgrade, liquidity drops and marks become judgemental. This is where the IPV team’s view matters most.
  • Reserve moves are not trading PnL. Explain what changed in the inputs, and confirm it follows the approved methodology.

5.5 Equity Derivatives: explaining non-linear PnL

With options, PnL is no longer position × price move. It depends on how far the market moved, how fast, and what happened to volatility. Chapter 6 noted that equity derivatives PnL comes from structuring margins, hedging, and volatility and correlation exposure. The commentary now has to translate the Greeks into plain language.

What drives the PnL

GreekWhat it measuresHow to describe it
DeltaSensitivity to the underlying price.“Long $60m delta lost as the index fell.”
GammaChange in delta as the underlying moves. Long gamma gains from large moves in either direction.“Long gamma gained on the size of the move.”
VegaSensitivity to implied volatility.“Long vega gained as implied vol rose.”
ThetaTime decay. The daily cost of owning options.“Paid $0.4m of time decay.”
Skew / cross-GreeksChanges in the shape of the volatility surface; vanna and volga effects.“Long downside puts gained as skew steepened.”

On top of these sit dividend and repo exposure, Day-1 PnL on new structured trades, and model and bid-offer reserves.

The day

A sharp sell-off. The equity index fell 2.1% and one-month implied volatility rose 2.5 points, with downside skew steepening. The desk started the day long $60m of delta, long gamma, and long vega. It also issued a $50m structured note to a private bank distributor.

PnL driverPosition / riskMarket movePnL ($m)
DeltaLong $60m at openIndex −2.1%−1.26
GammaLong $45m delta per 1% move2.1% move+0.99
VegaLong $0.8m per vol pointImplied vol +2.5 pts+2.00
SkewLong downside putsSkew steeper+0.30
Cross-Greeks (vanna/volga)—Spot down, vol up+0.15
ThetaLong options1 day−0.42
Dividends / repo——−0.05
Day-1 PnL$50m structured note1.2% margin+0.60
Reserves on new tradeModel and bid-offer—−0.20
Unexplained−0.06
Total+2.05
Weak commentary

Equity Derivatives +$2.05m. Desk was well positioned for the sell-off. Gains on vega and gamma offset delta losses. New structured trade also contributed.

Strong commentary

Equity Derivatives +$2.05m. The desk was long volatility into a 2.1% index sell-off: gains on vega (+$2.00m) and gamma (+$0.99m) outweighed the loss on opening delta (−$1.26m). New issuance added +$0.40m net of reserves.

Volatility +$2.45m: long vega of $0.8m per vol point gained $2.00m as one-month implied volatility rose 2.5 points. Downside skew steepened, adding $0.30m on long put positions, and cross-Greek effects added $0.15m.

Delta and gamma −$0.27m net: the opening long delta of $60m lost $1.26m on the 2.1% fall. Long gamma of $45m per 1% recovered $0.99m (½ × $45m × 2.1² ÷ 100), as the book’s delta shortened into the fall and was re-hedged.

Time decay and carry −$0.47m: theta −$0.42m, the daily cost of the long option position; dividend and repo −$0.05m.

New business +$0.40m net: $50m structured note issued at a 1.2% margin (+$0.60m Day-1), less $0.20m of model and bid-offer reserves taken at inception. Inputs are observable; no Day-1 deferral required.

Unexplained −$0.06m, within the desk tolerance.

What the controller should check

  • Tell the positioning story first. “Long volatility into a sell-off” lets a non-specialist understand the whole day before any Greek is mentioned.
  • Reconcile to the risk report. Opening delta, gamma, and vega in the commentary must match what Market Risk saw the night before.
  • Gamma needs the size of the move. Gamma PnL grows with the square of the move, so always quote the move alongside it.
  • Theta is the price of the position. A long-gamma desk that reports gains on big days but never mentions theta on quiet days is telling half the story.
  • Day-1 and reserves are shown gross. Report the margin, the reserve, and whether deferral applies. Netting them hides revenue quality.
  • Unexplained grows on large-move days, because Greeks are local approximations. That is expected, but it must be measured against tolerance, and full revaluation should confirm it.

