FINC6023 Financial Risk Management
The Complete Exam Bible · S1 2026FINC6023 Overview
- University of Sydney
- Semester 1 2026
- Postgraduate unit
- Closed book final
FINC6023 Financial Risk Management is a postgraduate University of Sydney unit about identifying, measuring, and managing market, credit, liquidity, and operational risk. Its central question is not merely how to calculate a risk number, but how to choose a model whose assumptions match the exposure.
- Assessment has three parts The mid-semester exam contributes twenty percent, the group report contributes forty percent, and the final exam contributes forty percent.
- Model choice comes first A correct calculation begins by deciding whether the exposure calls for parametric VaR, historical simulation, liquidity adjustment, portfolio aggregation, or a credit-loss model.
- Credit dominates the paper The available final paper devotes half of Part A and two of the three Part B questions to default probability, exposure, expected loss, and Credit VaR.
- Interpret every number A complete response states the horizon, confidence level, units, assumptions, and the decision implied by the computed risk measure.
How FINC6023 is assessed
| Component | Weight | Format |
|---|---|---|
| Mid-semester exam | 20% | Closed-book multiple-choice written test, one hour, in Week 7 |
| Report | 40% | Written report completed in groups of four to five students, with the unit outline saying four and the Canvas administration notes saying four to five, due in Week 12 |
| Final exam | 40% | Closed-book written exam with multiple-choice and short-answer questions, two hours |
The final is a closed-book, two-hour written exam. The available paper adds ten minutes reading time and has Part A with twenty multiple-choice questions for twenty marks, followed by Part B with three short-answer or calculation questions for thirty marks. A formula sheet and standard-normal table are provided. The available paper places half of Part A and two Part B questions in credit risk, with the remaining questions spread across VaR, liquidity, portfolio risk, stress testing, risk classification, and cash-flow mapping.
What FINC6023 covers
The whole subject → one exam-ready map. Each topic links to its free chapter guide.
Foundations of Financial Risk Management
What risk is, who manages it, and the market/credit/liquidity/operational risk familiesFoundations of Financial Risk Management establishes the language used throughout the unit. Risk concerns unexpected changes in the value of assets, liabilities, and business cash flows.
- Definition of risk = volatility of unexpected outcomes
- The risk families: market, credit, liquidity, operational (+ strategic/legal/reputational)
- Classifying a described exposure into the right family
- Why corporate hedging can add value: M&M imperfections (distress costs, taxes, info asymmetry)
- What risk management can and cannot do (good RM ≠ profit)
- Lo & Mueller (2010) five levels of uncertainty
Value at Risk: Parametric & Empirical
The VaR definition, delta-normal (parametric) VaR and historical/empirical VaRValue at Risk: Parametric and Empirical introduces the unit's central market-risk measure. A VaR statement is incomplete unless it gives a loss amount, a confidence level, and a holding period.
- The VaR definition and the (amount, confidence, horizon) triple
- Interpreting VaR: 'lose at least X on 1 day in 100', not at most
- Parametric (delta-normal) VaR = W · |z| · σ · √t
- Key z-quantiles: z₉₅ = 1.645, z₉₇.₅ = 1.960, z₉₉ = 2.326
- Threshold-return route R* = μ + z·σ, then VaR = |R*| · W
- Empirical/historical VaR: rank P&L and read the percentile
VaR Accuracy, Backtesting, Scaling & Expected Shortfall
Confidence/horizon scaling, the √T rule, backtesting, coherence and Expected ShortfallVaR Accuracy, Backtesting, Scaling and Expected Shortfall asks whether a reported VaR is useful and how it changes when the requested confidence level or horizon changes. Confidence conversion uses the ratio of normal quantiles.
