Value at Risk (VaR): Comprehensive Study

Guide

n
na
Prepared by Prof. [Link], Bangalore

Fi
July 27, 2025

of
Table of Contents
1 History and Evolution of VaR 3

ld
1.1 Origins and Early Development . . . . . . . . . . . . . . . . . . . . 3
1.2 The Birth of VaR at J.P. Morgan (1980s-1990s) . . . . . . . . . . 3

or
1.3 Popularization and Standardization (1990s) . . . . . . . . . . . . . 4
1.4 Regulatory Adoption and Basel Accords . . . . . . . . . . . . . . . 5
-W
1.5 Crisis and Criticism (2000s-2010s) . . . . . . . . . . . . . . . . . . . 5
1.6 Post-Crisis Evolution (2010s-Present) . . . . . . . . . . . . . . . . . 6
1.7 Key Figures in VaR Development . . . . . . . . . . . . . . . . . . . 7
1.8 Timeline of VaR Development . . . . . . . . . . . . . . . . . . . . . 7
1.9 Current State and Future Directions . . . . . . . . . . . . . . . . . 7
t h

2 VaR Basics 9
Pa

2.1 Q1: What is Value at Risk (VaR)? . . . . . . . . . . . . . . . . . . 9

2.2 Q2: What are the mathematical foundations of VaR? . . . . . . . 9
2.3 Q3: How do you convert VaR from one time horizon to another? 10
2.4 Q4: How do you convert VaR from one confidence level to another? 11
in

3 VaR Calculation Methods 13

3.1 Q5: What are the three main VaR calculation methods? . . . . . 13
a

3.2 Q6: How do you calculate VaR using the Historical Simulation
nt

Method? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.3 Q7: How do you calculate VaR using the Variance-Covariance
ou

Method? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4 Q8: How do you calculate VaR using Monte Carlo Simulation? 15
eM

4 VaR Uses and Applications 16

4.1 Q9: What are the primary uses of VaR in finance? . . . . . . . . 16

5 VaR Limitations 17
Th

5.1 Q10: What are the major limitations of VaR? . . . . . . . . . . . 17

6 VaR for a Single Asset 18

6.1 Q11: How do you calculate VaR for a single asset using different
methods? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1
 Value at Risk (VaR) - Comprehensive Study Guide 2

7 VaR for Two Assets 19

7.1 Q12: How do you calculate VaR for a two-asset portfolio? . . . . 19

8 VaR for Multiple Assets 20

8.1 Q13: How do you calculate VaR for a multi-asset portfolio? . . 20

n
9 Expected Shortfall (Conditional VaR) 22

na
9.1 Q14: What is Expected Shortfall and how is it calculated? . . . 22

10 VaR Illustrations and Problems for Learners 23

Fi
10.1 Simple Problem: Single Asset VaR . . . . . . . . . . . . . . . . . . 23
10.2 Intermediate Problem: Two-Asset Portfolio VaR . . . . . . . . . . 23
10.3 Complex Problem: Multi-Asset VaR with Horizon and Confi-
dence Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

of
10.4 CFA-Style Practice Problems . . . . . . . . . . . . . . . . . . . . . . 25
10.4.1 Problem 1: Conceptual Understanding . . . . . . . . . . . . 25

ld
10.4.2 Problem 2: Historical VaR Calculation . . . . . . . . . . . . 25
10.4.3 Problem 3: Variance-Covariance (Parametric) VaR . . . . 25
10.4.4 Problem 4: Portfolio VaR (Two Assets) . . . . . . . . . . . 25

or
10.4.5 Problem 5: Monte Carlo Simulation (Conceptual) . . . . . 26
10.4.6 Problem 6: CFA Topic Integration/Interpretation . . . . . 26
-W
10.4.7 Problem 7: Quick CFA Quantitative Practice . . . . . . . 26
t h
Pa
a in
nt
ou
eM
Th
 Value at Risk (VaR) - Comprehensive Study Guide 3

1 History and Evolution of VaR

1.1 Origins and Early Development
The Pre-VaR Era (1950s-1980s):
Before VaR became the standard risk measure, financial institutions relied on various

n
ad-hoc methods to measure risk:

na
• Notional Exposure: Simple sum of position sizes without considering correlations

• Sensitivity Analysis: Greeks in options trading (Delta, Gamma, Vega)

Fi
• Scenario Analysis: What-if analysis for specific market moves

• Stop-Loss Limits: Maximum allowable losses per position or trader

of
These methods had significant limitations:

ld
• No unified risk metric across different asset classes

• Difficulty in aggregating risks

• No probabilistic interpretation

or
-W
• Limited consideration of portfolio effects

1.2 The Birth of VaR at J.P. Morgan (1980s-1990s)

Key Catalysts:
t h

1. Market Volatility: The 1970s and 1980s saw increased market volatility due to:
Pa

• End of Bretton Woods system (1971)

• Oil crises (1973, 1979)
• High inflation periods
in

• Deregulation of financial markets

a

2. Trading Volume Growth: Explosive growth in derivatives and proprietary trad-

nt

ing

3. Regulatory Pressure: Need for better risk management after various trading
ou

scandals

4. Computing Power: Advancement in computational capabilities

eM

J.P. Morgan’s Innovation:

In the late 1980s, J.P. Morgan developed what would become VaR under the leadership
of:
Th