5.6 Rates: when the net number misleads

Rates is where many new controllers struggle, because the same market can move in several ways at once. Yields at different maturities rarely move by the same amount. Swaps, government bonds, and futures move relative to one another. And as Chapter 6 noted, rates PnL comes from market making, risk warehousing, structuring, and relative value and curve positioning. A rates commentary that says “rates rallied” has explained almost nothing.

What drives the PnL

  • Outright duration: net DV01 × parallel move in yields.
  • Curve: DV01 by maturity bucket × the move at each point. Steepeners, flatteners, and butterflies make or lose money with no net DV01 at all.
  • Basis: swap spreads (bond versus swap), tenor basis, cross-currency basis, and futures basis.
  • Volatility: vega, gamma, and theta on swaptions and caps/floors.
  • Carry and roll-down, new client trades, and XVA charges passed to the XVA desk.

The day

Weak economic data triggered a bull steepening: front-end yields fell sharply while long-end yields barely moved. The desk was positioned for exactly this through a 2s10s steepener: received in the 2-year sector and paid in the 10-year sector, with almost no net DV01. In the table, a positive DV01 means the desk gains when yields fall.

BucketDV01 ($k/bp)Yield move (bp)PnL ($m)
2-year+85−9+0.77
5-year+20−7+0.14
10-year−110−4−0.44
30-year+10−1+0.01
Net+5+0.48
PnL driverPosition / riskMarket movePnL ($m)
Curve / delta (table above)2s10s steepener, net DV01 +$5kBull steepening+0.48
Swap spreadsLong bonds vs pay swaps, $30k/bpSpreads +2bp+0.06
Cross-currency basis−$12k/bpBasis moved 1.5bp against−0.02
Swaption vegaLong $60k per bp of normal volImplied vol +3bp+0.18
Swaption gammaLong gammaLarge front-end move+0.07
Swaption thetaLong options1 day−0.09
Carry and roll-down——+0.05
New client trade$200m 10y swap, DV01 $170k0.75bp margin+0.13
XVA chargeTransferred to XVA desk—−0.05
Unexplained+0.01
Total+0.82
Weak commentary

Rates +$0.82m. Gains as rates rallied on weak data. Swaption book and client flow also positive.

Strong commentary

Rates +$0.82m, driven by the desk’s 2s10s steepener (+$0.48m) in a bull-steepening market, long swaption volatility (+$0.16m net), and one new client swap (+$0.08m net of XVA).

Curve +$0.48m: the PnL came from curve shape, not direction. Net DV01 was only +$5k/bp, so a parallel 5bp rally would have earned about $0.03m. The desk was received $85k/bp in 2-years and paid $110k/bp in 10-years. 2-year yields fell 9bp (+$0.77m) while 10-year yields fell only 4bp (−$0.44m); 5-year and 30-year buckets added $0.15m. The 2s10s curve steepened 5bp.

Basis +$0.04m: swap spread position gained $0.06m as spreads widened 2bp; cross-currency basis lost $0.02m.

Swaptions +$0.16m: long vega gained $0.18m as implied volatility rose 3bp, gamma added $0.07m on the front-end move, and theta cost $0.09m.

Client franchise +$0.08m net: a corporate client paid fixed on a $200m 10-year swap (DV01 $170k) at a 0.75bp margin (+$0.13m), less a $0.05m CVA/FVA charge transferred to the XVA desk. The trade was hedged in the 10-year bucket by close.

Carry and roll-down +$0.05m. Unexplained +$0.01m. The steepener remains in place and is within curve limits.