- Scaling confidence: VaR_c2 = VaR_c1 · (z_c2 / z_c1)
- Scaling horizon: the √T rule and when it is valid (i.i.d. returns)
- Trend → √T underestimates; mean reversion → √T overestimates
- Precision of estimates: σ estimated more precisely than μ
- Binomial backtesting: exceptions, E[x] = pT, z = (x − pT)/√(p(1−p)T)
- Type I vs Type II error; Kupiec, Christoffersen, and Basel model controls
Liquidity Risk & Liquidity-Adjusted VaR
Trading vs funding liquidity, the bid-offer spread and liquidity-adjusted VaRLiquidity Risk and Liquidity-Adjusted VaR distinguishes trading liquidity from funding liquidity. Trading liquidity concerns how quickly a position can be sold and how much its price moves during liquidation.
- Trading liquidity vs funding liquidity
- Proportional bid-offer spread s = (offer − bid)/mid
- Normal-market LVaR = VaR + ½ Σ sᵢ wᵢ [OFF-SHEET — memorise]
- Stressed-market LVaR = VaR + ½ Σ (μᵢ + λσᵢ) wᵢ [OFF-SHEET]
- Optimal unwinding: market impact vs price risk trade-off
- Liquidity black holes, positive/negative feedback trading, the leverage-deleverage cycle
Portfolio VaR: Variance, Diversification & Component VaR
Portfolio variance w′Σw, diversified vs undiversified VaR, and marginal/component VaRPortfolio VaR: Variance, Diversification and Component VaR moves from one position to a portfolio whose risks interact. Portfolio volatility depends on each position's weight and volatility plus every covariance.
- Two-asset variance σ_p² = 2ρ(w₁σ₁)(w₂σ₂) + (w₁σ₁)² + (w₂σ₂)²
- Matrix form σ_p² = w′Σw
- Portfolio VaR = W · z · σ_p · √T
- Diversified vs undiversified VaR and the diversification benefit
- Cross-check VaR_p = √(VaR_A² + VaR_B² + 2ρ·VaR_A·VaR_B)
- Marginal and component VaR; components sum to total and a negative component is a hedge
Multivariate Models, Correlations & Copulas
Factor models, the multivariate normal, the Gaussian copula and Cholesky factorisationMultivariate Models, Correlations and Copulas explains how dependence enters portfolio and credit models.
- Single-factor model Rᵢ = αᵢ + βᵢR_m + εᵢ → σᵢ² = βᵢ²σ_m² + σ_εi²
- Covariance from a factor model σᵢⱼ = βᵢβⱼσ_m²
- Correlation ρ = cov(V₁,V₂)/(SD(V₁)SD(V₂))
- Independence ≠ zero correlation (the triangular-distribution example)
- Bivariate-normal conditionals μ_{x|y} and σ_{x|y}
- Gaussian copula: percentile-to-percentile mapping onto standard normals; Sklar's theorem
Forecasting Volatility & Correlations
Implied volatility, EWMA, ARCH and GARCH(1,1) with long-run variance and mean reversionForecasting Volatility and Correlations supplies the changing volatility input used by market-risk models.
- Volatility as the SD of log returns; annualise by √252
- Implied volatility and the VIX (risk-neutral)
- Equally-weighted estimator σ_n² = (1/m)Σ u_{n−i}²
- EWMA σ_n² = λσ_{n−1}² + (1−λ)u_{n−1}² (λ = 0.94)
- ARCH(m) and GARCH(1,1) σ_n² = ω + αu_{n−1}² + βσ_{n−1}²
- Long-run variance V_L = ω/(1−α−β); α+β < 1 gives mean reversion
Historical Simulation, Mapping & the Linear/Quadratic Model
Historical simulation, extreme-value theory, cash-flow mapping and the delta-gamma modelHistorical Simulation, Mapping and the Linear or Quadratic Model covers portfolios that do not fit a simple delta-normal approximation.
- Historical simulation: replay factor moves, rank P&L, and apply weighting or volatility updating
- Extreme-value theory: GPD tail and the power law
- Interest-rate mappings: principal, duration, cash-flow, PCA
- Duration VaR: ΔP = −D·P·Δy, VaR = D·P·z·σ_y
- Cash-flow mapping by variance-matching (CORE Part-B)
- Linear and quadratic option models, Monte Carlo simulation, and bootstrap confidence intervals
Operational Risk, Stress Testing & Scenario Analysis
Operational-risk loss events, and stress testing versus scenario analysisOperational Risk, Stress Testing and Scenario Analysis covers loss mechanisms that a market VaR number does not describe adequately. Operational risk arises from failed or inadequate people, processes, systems, or external events.