• Dennis Weatherstone: CEO who wanted a single-page risk report every morning

• Risk Management Team: Led by Till Guldimann and his colleagues

 Value at Risk (VaR) - Comprehensive Study Guide 4

The ”4:15 Report”:

Dennis Weatherstone famously requested a single-page report by 4:15 PM each
day that would summarize the firm’s trading risks. This led to the development of:

• A unified risk measure across all trading desks

n
• 95% confidence level, 1-day horizon

na
• Aggregation of risks across different asset classes

• Daily monitoring and reporting

Fi
This became the template for modern VaR systems.

of
1.3 Popularization and Standardization (1990s)

ld
RiskMetrics (1994):
J.P. Morgan made a revolutionary decision to make their VaR methodology public:

or
• October 1994: Launch of RiskMetrics technical document
-W
• Free Distribution: Made available to financial institutions worldwide

• Daily Data: Provided volatility and correlation estimates

• Software: Free software for VaR calculations

h

Impact of RiskMetrics:
t

• Standardized VaR methodology across the industry

Pa

• Enabled smaller institutions to implement sophisticated risk management

• Created a common language for risk communication

in

• Facilitated regulatory adoption

a

Academic Development:
nt

The 1990s saw significant academic contributions:

• Philippe Jorion: ”Value at Risk” (1997) - First comprehensive textbook

ou

• Kevin Dowd: ”Beyond Value at Risk” (1998) - Extensions and alternatives

eM

• Paul Embrechts: Extreme value theory applications

• Alexander McNeil: Quantitative risk management frameworks

Th
 Value at Risk (VaR) - Comprehensive Study Guide 5

1.4 Regulatory Adoption and Basel Accords

Basel I (1988):
• Focused primarily on credit risk

• Simple risk-weighted assets approach

n
• No explicit market risk capital requirements

na
Basel II (2004):

Fi
• Market Risk Amendment (1996): First regulatory recognition of VaR

• Banks could use internal VaR models for capital calculations

of
• 99% confidence level, 10-day holding period

• Multiplication factor based on backtesting performance

ld
Key Requirements under Basel II:

or
• Minimum 1-year historical data
-W
• Daily VaR calculation

• Independent risk management function

• Regular backtesting and validation

h

1.5 Crisis and Criticism (2000s-2010s)

t

Major Financial Crises:

Pa

Several crises highlighted VaR limitations:

1. Long-Term Capital Management (1998):

in

• Sophisticated VaR models failed to predict losses

• Extreme leverage (30:1) amplified small probability events
a

• Correlation breakdown during crisis

nt

• Near-collapse required Federal Reserve intervention

ou

2. Dot-com Crash (2000-2002):

• VaR models underestimated technology sector risks

eM

• Correlation increases during market stress

• Historical data didn’t capture bubble dynamics

3. Global Financial Crisis (2007-2009):

Th

• VaR failed to capture systemic risk

• Mortgage-backed securities showed unexpected correlations
• ”100-year floods” occurred repeatedly
 Value at Risk (VaR) - Comprehensive Study Guide 6

• Banks with sophisticated VaR systems still failed

Criticisms that Emerged:

• Procyclicality: VaR increases during crises, forcing asset sales

n
• Tail Risk: Ignores extreme losses beyond confidence threshold

na
• Model Risk: Overreliance on historical data and assumptions

• Systemic Risk: Doesn’t capture interconnected failures

Fi
1.6 Post-Crisis Evolution (2010s-Present)

of
Basel III (2010):
Significant enhancements to address VaR limitations:

• Stressed VaR: Additional capital requirement based on 12-month stressed period

ld
• Incremental Risk Charge: For specific risks not captured in VaR

or
• Comprehensive Risk Measure: For correlation trading portfolios
-W
• Fundamental Review of Trading Book (2019): Move toward Expected Short-
fall

Expected Shortfall Adoption:

Basel III.1 (implemented 2022-2023) introduced major changes:
h

• Replacement of VaR with Expected Shortfall for regulatory capital

t
Pa

• 97.5% confidence level (equivalent to 99% VaR for normal distributions)

• Better capture of tail risk

• Coherent risk measure properties

in

Technological Advancements:
a

• Machine Learning: Advanced modeling techniques

nt

• Big Data: Incorporation of alternative data sources

ou

• Real-time Monitoring: Intraday VaR calculations

• Cloud Computing: Scalable risk infrastructure

eM
Th
 Value at Risk (VaR) - Comprehensive Study Guide 7

1.7 Key Figures in VaR Development

Pioneers:
• Dennis Weatherstone: J.P. Morgan CEO who commissioned the first VaR sys-
tem

n
• Till Guldimann: Led the technical development at J.P. Morgan

na
• Jacques Longerstaey: Co-developer of RiskMetrics methodology

• Peter Zangari: Contributed to RiskMetrics technical framework

Fi
Academic Contributors:
• Philippe Jorion: UC Irvine professor, authored seminal VaR textbook

of
• Paul Embrechts: ETH Zurich, extreme value theory applications

ld
• Alexander McNeil: ETH Zurich, quantitative risk management

• Kevin Dowd: University of Nottingham, VaR extensions and alternatives

or
-W
1.8 Timeline of VaR Development
Year Milestone
1980s J.P. Morgan develops internal VaR system for trading
risk
h