What the controller should check

  • Never explain a rates desk with net DV01 alone. Here, direction explained about $0.03m of a $0.48m curve gain. The bucketed view is the explanation.
  • Name the curve trade. “2s10s steepener” is how the desk, Risk, and management all think about the position. Use the same language.
  • Describe the curve move correctly. Bull or bear (direction), steepening or flattening (shape). The four combinations have very different PnL consequences.
  • Keep basis separate from delta. Swap spread, tenor, and cross-currency basis are distinct risk factors with distinct limits.
  • Check end-of-day curves. Rates PnL is highly sensitive to curve construction and snap times. A curve methodology or timing change should be disclosed as its own line.
  • Show XVA explicitly on new client trades, so the franchise margin is not overstated.

5.7 Macro / FICC: explaining by theme

A macro desk trades rates, FX, and often commodities and options across developed and emerging markets. Chapter 6 described its revenue as market making, flow intermediation, financing, structuring, and trading and risk warehousing. The difficulty for the controller is that the desk does not think in asset classes. It thinks in themes: “the central bank will cut”, “the dollar will weaken”, “inflation will stay sticky”. A single view is expressed through several instruments, and a single data release can move all of them.

The best macro commentary therefore has two layers: the theme that explains why positions moved together, and the risk-factor detail that proves it.

What drives the PnL

  • G10 and emerging market rates: bucketed DV01 by currency.
  • FX spot and options: positions by pair, plus vega, gamma, and theta.
  • Inflation, commodities, and cross-asset positions.
  • Client franchise across all products, which is often strongest on volatile days.
  • Carry, especially in emerging markets, and reserves that widen when liquidity falls.

The day

US inflation data came in above expectations. Markets priced out near-term rate cuts: 2-year Treasury yields rose 14bp and 10-year yields rose 9bp. The dollar strengthened broadly; USD/MXN rose 1.6%, USD/JPY rose 1.1%. Mexican swap rates rose 12bp, breakeven inflation rose 5bp, gold fell 1.2%, and FX implied volatility rose. The desk was positioned for a “soft landing”: received in front-end US and Mexican rates, and long emerging market currencies against the dollar. Its one-day 99% VaR was $2.1m.

Theme / driverPosition / riskMarket movePnL ($m)
Theme 1: lower US rates
US 2-yearReceived, DV01 +$120k/bp+14bp−1.68
Mexico ratesReceived, DV01 +$35k/bp+12bp−0.42
Theme 2: weaker dollar
USD/MXNLong MXN, $90mUSD/MXN +1.6%−1.44
GoldLong $20m−1.2%−0.24
Offsets and hedges
US 10-yearPaid, DV01 −$60k/bp+9bp+0.54
USD/JPYLong $70m+1.1%+0.77
US inflation breakevensLong, $40k/bp+5bp+0.20
FX optionsLong vega $110k/vol pt; long gammaVol +0.9 pts+0.24
Franchise, carry, other
Client franchiseElevated volumes, all products—+0.65
CarryEM rates and FX—+0.12
Bid-offer reserveEM liquidity—−0.15
FX reval and unexplained−0.11
Total−1.52

FX options +$0.24m comprises vega +$0.10m, gamma +$0.22m, and theta −$0.08m. The final line comprises FX revaluation of non-dollar PnL −$0.04m and unexplained −$0.07m.

Weak commentary

Macro −$1.52m. Difficult day following the US CPI surprise. Losses in rates and EM FX partly offset by gains in USD/JPY, options, and strong client flow.

Strong commentary

Macro −$1.52m following the upside surprise in US inflation. The desk’s “soft landing” positioning (received front-end rates, long EM currencies) lost $3.78m. Hedges and diversifying positions recovered $1.75m, and client franchise added $0.65m.

Theme 1, lower US rates, −$2.10m: received US 2-year (DV01 $120k/bp) lost $1.68m as yields rose 14bp. Received Mexican rates (DV01 $35k/bp) lost $0.42m on a 12bp sell-off in sympathy.

Theme 2, weaker dollar, −$1.68m: long MXN ($90m) lost $1.44m as USD/MXN rose 1.6%. Long gold ($20m) lost $0.24m on a 1.2% fall. Both are expressions of the same dollar view and moved together.