- Operational risk definition and major loss events (Barings, SocGen, UBS, NAB)
- Why VaR is insufficient: describes 'normal' conditions and only a minimum tail loss
- Stress testing: non-statistical, no probability attached
- Scenario analysis: historical / prospective / portfolio-driven
- Moving one variable at a time ignores correlations (= sensitivity testing)
- Typical shocks: ±100bps yield curve, ±20% equity/vol, ±20bps spreads
Credit Risk: Estimating Default Probabilities
Rating transition matrices, hazard rates, recovery, the Merton model and PD from spreadsCredit Risk: Estimating Default Probabilities turns ratings, market prices, and firm value into measures of default risk. The building blocks are probability of default, exposure at default, and loss given default.
- The three drivers: PD, recovery/LGD, exposure (EAD)
- Ratings, transition matrices and two-period default via matrix powers
- Cumulative vs marginal PD; hazard (conditional) = [cum(t) − cum(t−1)]/[1 − cum(t−1)]
- N-year cumulative from marginals: PD = 1 − Π(1 − dᵢ) [product is OFF-SHEET]
- Continuous survival and spread-implied intensity: V(t) = e^(−λ̄t), Q(t) = 1 − e^(−λ̄t), λ̄ ≈ s/(1 − R)
- PD from bond prices and the separate jobs of physical and pricing probabilities
Credit Risk: Credit Exposure & Credit VaR
Credit exposure, CVA, expected credit loss and Credit VaR = WCL − ECLCredit Risk: Credit Exposure and Credit VaR combines the chance of default with the amount that would be lost. Exposure depends on contract value at the time of default.
- Exposure by contract type (always-asset, always-liability, can-be-either)
- Greatest exposure when the contract is in-the-money to you
- Wrong-way vs right-way risk
- Mitigation: netting, collateral, downgrade triggers; CVA
- Expected credit loss E[CL] = Σ PD·EAD·LGD (independent of correlation)
- Credit VaR = WCL − ECL; default correlation raises unexpected loss, not expected loss
ABSs, CDOs & the 2007-08 Financial Crisis
Securitisation, the ABS waterfall and tranches, CDOs and the crisis chainABSs, CDOs and the Financial Crisis is the narrative capstone of the unit. Securitisation transfers a pool of loans to a special-purpose vehicle and issues claims on the pool's cash flows. Tranching creates a waterfall.
- Securitisation and the ABS structure: Senior/Mezzanine/Equity tranches
- The waterfall: senior paid first, equity absorbs first losses
- Originate-and-distribute and the moral-hazard problem
- ABS CDO, CDO² and synthetic CDO (built from CDSs)
- The crisis chain: subprime → bubble → 2006 slump → default → flight to quality
- Key modelling mistakes: AAA risk understated, correlation rises in stress
The unit begins with risk families and corporate hedging, develops parametric and historical Value at Risk, then tests the reliability of those measures through horizon scaling, backtesting, and Expected Shortfall.
The middle of the unit extends the same logic to real portfolios. Liquidity-adjusted VaR adds trading-cost risk to the market-loss estimate.
Portfolio VaR uses covariance and correlation to capture diversification, while factor models and copulas describe dependence across many positions. EWMA and GARCH supply changing volatility estimates. Historical simulation, Monte Carlo methods, and cash-flow mapping provide alternatives when a simple linear-normal model is unsuitable.
The final block moves from market risk to credit risk.
Students distinguish real-world and risk-neutral default probabilities, recover cumulative and conditional default rates, value exposure, compute expected credit loss, and build a discrete loss distribution for Credit VaR. Securitisation and the financial crisis then show why correlation, incentives, and model limits matter outside a calculation.
The official assessment has three components.
