Practitioner course in risk management
Value at Risk
How value at risk is calculated, tested and misused: historical simulation, variance-covariance and Monte Carlo methods, volatility-weighted variants, time scaling, regulatory backtesting, and what VaR does not tell you.
Value at risk compresses the risk of a portfolio into one number: a loss that should be exceeded only on a small share of days. Since the 1990s it has been used by banks, regulators and asset managers to set limits and capital. It is useful, easy to misread, and the subject of this course.
How is value at risk calculated?
There are three standard methods. Historical simulation applies past changes in market prices to today’s portfolio and reads off the loss at the chosen percentile. The parametric method assumes a distribution, usually normal, and uses volatilities and correlations. Monte Carlo simulates many possible scenarios from a model. You will implement all three on one portfolio and compare them.
How do you know if a VaR model is working?
By backtesting: counting how often actual losses exceeded VaR and testing whether the count, and the timing of those exceptions, is consistent with the model. Too many exceptions, or exceptions that cluster, mean the model is failing. The course covers the statistical tests and the regulatory traffic-light approach.
Who is this course for?
Risk analysts, portfolio managers and traders who use or report VaR. It counts toward the Certificate in Quantitative Risk Management, the Certificate in Quantitative Portfolio Management and the Professional Diploma in Algorithmic Trading.
What you will be able to do
- Calculate VaR by historical simulation, the parametric method and Monte Carlo.
- Use volatility-weighted and filtered historical simulation.
- Backtest a VaR model with the Kupiec and Christoffersen tests.
- Aggregate VaR across positions and explain diversification effects.
- Explain the weaknesses of VaR and when expected shortfall is preferred.
Syllabus, module by module
What VaR measures
Definition, confidence level, horizon and the quantile of the loss distribution.
Parametric VaR
The variance-covariance method, delta-normal VaR and its assumptions.
Historical simulation
Plain, age-weighted and volatility-filtered historical VaR.
Monte Carlo VaR
Simulating risk factors, full revaluation and non-linear positions.
Aggregation
Component, marginal and incremental VaR; time scaling and its limits.
Backtesting
Exceptions, the Kupiec and Christoffersen tests and the Basel traffic light.
Limits of VaR
Tail blindness, non-subadditivity and the move to expected shortfall.
Project lab
Supervised work on the assessed VaR project.
How the certificate is earned
An assessed project: you build and backtest three VaR models for a multi-asset portfolio over a period that includes a crisis, and submit the code with a report. On a pass you receive the Course certificate in Value at Risk, issued by the Institute with a verification code. Read how certification works.
Before you start
Probability and statistics to the level of distributions and quantiles, and working Python.
This course counts toward
- Certificate in Quantitative Portfolio ManagementSix courses and a capstone on building portfolios with numbers: mean-variance and its failures, robust optimisation, Black-Litterman views, factors, attribution and risk.
- Certificate in Quantitative Risk ManagementA financial risk management certification in five courses and a capstone on measuring and governing market risk: VaR and expected shortfall, volatility models, stress tests and the validation of the models themselves.
- Professional Diploma in Algorithmic TradingThe top credential on the practitioner path. Eight courses and a 40-hour capstone covering everything between a trading idea and a book you could defend to a risk committee.
Questions applicants ask
What does a VaR figure mean?
A one-day 99% VaR of 1 million means that, if the model is right, losses should exceed 1 million on about one day in a hundred. It says nothing about how large losses are on those days.
Which VaR method is best?
Each method trades off assumptions against data needs. Historical simulation avoids a distribution assumption but depends on the window; parametric methods are fast but understate fat tails; Monte Carlo handles options but is costly. The course tests all three on the same portfolio.