Offsets +$1.75m: paid US 10-year (DV01 $60k/bp) gained $0.54m, leaving the desk a net loser on a bear-flattening US curve. Long USD/JPY ($70m) gained $0.77m. Long breakeven inflation gained $0.20m (+5bp). FX options gained $0.24m (gamma +$0.22m, vega +$0.10m, theta −$0.08m).

Client franchise +$0.65m on elevated post-data volumes across rates and FX. Carry +$0.12m.

Reserves and other −$0.26m: EM bid-offer reserve increased $0.15m as liquidity thinned; FX revaluation −$0.04m; unexplained −$0.07m, within tolerance.

Risk context: the actual loss is 72% of one-day VaR of $2.1m, so there is no exception on actual PnL. Excluding franchise income, the loss was $2.17m, marginally above VaR; flagged to Market Risk for the hypothetical PnL backtest. The desk reduced the US 2-year position by roughly one third by close.

What the controller should check

  • Group by theme, prove by risk factor. Rates and FX losses that look unrelated in an asset-class view are one correlated bet. Saying so is the insight management needs.
  • Correlation is the real risk. The desk’s diversification worked only partly: $3.78m of losses against $1.75m of offsets. That ratio is worth reporting.
  • Give risk context on large days. Compare the loss to VaR and to limits, and distinguish actual PnL from hypothetical PnL, which excludes fees and intraday trading. A day can pass one test and fail the other.
  • Franchise can mask position losses. Strong client income on a volatile day should be shown separately so the position loss is visible.
  • Emerging market marks and reserves. On stressed days, EM closing levels and bid-offer are less reliable. Confirm snap times and sources.
  • Record what changed. If the desk cut risk, say so. Tomorrow’s reader needs to know today’s positions are not yesterday’s.

Quick reference: the language of each desk

The same formula applies everywhere. Only the vocabulary changes. The table also covers desks from Chapter 6 that are not worked through above.

DeskMain risk measureMarket move to quoteTypical extra items
Cash EquitiesPosition valuePrice change (%)Commissions, block discounts, dividends
Prime BrokerageBalancesSpread (bp), days accruedSpecials, repricing, balance changes
FXPosition per pairPips or %Forward points, amendments
Credit TradingCS01, jump-to-defaultSpread change (bp), price pointsBond-CDS basis, issuer events, reserves
Equity DerivativesDelta, gamma, vega, thetaSpot %, implied vol pointsSkew, dividends, correlation, Day-1
RatesBucketed DV01, vegaYield change by tenor (bp)Curve, swap spreads, basis, XVA
Macro / FICCAll of the above, by themeCross-assetCorrelation, EM liquidity, carry
CommoditiesDelta by contract monthPrice change, curve shapeContango / backwardation, physical, seasonality
Securitized ProductsSpread duration, prepaymentSpread (bp), pricePrepayment speeds, model reserves, liquidity
◆
SECTION 06

Common mistakes and how to fix them

MistakeWhy it is a problemThe fix
Restating the attributionThe reader learns the numbers twice and the reason never.Add the position and the market move behind each number.
Passing on the trader’s explanationRemoves independence, the whole point of Product Control.Verify positions and moves from independent data before writing.
Explaining only lossesUnexplained profits are where mismarking and unauthorised risk hide.Apply the same rigour and thresholds to gains.
Netting offsetting driversA small net hides large gross risk.Show material gross drivers even when the total is small.
Ignoring carry and thetaRun-rate items explain most PnL on quiet days.Report them daily so that changes stand out.
Burying one-offsDay-1, reserve releases, and remarks inflate apparent trading performance.Give each its own line and reason.
Tolerating a growing residualSignals missing risk factors or valuation issues.Track unexplained against tolerance; escalate breaches.
Yesterday’s comment, copiedStale commentary is noticed quickly and destroys credibility.Write from today’s attribution every day.
Jargon and desk slangReaders outside the desk cannot follow.Plain language; define abbreviations; state sign conventions.
Open items that vanishReaders lose trust in the process.Carry open items forward until explicitly closed.
◆
SECTION 07

The daily workflow

Good commentary is the output of a disciplined process, not a writing exercise at the end of it. A typical daily cycle looks like this.