1994 RiskMetrics methodology published and made freely

available
t

1996 Basel Market Risk Amendment allows internal VaR

Pa

models
1997 Philippe Jorion publishes first comprehensive VaR text-
book
in

1998 LTCM crisis highlights VaR limitations

2004 Basel II formally incorporates VaR for regulatory capital
2007-09 Financial crisis exposes systemic VaR failures
a

2010 Basel III introduces stressed VaR requirements

nt

2016 Basel III.1 replaces VaR with Expected Shortfall

2022-23 Implementation of Expected Shortfall in regulatory
ou

frameworks

1.9 Current State and Future Directions

eM

Modern VaR Implementation:

• Integration with stress testing and scenario analysis
Th

• Real-time risk monitoring and alerts

• Machine learning enhanced parameter estimation

• Multi-asset class and multi-currency capabilities

 Value at Risk (VaR) - Comprehensive Study Guide 8

• Regulatory and economic capital integration

Emerging Trends:

• Climate Risk: Incorporating environmental risk factors

• Cyber Risk: Quantifying operational risk from cyber threats

n
na
• Model Risk: Better measurement and management of model uncertainty

• Systemic Risk: Network effects and contagion modeling

Fi
• Alternative Data: Social media sentiment, satellite imagery, etc.

Lessons Learned:

of
• VaR is a valuable but incomplete risk measure

• Must be supplemented with stress testing and scenario analysis

ld
• Regular model validation and backtesting are essential

or
• Risk managers must understand and communicate model limitations
-W
• Regulatory requirements continue to evolve

Evolution of Risk Reporting:

1980s - J.P. Morgan Original:
h

• Single page daily report

t
Pa

• 95% confidence, 1-day horizon

• Basic position and VaR information

in

2020s - Modern Risk Dashboard:

• Real-time interactive dashboards

a
nt

• Multiple confidence levels and horizons

• VaR, Expected Shortfall, stress testing results

ou

• Drill-down capabilities by desk, region, asset class

eM

• Integration with P&L attribution and backtesting

• Regulatory and economic capital views

The history of VaR demonstrates both the power of financial innovation and the
Th

importance of continuous improvement in risk management practices. While VaR revo-

lutionized risk management, its evolution continues as markets and technologies advance.
 Value at Risk (VaR) - Comprehensive Study Guide 9

2 VaR Basics
2.1 Q1: What is Value at Risk (VaR)?

Value at Risk (VaR) is a statistical measure that quantifies the maximum potential

n
loss that an asset, portfolio, or firm might experience over a specified time period at
a given confidence level under normal market conditions.

na
Answer: VaR answers the question: ”What is the worst loss we might expect over a

Fi
given time period with a certain level of confidence?”
Key Components:
• Time Horizon: The period over which risk is measured (1 day, 10 days, 1 month)

of
• Confidence Level: Probability that losses will not exceed VaR (typically 95% or
99%)

ld
• Amount at Risk: Maximum expected loss in monetary terms

or
Example: A 1-day VaR of $1 million at 95% confidence means:
-W
• There is a 95% probability that daily losses will not exceed $1 million

• There is a 5% probability that daily losses will exceed $1 million

• On average, losses greater than $1 million should occur once every 20 trading
h

days
t
Pa

2.2 Q2: What are the mathematical foundations of VaR?

Answer: VaR is fundamentally based on probability distributions and percentiles.
For a random variable X representing returns, VaR at confidence level α is:
in

VaRα = −FX−1 (1 − α)
a
nt

Where:
• FX−1 is the inverse cumulative distribution function (quantile function)
ou

• α is the confidence level (e.g., 0.95 for 95%)

• The negative sign converts returns to losses

eM

Step-by-step illustration:
1. Define the return distribution
Th

2. Determine the (1 − α) quantile

3. Convert to loss terms

4. Scale by portfolio value

 Value at Risk (VaR) - Comprehensive Study Guide 10

2.3 Q3: How do you convert VaR from one time horizon to
another?
Answer: Under the parametric approach and assuming independent and identically
distributed (i.i.d.) returns with no serial correlation, VaR can be scaled across different
time horizons using the square root of time rule. This assumes that volatility scales with

n
the square root of time.

na
The formula to convert VaR from a shorter horizon to a longer horizon is:
√
VaRT = VaR1 × T

Fi
Where:

• VaRT is the VaR over T periods

of
• VaR1 is the VaR over 1 period (e.g., 1 day)

• T is the number of periods (e.g., 5 for 5 days, 252 for annual assuming 252 trading

ld
days)

or
Important Assumptions:
-W
• Returns are i.i.d. (independent and identically distributed)

• No mean drift (or negligible for short horizons)

• Normal distribution of returns

h

Limitations:
t

• Does not hold for non-normal distributions or fat-tailed risks

Pa

• Ignores autocorrelation in returns

• May underestimate extreme long-horizon risks in practice

in

Converting 1-Day VaR to Multi-Day and Annual VaR:

a

Given:
nt

• 1-day VaR at 95% confidence: $10,000

ou

• Portfolio value: $1,000,000

• Assume 252 trading days in a year

eM

To 5-Day VaR:
√
VaR5 = $10, 000 × 5 (1)
= $10, 000 × 2.236 (2)
Th

= $22, 360 (3)

 Value at Risk (VaR) - Comprehensive Study Guide 11

To 10-Day VaR (Basel Standard):

