Chapter 5 · Market Risk

12 of 100 exam questions (~12%) · types of market risk · hedging and limits · mean/median/mode · variance and standard deviation · normal distribution · alpha and beta · VaR · stress vs scenario
← Back to quiz
Why this chapter matters. 12 of 100 exam questions (~12%) on the v12 weighting (the Edition 11 workbook itself states 15; the bank uses v12 weights). It is the statistics chapter and feeds VaR, which reappears in later chapters. Trap zones: mean vs median vs mode (and which resists outliers), variance is the square of the standard deviation, population vs sample formula (n vs n-1), the 68 / 95.5 / 99.75 rule, beta vs alpha, VaR does not say how bad losses can get, the three VaR methods and what each assumes, and stress test = one variable; scenario = several.

5.1 Defining and identifying market risk

▼

What market risk is syllabus 5.1.1

Financial institutions make profits from price uncertainty; the risk of loss it brings is market risk. The workbook gives two definitions: "the risk of loss arising from changes in the value of financial instruments", and the BIS's more general version: the risk of losses in on- and off-balance sheet positions arising from movements in market prices. The aim is not to eradicate market risk but to understand and quantify it so an informed decision can be made about whether it is acceptable. Understanding it is also important for pricing products such as futures and options.

For an investment in a company's shares, direct factors reflect the company's own performance (balance sheet health, management vision and strength); indirect factors affect it indirectly (interest rate levels, economic events, political, sector sentiment and environmental effects).

The seven types of market risk syllabus 5.1.1

TypeDefinition and key points
Volatility riskPrice movements more uncertain than usual, affecting pricing. All priced instruments suffer it, but it especially affects options pricing: in a volatile market options are harder to price and become more expensive
Market liquidity riskLoss through being unable to trade in a market or get a price on a desired product when required; caused by lack of supply or demand or a shortage of market makers
Currency riskAdverse exchange rate movements; affects any portfolio or instrument with cash flows in a currency other than the firm's or investor's base currency. Also inherent in trading cryptoassets, which often display far higher volatility than fiat currencies
Basis riskOne exposure is hedged with an offsetting exposure in another instrument that behaves in a similar, but not identical, way. If the two were truly equal and opposite there would be no risk; basis risk exists to the extent they do not exactly mirror each other
Interest rate riskAdverse interest rate movements; directly affects fixed-income securities, futures, options and forwards, and may indirectly affect other instruments
Commodity price riskAdverse commodity price movement. Differs from other drivers because supply is often concentrated in a few suppliers, which can magnify volatility; liquidity fluctuations exacerbate it; storage ease and cost vary (gold, electricity, wheat). Commodity prices generally show higher volatility and larger price discontinuities (leaps) than most traded securities
Equity price riskReturns come from capital growth and income (dividends). Risk to capital: price falls or fails to keep pace with inflation or safer investments. Risk to income: dividends may not keep pace with inflation, fall or not be paid; unlike bond coupons, dividends are not compulsory
Memory hook for the list: V-M-C-B-I-C-E = Volatility, Market liquidity, Currency, Basis, Interest rate, Commodity, Equity.
Quick check
A hedge uses an instrument that moves similarly to, but not exactly like, the exposure hedged. The residual risk is:

Boundary issues and worked examples syllabus 5.1.2 - 5.1.3

Boundary issues: the types overlap, so it is not straightforward to say which factors cause which movements. Examples: liquidity risk from lack of supply or demand also causes price level risk; rising volatility risk worsens price level risk for buyers and sellers; interest rate risk indirectly affects the real economy and therefore markets.

The workbook names three key drivers, currency, interest rate and liquidity, and illustrates each:

DriverExample
CurrencyA Chinese investor buys a US equity portfolio. The investment returns +10% in dollars, but the dollar loses 8% against the pound, so the investor's return is only 2% (the workbook's simple subtraction)
Interest rateA long-term bond bought in Hong Kong. Bond prices move inversely to interest rates, so a rate rise cuts prices. Short-term bonds (under three years) move relatively little; long-term bonds (over ten years) more, because the distant future is unknown and riskier
LiquidityAn investor buys commercial property to rent out; because of liquidity problems in the property market it takes much longer to sell, and even a price cut may not help if no one wants it or mortgage credit is hard to get
Worked example (the workbook's currency case)
A foreign investor's US equity portfolio returns 10% in dollars, but the dollar falls 8% against the investor's currency. What is the approximate return in the investor's currency?
  1. 1 Local (dollar) return = +10%.
  2. 2 Currency movement = -8% (dollar weakens).
  3. 3 Return in the investor's currency ≈ 10% - 8% = 2%. A good investment result is largely lost to currency risk.
Quick check
If interest rates rise, the price of a long-term bond will typically:

5.2 Managing market risk: hedging, limits, diversification, electronic trading

▼

Hedging syllabus 5.2.1

Hedging reduces the risk of adverse price movements by taking an offsetting position in a related product; it is a way of insuring against market risk. The main tools are derivatives, especially futures and options.

Example: an investor who owns an equity can buy a put option costing a fraction of the equity's price. It gives the right, not the obligation, to sell the stock at a set price (the strike price) within a given time, protecting against a market fall. The decision to hedge is a trade-off between the risk of adverse movement and the cost of the hedge (here the option price). Perfect offset is hard, and hedging can introduce or worsen other risks: basis, credit and operational risk.

Market risk limits (stop-loss limits) syllabus 5.2.1

The market equivalent of credit limits. The maximum loss a firm will accept on a portfolio or transaction is the market risk limit or stop-loss limit, expressed in VaR or as an absolute number of the instrument traded. Their effectiveness depends on the accuracy of the risk measurement. Over-simplified measurement causes two problems:

  • limits usually have to be inflated to accommodate measurement error, hurting potential profit
  • traders may exploit the inaccuracy and take risks they know the measure misses

With high-quality data limits can be very effective. Professionals may see them as restrictive, but they can also be empowering: they set the firm's risk appetite and give explicit authority to take specified risk. For electronic trading, position limits are an extremely effective mitigant because machines can trade up to, but not beyond, the limit with great speed and accuracy.

Diversification syllabus 5.2.1

A security held alone has risk given by the standard deviation of its returns. Combining two in varying proportions gives a portfolio whose standard deviation is, in almost all cases, lower than the weighted average of the two individual standard deviations. Workbook illustration: shares in a sun cream factory and an umbrella business have earnings that move inversely (negative correlation), so combined earnings are less volatile. Diversification means combining securities whose returns ideally move in opposite directions or, if the same direction, at least not to the same degree. Multi-stock portfolios use correlation to quantify it.

Trap: diversification lowers portfolio standard deviation below the weighted average of the parts; it does not raise expected return by itself.

Electronic and high-frequency trading (HFT) syllabus 5.2.1

TypeApproach
Model-based HFTMathematical models predict prices over minutes or hours, run on very fast computers co-located in the market's data centre. Speed lets liquid positions be unwound at short notice, so they can offer tight spreads and only need a small profit per trade. This permits small position limits; small limits plus fast liquidation greatly reduce market risk, so such firms can act as market makers committing to a minimum volume of trades daily. They hold positions for minutes or even hours
Pure-speed HFTPursue speed above all else, investing heavily in infrastructure such as microwave links to beat fibre networks. Speed alone is not a sustainable advantage because rivals can replicate it. They hardly hold positions for any time

Flash crashes are sometimes blamed on HFT generally in the media, because the distinction between the two types is not always understood.

Quick check
Which control is described as an extremely effective mitigant of market risk in electronic trading?

5.3 The market risk function + central tendency

▼

The market risk management function syllabus 5.2.2

Like credit risk, market risk needs an independent risk management function working company-wide. An effective one has:

  • ownership of the firm's market risk management policy
  • proactive management involvement in market risk issues
  • defined escalation procedures for rising trading losses and limit breaches
  • independent validation of market pricing and of the adequacy of VaR models
  • VaR not used alone but combined with stress testing and scenario analysis
  • independent daily monitoring of risk utilisation through daily P&L and review of front-office closing prices ("independent" = a separately accountable function reporting directly to senior management)

Its role varies: a bank trading on its own account has limits applied "across the board", whereas each fund in an investment manager could have a different risk appetite.

Why statistics: central tendency and dispersion syllabus 5.2.3

Unlike operational risk, market risk analysis can draw on large amounts of data, which is a "mixed blessing": the bigger the dataset the harder it is to separate expected behaviour from outliers, so samples are often used. Three questions about a sample: its typical value (central tendency), how far other values stray (dispersion), and how well the sample mirrors the population (built into the standard deviation formula). A normal (Gaussian) distribution is the familiar bell shape seen in data such as IQ or adult heights: many values near the middle, fewer at each extreme. Its predictive properties would be useful if market returns also fitted it.