A closed-book multiple-choice mid-semester exam contributes twenty percent. A written report completed in groups of four to five students contributes forty percent, with the unit outline saying four and the Canvas administration notes saying four to five. A closed-book final exam contributes forty percent and lasts two hours.
The available paper adds ten minutes reading time and specifies twenty multiple-choice marks followed by three ten-mark short-answer or calculation questions. A formula sheet and standard-normal table are supplied, so preparation should focus on method selection, off-sheet relationships, clear working, and interpretation rather than isolated formula recall.
Worked example · free
Parametric VaR, then convert confidence and horizon (closed-book, formula-sheet style)
- 2Write the delta-normal formula: VaR = W · |z| · σ · √t, with W the position value, σ the per-period volatility and t the horizon. Identify W = 2,000,000, σ = 0.012, and for part (a) z = 1.645 (95%), t = 1 day.
- 2Compute the 1-day 95% VaR: 2,000,000 × 1.645 × 0.012 = $39,480.
- 1Scale the confidence level by the z-ratio (independent of horizon): multiply by z₉₉/z₉₅ = 2.326/1.645 = 1.414.
- 1Scale the horizon by the square-root-of-time rule (valid only for i.i.d. returns with no mean reversion or trend): multiply by √10 = 3.162.
- 2Combine: 39,480 × 1.414 × 3.162 ≈ $176,500.
Key terms
- Value at Risk (VaR)
- The loss on a position that will not be exceeded with confidence X over horizon t — quoted as a (currency amount, confidence, horizon) triple. A '99%, 1-day, $1m VaR' means you expect to lose at least $1m on about 1 day in 100, not at most.
- Expected Shortfall (ES / CVaR)
- The average loss given that the loss has exceeded VaR — the mean of the tail beyond the VaR quantile. ES ≥ VaR and, unlike VaR, it is coherent (sub-additive), so it cannot be 'gamed' by splitting a portfolio.
- Liquidity-adjusted VaR (LVaR)
- Ordinary VaR plus a half-spread liquidation cost: LVaR = VaR + ½ Σ sᵢ wᵢ (normal markets). It is examinable but NOT on the provided formula sheet, so it must be memorised; the stressed version replaces s with μ + λσ of the spread.
- Default probability (PD), LGD and EAD
- The three credit-risk drivers: PD is the probability the counterparty defaults, LGD = 1 − recovery is the loss given default, and EAD (credit exposure) is the amount at risk at default. Expected credit loss = PD × EAD × LGD.
- Credit VaR
- The worst credit loss at a confidence level minus the expected credit loss, Credit VaR = WCL − ECL. Default correlation raises the unexpected (tail) loss and hence Credit VaR, but leaves the expected loss unchanged.
- Backtesting
- Backtesting compares realised losses with forecast VaR and counts exceptions to judge whether a risk model is calibrated and whether exceptions cluster through time.
- Component VaR
- Component VaR is the contribution of one position to total portfolio VaR. The component amounts sum to portfolio VaR and can reveal a position that acts as a hedge.
- Cash-flow mapping
- Cash-flow mapping replaces an off-vertex fixed-income cash flow with positions at standard maturity vertices while preserving present value and matching return variance.
- Default correlation
- Default correlation measures the tendency of obligors to default together. It changes the tail of the credit-loss distribution even when each obligor's marginal default probability is unchanged.
- Expected credit loss
- Expected credit loss is the probability-weighted mean loss, commonly computed from probability of default, exposure at default, and loss given default.
- GARCH
- GARCH is a conditional-variance model in which tomorrow's variance depends on a constant, the latest squared shock, and the previous conditional variance.
- Hazard rate
- A hazard rate is the probability or intensity of default during a period conditional on the borrower having survived to the start of that period.
- Historical simulation
- Historical simulation applies observed risk-factor changes to the current portfolio, ranks the resulting profit and loss observations, and reads risk measures from the empirical tail.
- Stress testing
- Stress testing evaluates portfolio outcomes under extreme but plausible scenarios that are not described adequately by a single VaR quantile.
FINC6023 FAQ
What makes this unit difficult?