√
VaR10 = $10, 000 × 10 (4)
= $10, 000 × 3.162 (5)
= $31, 620 (6)

n
To Annual VaR (252 Days):

na
√
VaR252 = $10, 000 × 252 (7)
= $10, 000 × 15.875 (8)

Fi
= $158, 750 (9)

Interpretation:

of
• The 5-day VaR suggests a 5% chance of losing more than $22,360 over 5 days

• Annual VaR indicates a 5% chance of losing more than $158,750 over a year

ld
• Note: For long horizons
√ like annual, mean drift should be considered: VaRT =

or
P0 × (−µT + zα σ T )
-W
2.4 Q4: How do you convert VaR from one confidence level to
another?
Answer: Converting VaR between confidence levels requires adjusting the critical value
h

(z-score) from the assumed distribution. For the parametric method assuming normal
distribution, the VaR scales with the ratio of the z-scores.
t

The conversion formula is:

Pa

z α2
VaRα2 = VaRα1 ×
z α1
Where:
in

• VaRα2 is VaR at new confidence level α2

a

• VaRα1 is known VaR at confidence level α1

nt

• zα is the z-score for confidence level α

ou

Common z-scores:

• 90%: 1.28
eM

• 95%: 1.65

• 99%: 2.33
Th

• 99.9%: 3.09

Assumptions:

• Normal distribution of returns

 Value at Risk (VaR) - Comprehensive Study Guide 12

• Zero or negligible mean return

• Constant volatility

Limitations:

• Inaccurate for non-normal distributions

n
na
• Does not account for skewness or kurtosis

• Better to recalculate from raw data when possible

Fi
Converting Between Confidence Levels:
Given:

of
• 95% 1-day VaR: $15,000

ld
• Portfolio value: $1,000,000

• Assumed normal distribution with zero mean

From 95% to 99%:

or
-W
z95% = 1.65 (10)
z99% = 2.33 (11)
2.33
VaR99% = $15, 000 × (12)
1.65
h

= $15, 000 × 1.412 (13)

t

= $21, 180 (14)

Pa

From 95% to 90%:

z90% = 1.28 (15)

in

1.28
VaR90% = $15, 000 × (16)
1.65
a

= $15, 000 × 0.776 (17)

nt

= $11, 640 (18)

From 99% to 99.9%:

ou

z99.9% = 3.09 (19)

eM

3.09
VaR99.9% = $21, 180 × (20)
2.33
= $21, 180 × 1.326 (21)
= $28, 085 (22)
Th

Interpretation:

• Higher confidence levels result in larger VaR values

 Value at Risk (VaR) - Comprehensive Study Guide 13

• The ratio depends only on the z-scores for normal distributions

• For non-zero mean: Adjust as VaRα2 = −µ + σzα2 after backing out σ from
original VaR

n
3 VaR Calculation Methods

na
3.1 Q5: What are the three main VaR calculation methods?
Answer: The three primary methods are:

Fi
1. Historical Simulation Method

of
2. Variance-Covariance (Parametric) Method

3. Monte Carlo Simulation Method

ld
3.2 Q6: How do you calculate VaR using the Historical Simu-

or
lation Method?
Answer: This method uses actual historical return data without making distributional
-W
assumptions.
Step-by-step procedure:

1. Data Collection: Gather historical returns for n periods

h

R1 , R 2 , R 3 , . . . , R n
t
Pa

2. Sort Returns: Arrange returns from worst (most negative) to best

R(1) ≤ R(2) ≤ R(3) ≤ . . . ≤ R(n)

in

3. Find Percentile: For confidence level α, find the (1 − α) × n observation

Position = (1 − α) × n
a
nt

4. Calculate VaR:
VaR = −R((1−α)×n) × Portfolio Value
ou

Detailed Example:
eM

Suppose we have 100 historical daily returns and want 95% VaR:
Step 1: Historical returns (sample):

3.2%, −1.8%, 0.5%, −4.1%, 2.1%, −0.3%, . . . , 1.7% (23)

Th

Step 2: After sorting (worst to best):

− 4.1%, −3.7%, −3.2%, −2.8%, −2.1%, . . . , 3.2% (24)

 Value at Risk (VaR) - Comprehensive Study Guide 14

Step 3: For 95% confidence:

Position = (1 − 0.95) × 100 = 5 (25)

Step 4: The 5th worst return is -2.1%, so:

VaR = −(−2.1%) × $1, 000, 000 = $21, 000

n
(26)

na
Interpretation: There is a 5% chance of losing more than $21,000 in one day.