Mean, median and mode syllabus 5.2.3

MeasureHowWorkbook example
MeanSum of all values ÷ number of observations (the everyday "average")Returns of 7%, 8%, 9%, 10%, 11%, 12%: total 57 ÷ 6 = 9.5%
MedianMiddle item of the sorted data; for an even count, the average of the two middle items2, 5, 8, 9, 12, 13, 15, 16, 18: median = 12. With 19 added (ten values): (12 + 13) ÷ 2 = 12.5
ModeMost frequently occurring value; there may be no mode, or two (bimodal), three (trimodal) or more (multimodal)Classic car values $100k, $125k, $100k, $115k, $135k, $95k, $100k: mode = $100k

Which to use? The mean is most common but is influenced by outliers. The median is not, so it is a more robust measure for skewed or non-normal data. Workbook illustration: with weekly income (2006), a few millionaires push the mean (£443) above the median (£362), so around 60% of people are below the mean; consider both. The mode is useful for qualitative data (e.g. Morse's letter counts, where five letters each appeared about 8,000 times).

Trap: outliers drag the mean; the median resists them. Positive skew means mean above median.
Quick check
The ordered data 2, 5, 8, 9, 12, 13, 15, 16, 18, 19 has a median of:

5.4 Dispersion: range, IQR, quartile deviation, variance, SD

▼

Measures of dispersion at a glance syllabus 5.2.3

MeasureCentred onIdea
Range-Highest value minus lowest. Simplest; distorted by extreme values and ignores the numbers in between
Inter-quartile rangeMedianDifference between the 25th and 75th percentile positions
Quartile deviationMedianHalf the difference between upper and lower quartiles: (Q3 - Q1) / 2; not influenced by extremely high or low scores
VarianceMeanAverage of the squared differences from the mean
Standard deviationMeanSquare root of the variance; the most commonly used measure

Quartiles split ordered data into quarters, deciles into tenths, percentiles into hundredths. A fund in the third quartile ranks in the top half; one in the first quartile has returns exceeded by 75% of the sample. In the workbook's 12-fund example (returns 3.5% to 9%), the median is 5.875%. The smaller the inter-quartile range, the less difference between the funds.

Worked example: range (from the workbook)
A fund's annual returns over ten years are 7, 6, 5, 2, 3, 3, 6, 9, 11 and 13 (%). What is the range?
  1. 1 Highest = 13, lowest = 2.
  2. 2 Range = 13 - 2 = 11. It ignores the eight values in between.

Quartile deviation: the four steps syllabus 5.2.3

  1. Order the data and find the median, Q2.
  2. Split the data at Q2 into a lower and upper half (if the number of items is odd, include Q2 in both halves).
  3. Q1 = median of the lower half; Q3 = median of the upper half.
  4. Quartile deviation = (Q3 - Q1) ÷ 2.
Worked example (from the workbook)
Returns: 49, 54, 59, 60, 62, 65, 65, 68, 77, 83, 84, 89, 90 (n = 13). Find the quartile deviation.
  1. 1 Q2 = median = the 7th value = 65.
  2. 2 Lower set (including Q2): 49, 54, 59, 60, 62, 65, 65 → Q1 = 4th value = 60. Upper set: 65, 68, 77, 83, 84, 89, 90 → Q3 = 4th value = 83.
  3. 3 Quartile deviation = (83 - 60) ÷ 2 = 11.5.
Quick check
If Q1 = 60 and Q3 = 83, the quartile deviation is:

Variance and standard deviation syllabus 5.2.3

Steps to calculate by hand:

  1. Find the mean.
  2. Find each value's deviation from the mean.
  3. Square each deviation.
  4. Divide the sum of squared deviations by the number of observations to get the (population) variance.
  5. Take the square root to get the standard deviation.
BasisStandard deviation formula
Population√[ Σ(x - mean)² ÷ n ]
Sample√[ Σ(x - mean)² ÷ (n - 1) ]

For a sample, the divisor is reduced by one to allow for sampling error, since a small dataset might not represent the whole population. Because the standard deviation is the square root of the variance, it is in the same units as the returns, which makes it easier to interpret (variance is in different units). It represents the average amount by which values deviate from the mean.