The challenge is breadth plus model selection. Similar-looking stems may require parametric VaR, historical simulation, a liquidity adjustment, portfolio covariance, or a credit-loss distribution. Students improve fastest when they label the risk family and assumptions before entering numbers, then finish every calculation with units and an interpretation.
What is the official assessment structure?
The official record lists a closed-book mid-semester examination worth twenty percent, a written group report worth forty percent, and a closed-book final examination worth forty percent. The three published weights sum to one hundred percent. Current operational details should still be checked in the unit outline and Canvas instructions.
How is the final paper organised?
The available paper allows two hours plus ten minutes reading time. Part A contains twenty multiple-choice questions worth one mark each. Part B contains three short-answer or calculation questions worth ten marks each. Multiple-choice answers go on the computer answer sheet, while Part B responses go in the answer book.
Which topics dominate the available final paper?
Credit risk supplies half of Part A and two of the three Part B questions. The remaining multiple-choice questions emphasise VaR definition and scaling, liquidity adjustment, portfolio VaR, stress testing, and risk-family classification. The third Part B question uses cash-flow mapping followed by portfolio VaR.
What does the formula sheet change about revision?
It reduces the value of memorising formulas that are already printed and increases the value of recognising the correct model. You should still carry relationships absent from the sheet, especially liquidity-adjusted VaR, cumulative default probability, Merton distance to default, bond-implied risk-neutral default probability, and the Credit VaR definition.
How should I practise calculation questions?
Use the same sequence every time: name the risk measure, write the symbolic relation, align the time and percentage units, substitute, retain precision, round once, and interpret the result. For Credit VaR, build and sort the complete loss distribution before selecting the confidence quantile and subtracting expected loss.
How should I use the ten-minute reading period?
First classify the three Part B questions and mark the formulas or tables they require. Next scan Part A for quick conceptual items and longer calculations. Note any unit conversions, confidence levels, and recovery assumptions. Begin writing only after you know which questions need the most arithmetic and checking time.
When will the unit run again?
The official handbook record shows one availability in Semester 1 for 2026. Because the unit runs once each year in Semester 1, the next sitting is expected in 2027. Confirm the future availability and dates on the current University of Sydney unit page before planning enrolment or assessment timing.
Questions students ask in each FINC6023 chapter
- What is the single best test for classifying a risk in an MCQ?
- Ask what generates the loss. A change in a market price or rate is market risk; a counterparty failing to pay is credit risk; an inability to trade or fund a position is liquidity risk; a failure of an internal process, person or system is operational risk. The same underlying event can carry more than one family. Chapter 01 · Foundations of Financial Risk Management
- Why is historical VaR often bigger than parametric VaR at 99%?
- Because real return distributions usually have fatter tails than a normal curve. The delta-normal method assumes normality and so understates how often extreme losses occur, whereas the historical method reads the actual observed tail — at high confidence the gap widens. Chapter 02 · Value at Risk: Parametric & Empirical
- When is the √T rule wrong?
- It assumes returns are independent and identically distributed with stable variance. If returns trend, positive autocorrelation can make √T understate long-horizon risk because losses reinforce one another. If returns mean-revert, negative autocorrelation can make it overstate risk because moves partly cancel. State the direction of the bias and use an autocorrelation-aware variance when supplied. Chapter 03 · VaR Accuracy, Backtesting, Scaling & Expected Shortfall
- Is the LVaR formula on the exam formula sheet?
- No — the unit flags that LVaR is one of the basic formulas that is NOT on the provided sheet, yet it IS examinable. You must memorise both the normal-market form (½ × spread × position) and the stressed-market form (replace the fixed spread with its mean plus a confidence multiple of its standard deviation). Chapter 04 · Liquidity Risk & Liquidity-Adjusted VaR
- What does a negative component VaR mean?
- It means that position is reducing the book's overall risk — it acts as a hedge. Because component VaRs must sum to the total portfolio VaR, a hedging position contributes a negative slice, so adding more of it lowers total VaR. Chapter 05 · Portfolio VaR: Variance, Diversification & Component VaR
- Do I have to compute copulas in the exam?