Fi
3.3 Q7: How do you calculate VaR using the Variance-Covariance
Method?

of
Answer: This parametric method assumes returns are normally distributed.
Mathematical Framework:

VaR = |µ − zα × σ| × Portfolio Value

ld
Where:

or
• µ = Expected return (often assumed to be 0 for short horizons)
-W
• zα = Z-score corresponding to confidence level α
• σ = Standard deviation of returns
Step-by-step calculation:
h

1. Calculate Mean Return:

1X
n
t

µ= Ri
Pa

n i=1

2. Calculate Standard Deviation:

v
u
1 X
in

u n
σ=t (Ri − µ)2
n − 1 i=1
a
nt

3. Determine Z-score:
• 90% confidence: z0.90 = 1.28
ou

• 95% confidence: z0.95 = 1.65

• 99% confidence: z0.99 = 2.33
eM

4. Calculate VaR:
VaR = (µ − zα × σ) × Portfolio Value
Th

Detailed Example:
Portfolio value: $500,000
Historical data: 250 daily returns
 Value at Risk (VaR) - Comprehensive Study Guide 15

Step 1: Calculate statistics

µ = 0.08% (daily) (27)

σ = 1.8% (daily) (28)

Step 2: For 95% confidence, z0.95 = 1.65

n
Step 3: Calculate VaR

na
VaR = |0.08% − 1.65 × 1.8%| × $500, 000 (29)
= |0.08% − 2.97%| × $500, 000 (30)

Fi
= | − 2.89%| × $500, 000 (31)
= $14, 450 (32)

of
Interpretation: There is a 5% probability of losing more than $14,450 in one
day.

ld
3.4 Q8: How do you calculate VaR using Monte Carlo Simula-

or
tion?
-W
Answer: This method generates thousands of potential future scenarios through random
simulation.
Step-by-step procedure:

1. Model Selection: Choose probability distribution for returns

h

2. Parameter Estimation: Estimate distribution parameters from historical data

t

3. Random Generation: Generate N random scenarios (typically N ≥ 10, 000)

Pa

4. Portfolio Valuation: Calculate portfolio value for each scenario

5. Loss Calculation: Convert to losses
in

6. VaR Estimation: Find the (1 − α) percentile of simulated losses

a

Mathematical Implementation:
nt

For normal distribution:

R i = µ + σ × Zi
ou

Where Zi ∼ N (0, 1) are standard normal random variables.

Monte Carlo VaR Calculation:

eM

Given:

• Portfolio value: $1,000,000

• Expected daily return: µ = 0%

Th

• Daily volatility: σ = 2%

• Confidence level: 95%

 Value at Risk (VaR) - Comprehensive Study Guide 16

• Simulations: N = 10, 000

Simulation Process:

1. Generate 10,000 random returns: Ri = 0% + 2% × Zi

2. Calculate portfolio values: Vi = $1, 000, 000 × (1 + Ri )

n
3. Calculate losses: Li = $1, 000, 000 − Vi

na
4. Sort losses in descending order

Fi
5. VaR = 500th largest loss (5th percentile)

Sample Results: If the 500th largest loss is $32,890, then:

of
VaR95% = $32, 890

ld
4 VaR Uses and Applications

or
4.1 Q9: What are the primary uses of VaR in finance?
-W
Answer: VaR serves multiple critical functions in financial risk management:

1. Risk Measurement and Monitoring

• Daily risk assessment

h

• Tracking risk exposure changes

t

• Risk dashboard reporting

Pa

2. Risk Limits and Controls

• Setting position limits

• Defining risk budgets
in

• Automated trading stops

a

3. Capital Allocation
nt

• Determining economic capital requirements

ou

• RAROC (Risk-Adjusted Return on Capital) calculations

• Regulatory capital compliance
eM

4. Performance Evaluation

• Risk-adjusted performance metrics

• Sharpe ratio enhancements
Th

• Benchmark comparisons

5. Regulatory Reporting

• Basel III compliance

 Value at Risk (VaR) - Comprehensive Study Guide 17

• Stress testing requirements

• Disclosure obligations

5 VaR Limitations

n
5.1 Q10: What are the major limitations of VaR?

na
Answer: VaR has several important limitations that users must understand:

Fi
Critical Limitations:

1. Tail Risk Ignorance

of
• VaR provides no information about losses beyond the confidence threshold
• Doesn’t capture ”black swan” events

ld
• Can underestimate extreme market conditions
2. Model Risk

or
• Parametric VaR assumes normal distributions
-W
• Historical VaR assumes past patterns repeat
• Monte Carlo VaR depends on model assumptions
3. False Sense of Security
• High confidence levels (99%) still leave 1% probability of exceedance
h

• Exceedances can be much larger than VaR suggests

t

• May encourage excessive risk-taking

Pa

4. Coherence Issues
• VaR is not a coherent risk measure
in

• Fails subadditivity property

• Portfolio VaR can exceed sum of individual VaRs
a

5. Procyclicality
nt

• VaR increases during market stress

ou

• Can amplify market volatility

• Forces selling when markets are down
eM

Illustration of VaR Limitation:

Consider a portfolio with 95% VaR of $10 million:

• VaR tells us there’s 5% chance of losing more than $10 million

Th

• VaR doesn’t tell us the loss could be $50 million or $100 million

• The expected loss given exceedance is unknown

 Value at Risk (VaR) - Comprehensive Study Guide 18

This is why Expected Shortfall (ES) was developed as a complement to VaR.