Worked example (from the workbook)
Annual returns: 13, 11, 2, 6, 5, 8, 7, 9, 7, 6 (%). Find the mean, variance and standard deviation (population basis).
  1. 1 Mean = 74 ÷ 10 = 7.4%.
  2. 2 Deviations: 5.6, 3.6, -5.4, -1.4, -2.4, 0.6, -0.4, 1.6, -0.4, -1.4. Squared: 31.36, 12.96, 29.16, 1.96, 5.76, 0.36, 0.16, 2.56, 0.16, 1.96. Sum = 86.4.
  3. 3 Variance = 86.4 ÷ 10 = 8.64.
  4. 4 Standard deviation = √8.64 ≈ 2.94% (the workbook gives the variance and describes the square-root step; the rounded value is my arithmetic).
Memory hook: SD = √variance. Sample uses n - 1; population uses n.
Quick check
A data set has a variance of 9. Its standard deviation is:

5.5 Standard deviation in use, normal and fat-tailed distributions

▼

Why standard deviation matters for risk syllabus 5.2.4

  • The variance of a stock's returns is used to calculate its beta, but it is in different units from the returns, so the SD is used for conceptualising dispersion.
  • The variability of an asset's or portfolio's returns is its volatility, a measure of its risk, expressed mathematically by the SD.
  • Higher SD = more volatile = higher risk; lower SD = lower risk.
  • Experience shows that for about two-thirds of the time the return is within one SD above or below the average.
Trap: an increase in volatility of returns is reflected in a HIGHER standard deviation.

The normal distribution and confidence intervals syllabus 5.2.5

Distribution analysis uses historical data to predict future events and relies on probability distributions. The normal distribution is symmetrical about its mean and is defined by two attributes: its mean and its standard deviation (sigma, σ). The weekly income distribution, by contrast, is positively skewed.

Within...Share of data (workbook figures)
±1 standard deviation68.26% (about two-thirds)
±2 standard deviations95.5%
±3 standard deviations99.75%

Confidence intervals: returns from primary instruments (not derivatives), driven by market factors, are often assumed normally distributed. That lets a model predict future performance to a given probability, linked to how many standard deviations from the mean you read off: the probability is the confidence interval. So instead of quoting only a mean, you can say "there is a 95.5% chance the return will be within two standard deviations of the mean".

Worked example
An instrument's returns are assumed normal with a mean of 8% and a standard deviation of 3%. Using the workbook's percentages, within what range would you expect returns about 95.5% of the time?
  1. 1 95.5% corresponds to ±2 standard deviations.
  2. 2 Two standard deviations = 2 × 3% = 6%.
  3. 3 Range = 8% ± 6% = 2% to 14%. (For about 68.26% of the time the range would be 8% ± 3% = 5% to 11%.) The 8% and 3% are illustrative inputs.
Quick check
For a normal distribution, about 95.5% of data lies within:

Fat-tailed distributions syllabus 5.2.5

In a normal distribution the tails touch the axis after about three to five standard deviations. Real returns often do not behave that way: many distributions have fat tails (more extreme outcomes than the normal implies; the weekly income graph is the workbook's illustration). Using the normal as a simplifying assumption when the data are fat-tailed means periodically very large deviations from expected returns.

Memory hook: normal assumption + fat-tailed reality = extreme losses turn up more often than the model says.

5.6 Probability, volatility, regression, correlation, alpha, beta, optimisation

▼

Probability and volatility syllabus 5.2.6

Probability = number of ways the event can occur ÷ total number of possible outcomes

Dice example: rolling a six = 1/6; rolling an even number = 3/6 = 1/2.

Volatility describes how variable prices are. Technically it is a measure of the standard deviation of an instrument's returns within a specific time horizon, often annualised, expressed as an absolute number ($5) or a fraction of the mean return (5%). The workbook notes that investment grade is less volatile than speculative grade, and bond prices are typically less volatile than equities.

Regression and the correlation coefficient syllabus 5.2.6

Regression analysis investigates relationships between variables and estimates the effect of one on another (e.g. price on demand), along with the statistical significance of the estimate. A scatter plot can show a clear relationship, but not which variable causes the change.