WhenStepWhat happens
T+0 closeFlash PnLTraders submit their estimate and their own view of the day’s drivers.
T+0 overnightBatch and valuationPositions, end-of-day market data, and risk sensitivities are produced by official systems.
T+1 earlyOfficial PnL and attributionProduct Control produces PnL, runs attribution, and checks completeness of trades and market data.
T+1 earlyFlash versus actualDifferences between the trader’s estimate and official PnL above threshold are investigated and explained.
T+1 morningDraft commentaryController writes commentary from the attribution, validated against independent market moves and the risk report.
T+1 morningDesk review and sign-offCommentary is discussed with the desk; disagreements are resolved or recorded; trader and controller sign off.
T+1 by deadlineDistributionPnL and commentary go to management, Market Risk, and Finance. Breaches of thresholds are escalated.
OngoingFollow-upOpen items are tracked to closure; themes feed into weekly and month-end reviews.

Two features of this process deserve emphasis.

Flash versus actual is a control. A persistent gap between what the trader thinks the desk made and what the official systems show means one of them has the wrong positions, prices, or models. The commentary should mention material differences and their cause.

Thresholds drive depth. No one can write a full page on every desk every day. Banks set materiality thresholds for when commentary is required, when it must be detailed, and when it must be escalated, for example by absolute PnL, by percentage of VaR, and by size of unexplained.

Where automation and AI help

Much of the workflow above can and should be automated. Attribution engines can supply position, market move, and PnL for each driver. Market data can be pulled automatically. Increasingly, AI tools can produce a first draft of the commentary directly from the attribution, in a consistent template.

This changes the controller’s job; it does not remove it. Automation is good at assembling “position × move = PnL” sentences. It is not accountable for whether the positions are complete, whether the marks are right, whether the explanation is consistent with the desk’s mandate, or whether something looks wrong. The controller’s value moves from typing the explanation to challenging it. Any automated draft should be reviewed, edited, and signed off by a named individual who understands the desk.

◆
SECTION 08

The controller’s checklist

Before sending any commentary, run through these questions.

  1. Does the first sentence give the total, the direction, and the main drivers?
  2. Does every material line state position, market move, and PnL?
  3. Do the named drivers explain at least 80% to 90% of the absolute PnL?
  4. Is client franchise shown separately from risk and inventory PnL?
  5. Have market moves been checked against independent data?
  6. Do the positions quoted agree with the risk report?
  7. Are Day-1 PnL, reserves, amendments, late bookings, and errors each disclosed?
  8. Is the unexplained amount stated and within tolerance?
  9. Is the result consistent with the desk’s mandate, limits, and VaR? If not, has it been escalated?
  10. Are open items listed with an owner, and have yesterday’s been closed?
  11. Would a reader outside the desk understand it without the spreadsheet?
  12. Has the flash versus actual difference been explained?
◆
TAKE AWAY

What to carry into tomorrow’s PnL sign-off

The daily PnL commentary is short, but it carries a great deal of weight. For management, it shows whether each desk is earning money the way the business was designed to. For Risk, it confirms that PnL is consistent with reported exposures and feeds backtesting. For regulators and auditors, it is evidence that an independent function understands and challenges the bank’s trading results every single day.

The method is the same on every desk: lead with the headline, explain position × market move, separate franchise from risk, disclose everything unusual, and state what you cannot explain. What changes as you climb from Cash Equities to Rates and Macro is the vocabulary and the number of moving parts: from price, to carry, to spreads, to Greeks, to curves, and finally to themes that cut across asset classes.

That is why Chapter 6 matters. A controller who understands how a desk makes money can read an attribution and see the story in it. A controller who does not can only repeat the numbers.

A PnL number tells you what happened. A good commentary proves that someone independent understands why.