6 VaR for a Single Asset

6.1 Q11: How do you calculate VaR for a single asset using

n
different methods?

na
Answer: For a single asset, VaR calculation depends on the chosen method:
Method 1: Parametric (Normal Distribution)

Fi
VaR = P0 × |µ − zα × σ|
Where:

of
• P0 = Current portfolio/position value

ld
• µ = Expected return

• σ = Return volatility

• zα = Critical z-value

or
-W
Method 2: Historical Simulation
Sort historical returns and find the (1 − α) percentile.
Method 3: Monte Carlo
Simulate return paths and find the (1 − α) percentile of simulated losses.
h

Complete Single Asset VaR Example:

t

Given:
Pa

• Stock position: $2,000,000

• Historical mean daily return: µ = 0.05%

in

• Historical volatility: σ = 2.5%

a

• Confidence level: 99%

nt

• Time horizon: 1 day

ou

Parametric Method:

z0.99 = 2.33 (33)

eM

VaR = $2, 000, 000 × |0.05% − 2.33 × 2.5%| (34)

= $2, 000, 000 × |0.05% − 5.825%| (35)
= $2, 000, 000 × 5.775% (36)
Th

= $115, 500 (37)

Historical Method: If we have 1000 historical returns, the 1st percentile (10th
worst return) gives us VaR.
 Value at Risk (VaR) - Comprehensive Study Guide 19

Interpretation: There is a 1% probability of losing more than $115,500 in one

day.

7 VaR for Two Assets

n
7.1 Q12: How do you calculate VaR for a two-asset portfolio?

na
Answer: For a two-asset portfolio, correlation between assets significantly affects VaR
calculation.

Fi
Portfolio Return Variance:

σp2 = w12 σ12 + w22 σ22 + 2w1 w2 σ1 σ2 ρ12

of
Where:

• w1 , w2 = Asset weights (w1 + w2 = 1)

ld
• σ1 , σ2 = Asset volatilities

or
• ρ12 = Correlation coefficient between assets
-W
Portfolio VaR:
VaRp = P0 × |µp − zα × σp |
Where:

• µp = w1 µ1 + w2 µ2 = Portfolio expected return

h

p
• σp = σp2 = Portfolio volatility
t
Pa

Two-Asset Portfolio VaR Calculation:

Given:
in

• Total portfolio value: $5,000,000

a

• Asset 1: Weight = 60%, σ1 = 3%, µ1 = 0.1%

nt

• Asset 2: Weight = 40%, σ2 = 4%, µ2 = 0.15%

ou

• Correlation: ρ12 = 0.3

• Confidence level: 95%

eM

Step 1: Portfolio Expected Return

µp = 0.6 × 0.1% + 0.4 × 0.15% (38)

= 0.06% + 0.06% = 0.12% (39)
Th
 Value at Risk (VaR) - Comprehensive Study Guide 20

Step 2: Portfolio Variance

σp2 = (0.6)2 (3%)2 + (0.4)2 (4%)2 + 2(0.6)(0.4)(3%)(4%)(0.3) (40)

= 0.36 × 0.0009 + 0.16 × 0.0016 + 2 × 0.24 × 0.0012 × 0.3 (41)
= 0.000324 + 0.000256 + 0.0001728 (42)

n
= 0.0007528 (43)

na
Step 3: Portfolio Volatility
√
σp = 0.0007528 = 2.744% (44)

Fi
Step 4: Portfolio VaR (95% confidence, z0.95 = 1.65)

of
VaR = $5, 000, 000 × |0.12% − 1.65 × 2.744%| (45)
= $5, 000, 000 × |0.12% − 4.528%| (46)
= $5, 000, 000 × 4.408% (47)

ld
= $220, 400 (48)

or
Diversification Benefit Analysis:
Undiversified VaR (weighted average of individual VaRs):
-W
VaR1 = $3, 000, 000 × 1.65 × 3% = $148, 500 (49)
VaR2 = $2, 000, 000 × 1.65 × 4% = $132, 000 (50)
VaRundiversif ied = $148, 500 + $132, 000 = $280, 500 (51)
h

Diversification Benefit:
t

Benefit = $280, 500 − $220, 400 = $60, 100 (52)

Pa

$60, 100
Percentage = = 21.4% (53)
$280, 500
in

8 VaR for Multiple Assets

a
nt

8.1 Q13: How do you calculate VaR for a multi-asset portfolio?

Answer: For n assets, we use matrix notation for efficient calculation.
ou

Portfolio Variance (Matrix Form):

σp2 = wT Σw
eM

Where:

• w = n × 1 vector of asset weights

Th

• Σ = n × n covariance matrix

• wT = transpose of weight vector

 Value at Risk (VaR) - Comprehensive Study Guide 21

Covariance Matrix:
 
σ12 σ1 2 σ1 3 · · · σ1n
σ 2 1 σ 2 σ 2 3 · · · σ2n 
 2 
 2
· · · σ3n 
Σ = σ 3 1 σ 3 2 σ 3 
 .. .. .. .. .. 
 . . . . . 

n
σn1 σn2 σn3 · · · σn2

na
Where σij = σi σj ρij for i ̸= j.