Correlation coefficientMeaningDiversification
+1Perfect positive: same direction, in perfect synchronisationThe only case where diversification benefits cannot be achieved
0Not correlatedDiversification benefits still available
-1Perfect negative: opposite direction at all times in the same proportionThe only case that can produce a risk-free return when combined

The value ranges from -1 through 0 to +1. Even imperfect or zero correlation gives diversification benefits.

Quick check
Diversification benefits CANNOT be achieved when two securities' returns have a correlation of:

Beta vs alpha syllabus 5.2.6

MeasureWhat it describesReading it
BetaThe relationship of a stock's or portfolio's returns with the market as a wholeBeta 0: not correlated with the market. Positive beta: generally follows the market. Negative beta: moves inversely to the market, generally falling when the market rises
AlphaOutperformance of a fund or portfolio against its benchmarkThe difference between the fund's expected returns (based on its beta) and its actual returns
Memory hook: Beta = follows the market; Alpha = beats the benchmark (after allowing for beta).
Quick check
Alpha is best described as the difference between a fund's:

Optimisation syllabus 5.2.6

A well-constructed portfolio can show much lower risk than the average of its assets' risks. Optimisation means portfolio construction techniques that obtain the best expected return from the right mix of correlations and variances. It is often called mean-variance (MV) optimisation: mean = expected return, variance = the measure of risk. It can be framed as either minimising risk for a given return or maximising expected return for a given risk.

5.7 Value-at-Risk (VaR) and its three approaches

▼

What VaR says, and what it does not syllabus 5.2.7

VaR expresses the maximum loss that can occur with a specified confidence over a specified period. Because there is uncertainty about the loss, VaR includes a confidence level that the loss will not be exceeded. Workbook example: a portfolio's one-week VaR of $1 million in 95 weeks out of 100 means it is predicted to lose less than $1 million at the 95% confidence level.

AdvantagesDisadvantage
Gives a statistical probability of potential loss; easily understood by non-risk managers; translates all portfolio risks into a common standard, allowing comparison across asset classes and firm-wide, cross-product exposuresDoes not specify how bad it could get: losing no more than $10m on 95 days out of 100 says nothing about the other five days, except that the loss will be at least $10m

Limits: banks, securities firms and investment managers use VaR to estimate portfolio market risk. The market risk function sets VaR limits and monitors them; breaches are escalated to the head of trading, the desk head or a risk committee.

Trap: VaR at 95% tells you the loss you should not exceed 95 times out of 100, not the worst case. That is why it must be combined with stress testing and scenarios.
Quick check
The main disadvantage of VaR as a measure of portfolio risk is that it:

Validation and back testing syllabus 5.2.7

VaR models can break down if their assumptions are violated or untrue; that danger is model risk. Models should be validated. Back testing compares the actual daily trading exposure with the previously predicted VaR figure. It tests the reliability of the VaR methodology, is usually performed daily by the financial reporting function, and if the gap between reality and estimate is unsatisfactory the VaR model must be revised.

The three VaR approaches compared syllabus 5.2.8

ApproachHow it worksAdvantagesDisadvantages
Historical simulationTake actual historic risk-factor changes (e.g. last 500 trading days), apply each to the current risk factor values to revalue the portfolio that many times, rank the results and pick the 5th percentile in the left tail for 95% VaRConceptually simple; no distribution assumption; no need to estimate volatilities and correlations (captured implicitly in the actual history)Assumes history will repeat itself; may look benign just before a crash
Parametric (analytical)Assumes returns are normally distributed and uses the standard deviation (volatility) to plot the curve and read the VaR at the required confidence levelSimplest method; little input data; no simulations so minimal computation timeInaccurate if returns are not normal; cannot be used for options, whose returns are not normally distributed
Monte Carlo simulationBuild a model (equations) for future returns and run many hypothetical trials using random numbers generated in a "controlled" way from the chosen distribution; typically 10,000 or more iterations; rank results and read the percentile (e.g. the worst 500 of 10,000 = worst 5%)Can give VaR for non-normally distributed instruments such as options (simulate the underlying, re-price the option each time)Needs a lot of computing power and time: 1,000 assets x 10,000 simulations = ten million simulations, before any sub-simulations to price options

The first two steps of historical simulation: identify the risk factors that affect returns (stock prices, volatilities, correlations), then select a sample of actual historic changes. Monte Carlo follows a similar algorithm, except that instead of picking a historical return it generates a random number to estimate the asset's return.