- No — the Week-13 Review states copulas are understand-and-interpret only. You should be able to explain the percentile-to-standard-normal mapping, why it separates marginals from dependence, and the role of the Gaussian copula in the crisis, but the calculations focus on factor models, correlation and bivariate normals. Chapter 06 · Multivariate Models, Correlations & Copulas
- Is the GARCH long-run-variance formula on the sheet?
- The basic GARCH relation V_L = ω/(1−α−β) appears, but the unit flags that you must be able to REARRANGE it (recover V_L from ω, or ω = γV_L) quickly under pressure — practise both directions, since the intercept ω = γV_L trips many students. Chapter 07 · Forecasting Volatility & Correlations
- When do I use the linear model versus delta–gamma or Monte Carlo?
- Use the linear (delta) model when the portfolio's value is roughly linear in the risk factors. For options and other convex payoffs the linear model misses curvature and skew, so switch to the quadratic delta–gamma approximation or, when payoffs are highly non-linear, full Monte Carlo revaluation. Chapter 08 · Historical Simulation, Mapping & the Linear/Quadratic Model
- Why isn't VaR enough on its own?
- VaR describes only 'normal' market conditions and reports the MINIMUM loss in the tail (the quantile), not how deep the tail goes. It cannot capture rare, joint, extreme events. Stress testing and scenario analysis fill that gap by deliberately examining severe situations VaR was never designed to see. Chapter 09 · Operational Risk, Stress Testing & Scenario Analysis
- What is the difference between the marginal, hazard and cumulative PD?
- The marginal (unconditional) PD is the chance of defaulting IN year t viewed from today, cum(t) − cum(t−1). The hazard (conditional) PD assumes survival to the start of year t and divides by the survivors. The cumulative PD is the chance of defaulting by year t, built as 1 − the product of yearly survival probabilities. Chapter 10 · Credit Risk: Estimating Default Probabilities
- Why does default correlation affect Credit VaR but not expected credit loss?
- Expected loss is a sum of marginal expectations, Σ PD·EAD·LGD, and expectation is linear — it does not care whether defaults happen together. Default correlation changes the SHAPE of the loss distribution: positive correlation makes simultaneous defaults more likely, fattening the tail and raising the worst-case loss (and hence Credit VaR) while leaving the mean unchanged. Chapter 11 · Credit Risk: Credit Exposure & Credit VaR
- Why was the AAA rating on senior tranches misleading?
- The senior tranche only suffers losses once the subordinated tranches below it are exhausted, which requires many loans to default together. Ratings models assumed defaults were largely independent (low correlation); in the crisis correlation surged, so simultaneous defaults became common and the supposedly safe AAA tranches took losses. Chapter 12 · ABSs, CDOs & the 2007-08 Financial Crisis
How to study for the exam
Start from the assessment weights. The mid-semester examination rewards quick recognition of early VaR, liquidity, and portfolio concepts, so practise those as timed multiple-choice sets. The group report requires a defensible modelling chain: define the portfolio, justify the risk factors, estimate VaR under more than one technique, test assumptions, and explain why results differ.
Keep a short record of data choices and limitations so the final discussion follows from the analysis rather than being added at the end. For the final examination, divide preparation by the paper's observed mix. Spend substantial time on default probability, exposure, expected loss, Credit VaR, and bond-implied risk-neutral probability.
Rehearse cash-flow mapping as a complete written solution because it combines interpolation, present value, variance matching, aggregation, and horizon scaling. Use shorter drills for VaR interpretation, confidence and horizon changes, liquidity-adjusted VaR, portfolio diversification, stress testing, and risk classification.
In every quantitative response, write the relation before substitution, keep decimals and percentages consistent, preserve intermediate precision, and finish with a sentence that names the horizon, confidence level, and practical meaning. In the final week, alternate one timed mixed set with one error-review session so the same sign, recovery, correlation, or quantile mistake is not repeated.
Your AI Finance tutor for FINC6023
Stuck on a hard FINC6023 question? Sia is AskSia’s AI Finance tutor — ask any FINC6023 Financial Risk Management question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.