Fi
Three-Asset Portfolio VaR:
Given:

of
• Portfolio value: $10,000,000

• Weights: w1 = 0.5, w2 = 0.3, w3 = 0.2

ld
• Volatilities: σ1 = 2%, σ2 = 3%, σ3 = 4%

or
• Correlations: ρ12 = 0.4, ρ13 = 0.2, ρ23 = 0.6
-W
• Confidence: 95%

Step 1: Weight Vector  

0.5
w = 0.3
0.2
h

Step 2: Covariance Matrix

t
Pa

σ12 = 2% × 3% × 0.4 = 0.024% (54)

σ13 = 2% × 4% × 0.2 = 0.016% (55)
σ23 = 3% × 4% × 0.6 = 0.072% (56)
in

 
0.04% 0.024% 0.016%
a

Σ = 0.024% 0.09% 0.072%

nt

0.016% 0.072% 0.16%

Step 3: Portfolio Variance
ou

  

0.04% 0.024% 0.016% 0.5
2
σp = 0.5 0.3 0.2  0.024% 0.09% 0.072%   0.3 (57)
eM

0.016% 0.072% 0.16% 0.2

= 0.0584% (58)

Step 4: Portfolio Volatility

Th

√
σp = 0.0584% = 2.417%
 Value at Risk (VaR) - Comprehensive Study Guide 22

Step 5: Portfolio VaR

VaR = $10, 000, 000 × 1.65 × 2.417% (59)

= $398, 805 (60)

n
9 Expected Shortfall (Conditional VaR)

na
9.1 Q14: What is Expected Shortfall and how is it calculated?

Fi
Expected Shortfall (ES) or Conditional VaR (CVaR) measures the expected
loss given that the loss exceeds the VaR threshold. It provides information about
tail risk that VaR ignores.

of
Mathematical Definition:

ld
ESα = E[L|L > VaRα ]

or
Where L represents losses and α is the confidence level.
Calculation Methods:
-W
1. Historical Method:

1 X
(1−α)n
ESα = L(i)
(1 − α)n i=1

Where L(i) are the sorted losses in descending order.

h

2. Parametric Method (Normal Distribution):

t

ϕ(zα )
ESα = µ + σ ×
Pa

1−α
Where ϕ(·) is the standard normal probability density function.
3. Monte Carlo Method: Average the simulated losses that exceed VaR.
in

Expected Shortfall Calculation:

a

Given:
nt

• Portfolio: $1,000,000

• 95% VaR: $25,000

ou

• 1000 historical observations

eM

• Normal distribution assumption: µ = 0, σ = 2.5%

Historical Method:

1. Identify losses exceeding VaR (worst 5% = 50 observations)

Th

2. Suppose these losses are: $25,100, $26,800, $28,500, ..., $45,000

3. ES = Average of these 50 losses = $32,400

 Value at Risk (VaR) - Comprehensive Study Guide 23

Parametric Method:

z0.95 = 1.65 (61)

ϕ(1.65) = 0.1031 (62)
0.1031
ES = $1, 000, 000 × 2.5% × (63)
0.05

n
= $1, 000, 000 × 2.5% × 2.062 (64)

na
= $51, 550 (65)

Interpretation:

Fi
• VaR tells us there’s 5% chance of losing more than $25,000

• ES tells us that when losses exceed $25,000, the average loss is $32,400 (his-

of
torical) or $51,550 (parametric)

ld
10 VaR Illustrations and Problems for Learners

or
This section provides a series of VaR illustrations and problems ranging from simple
to complex, with detailed solutions. These are designed to help learners practice and
-W
understand VaR concepts.

10.1 Simple Problem: Single Asset VaR

Problem: Calculate the 1-day 95% VaR for a single stock position worth $100,000. The
h

daily expected return is 0%, and the daily volatility is 1.5%. Use the parametric method
t

assuming normal distribution.

Pa

Solution:
Using the parametric formula:

VaR = P0 × |µ − zα × σ|
in

Where: - P0 = $100, 000 - µ = 0% - σ = 1.5% - z95% = 1.65

a
nt

VaR = $100, 000 × |0 − 1.65 × 0.015| = $2, 475 (66)

Interpretation: There is a 5% chance of losing more than $2,475 in one day.

ou

10.2 Intermediate Problem: Two-Asset Portfolio VaR

eM

Problem: Calculate the 1-day 99% VaR for a portfolio consisting of two assets: - Asset A:
$60,000, volatility 2%, expected return 0.05% - Asset B: $40,000, volatility 3%, expected
return 0.1% - Correlation between A and B: 0.5 Use the parametric method.
Solution:
Th

First, weights: - wA = 0.6 - wB = 0.4

Portfolio expected return:

µp = 0.6 × 0.05% + 0.4 × 0.1% = 0.07% (67)

 Value at Risk (VaR) - Comprehensive Study Guide 24

Portfolio variance:
σp2 = (0.6)2 (0.02)2 + (0.4)2 (0.03)2 + 2(0.6)(0.4)(0.02)(0.03)(0.5) = 0.000432 (68)
Portfolio volatility:
√
σp = 0.000432 ≈ 2.078% (69)

n
VaR at 99% (z = 2.33):

na
VaR = $100, 000 × |0.07% − 2.33 × 2.078%| ≈ $4, 772 (70)
Interpretation: There is a 1% chance of losing more than $4,772 in one day.