Worked example: reading the percentile
A firm runs a Monte Carlo VaR with 10,000 iterations and wants the 95% confidence level. Which outcome gives the VaR?
  1. 1 Rank the 10,000 hypothetical outcomes from worst to best.
  2. 2 5% of 10,000 = 500, so the worst 500 outcomes form the left-hand tail.
  3. 3 The VaR is read at the boundary of those 500 worst outcomes (the 5th percentile). The workbook's diagram uses exactly this: "the 500 worst outcomes from 10,000 iterations, i.e. worst 5%".
Memory hook: History needs no assumptions but assumes repetition; Parametric assumes normality (no options); Monte Carlo handles options but is heavy on computing.
Quick check
Which VaR approach cannot be used for options because it assumes normally distributed returns?

5.8 Scenario analysis and stress testing

▼

Stress test vs scenario syllabus 5.2.9

VaR does not say how bad things could get, so scenario analysis and stress testing are needed. The difference is the number of variables altered:

TechniqueWhat it does
Stress testAlters one variable at a time to see its effect on the rest of the portfolio; alters each relevant risk factor in exceptional but plausible ways
Scenario analysisLooks at how the portfolio behaves under multiple simultaneous changes

Risk factors to stress: general interest rates, individual stock prices, individual stock volatilities and the correlation coefficients between instruments. Scenarios may be invented (e.g. a doubling of volatility with a sharp rise in interest rates) or based on past external factors (e.g. large one-day market falls). Portfolios should face "extreme" market event scenarios to expose risks VaR may not capture.

Principles for good scenario and stress testing syllabus 5.2.9

  • Part of the market risk management process, run regularly, with scenarios chosen from historical data or empirical models of risk-factor changes.
  • Aim to assess the effect of sizeable changes in risk factors on holdings and financial condition, so include low-probability adverse stresses that could cause extraordinary losses.
  • Both quantitative and qualitative.
  • As far as possible institution-wide, covering market and non-market factors: prices, volatilities, market liquidity, historical correlations and assumptions under stress, and vulnerability to worst-case events such as the default of a large counterparty.
  • Helps the board and senior management judge the impact on earnings and capital; they should regularly review results and the major assumptions, and consider them when setting and reviewing policies and limits.
Quick check
The key difference between a stress test and a scenario analysis is the:

5.9 Ch 5 cheat sheet

▼

All the numbers + names

ItemAnswer
Market riskLoss from changes in the value of financial instruments (BIS: on- and off-balance sheet positions, market price movements)
7 typesVolatility · market liquidity · currency · basis · interest rate · commodity · equity
Basis riskHedge instrument similar but not identical to the exposure
Bond price and ratesMove inversely; longer maturity = bigger move
Currency example+10% local, dollar -8% → about 2%
HedgingOffsetting position (futures, options, e.g. put option); trade-off vs cost; can add basis, credit, operational risk
Stop-loss limitMaximum loss accepted; over-simple measures inflate limits and invite exploitation
Market risk functionIndependent; escalation; validate pricing and VaR; VaR + stress + scenario; daily P&L monitoring
Mean / median / modeAverage / middle value / most frequent; median resists outliers
ExamplesMean 9.5% · median 12.5 · mode $100k
RangeHighest minus lowest (example 13 - 2 = 11)
Quartile deviation(Q3 - Q1) ÷ 2 (example 11.5)
Variance / SDMean squared deviation (example 8.64) / its square root; sample uses n - 1
Normal curveSymmetrical; defined by mean and SD; 68.26% / 95.5% / 99.75% within 1 / 2 / 3 SD
Fat tailsMore extreme outcomes than a normal assumes
Correlation-1 to +1; only +1 removes diversification benefit; only -1 can give risk-free return
Beta / alphaSensitivity to the market / outperformance against benchmark
OptimisationMean-variance: minimise risk for a given return, or maximise return for a given risk
VaRMaximum loss at a confidence level over a period; does not show how bad it can get
Back testingCompare actual daily exposure with predicted VaR; usually daily by financial reporting
Historical simulationNo distribution assumption; assumes history repeats (e.g. 500 days)
ParametricAssumes normal; simplest; not for options
Monte Carlo10,000+ random trials; handles options; heavy computing (1,000 assets x 10,000 = ten million)
Stress vs scenarioOne variable at a time vs several at once