Fi
10.3 Complex Problem: Multi-Asset VaR with Horizon and
Confidence Conversion

of
Problem: For a three-asset portfolio valued at $200,000: - Asset 1: Weight 50%, volatil-
ity 1.8%, expected return 0.04% - Asset 2: Weight 30%, volatility 2.5%, expected return

ld
0.06% - Asset 3: Weight 20%, volatility 3.2%, expected return 0.08% - Correlations:
ρ12 = 0.4, ρ13 = 0.3, ρ23 = 0.6 Calculate the 1-day 95% VaR using parametric method.

or
Then convert to 10-day horizon and to 99% confidence level.
Solution:
-W
Portfolio expected return:
µp = 0.5 × 0.04% + 0.3 × 0.06% + 0.2 × 0.08% = 0.054% (71)
Covariance matrix elements:
σ12 = 0.018 × 0.025 × 0.4 = 0.00018 (72)
h

σ13 = 0.018 × 0.032 × 0.3 = 0.0001728 (73)

t

σ23 = 0.025 × 0.032 × 0.6 = 0.00048 (74)

Pa

Variance:
σp2 = (0.5)2 (0.018)2 + (0.3)2 (0.025)2 + (0.2)2 (0.032)2 (75)
in

+ 2(0.5)(0.3)(0.00018) + 2(0.5)(0.2)(0.0001728) + 2(0.3)(0.2)(0.00048) (76)

= 0.00032437 (77)
a

Volatility:
nt

√
σp = 0.00032437 ≈ 1.801% (78)
ou

1-day 95% VaR (z = 1.65):

VaR1,95% = $200, 000 × |0.00054 − 1.65 × 0.01801| ≈ $5, 835 (79)
eM

Convert to 10-day horizon:

√
VaR10,95% = $5, 835 × 10 ≈ $18, 450 (80)
Convert 1-day 95% to 1-day 99% (z99% = 2.33):
Th

2.33
VaR1,99% = $5, 835 × ≈ $8, 239 (81)
1.65
Interpretation: The calculations show how VaR increases with longer horizons and
higher confidence levels.
 Value at Risk (VaR) - Comprehensive Study Guide 25

10.4 CFA-Style Practice Problems

Below are sample CFA-style problems involving Value at Risk, including conceptual and
calculation questions.

10.4.1 Problem 1: Conceptual Understanding

n
Question: A portfolio has an annual 1% VaR of $45,000. Which statement is most

na
accurate?
A) The expected minimum loss over one year, 1% of the time, is $45,000. B) There
is a 99% probability that the expected loss over the next year is more than $45,000. C)

Fi
The likelihood of losing $45,000 over the next year is 1%.
Answer: A) The expected minimum loss over one year, 1% of the time, is
$45,000.

of
- Explanation: VaR states there is a 1% chance that loss will exceed $45,000 in one
year. In other words, 99% of the time, the loss should not exceed this value.

ld
10.4.2 Problem 2: Historical VaR Calculation

or
Problem: Suppose you have 100 days of daily portfolio returns. You arrange returns
from worst to best. For a 5% one-day VaR, which observation represents VaR?
-W
Answer: - 5% of 100 observations = observation number 5 (counting from the worst
return). - If the 5th worst return is -2.1%, and the portfolio is $1,000,000: - VaR =
2.1% × $1,000,000 = $21,000 - Interpretation: There is a 5% chance of losing more than
$21,000 in one day.
h

10.4.3 Problem 3: Variance-Covariance (Parametric) VaR

t

Question: A $500,000 portfolio has an expected daily return of 0% and daily volatility
Pa

(standard deviation) of 1.8%. Assuming a 95% confidence level (z = 1.65), calculate the
VaR.
Solution:
in

VaR = $500, 000 × (0% − 1.65 × 1.8%) = $500, 000 × (−2.97%) = −$14, 850
a

- Interpretation: There is a 5% chance that daily loss will exceed $14,850.

nt

10.4.4 Problem 4: Portfolio VaR (Two Assets)

ou

Problem: A portfolio has 60% in Asset X (volatility: 12%), 40% in Asset Y (volatility:
18%), with a correlation of 0.2. Calculate portfolio standard deviation and 95% VaR
with z = 1.65 (assume 0% expected return, portfolio value = $1,000,000).
eM

Solution: Portfolio standard deviation:

p
σp = (0.62 )(0.122 ) + (0.42 )(0.182 ) + 2(0.6)(0.4)(0.12)(0.18)(0.2) ≈ 0.1085 or 10.85%

VaR = $1, 000, 000 × 1.65 × 0.1085 ≈ $179, 025

Th
 Value at Risk (VaR) - Comprehensive Study Guide 26

10.4.5 Problem 5: Monte Carlo Simulation (Conceptual)

Question: Describe the steps to compute VaR using the Monte Carlo simulation in the
CFA context.
Answer: - Model return distribution for the asset/portfolio (often normal or log-
normal). - Simulate a large number (e.g., 10,000) of possible future returns. - For each

n
simulation, calculate the corresponding portfolio value/loss. - Sort the simulated losses;
for 95% VaR, pick the loss at the 5th percentile.

na
10.4.6 Problem 6: CFA Topic Integration/Interpretation

Fi
Question: If the one-day value at risk of a portfolio is $50,000 at a 95% probability
level, what does this mean?
Answer: - In only 5% of scenarios would you lose $50,000 or more in one day,

of
assuming models and assumptions hold.

10.4.7 Problem 7: Quick CFA Quantitative Practice

ld
Question: A $100 million portfolio, mean daily return of 0.05%, daily std deviation

or
1.2%. What is its 1-day 99% VaR (z = 2.33)?
Solution:
-W
VaR = $100, 000, 000×(0.05%−2.33×1.2%) = $100, 000, 000×(−2.746%) = −$2, 746, 000
t h
Pa
a in
nt
ou
eM
Th

This document provides a comprehensive overview of Value at Risk methodology. For

specific implementation details or regulatory requirements, consult current industry guide-
lines and regulatory publications.