A trading-desk risk manager's textbook on measuring and limiting size, not a strategy book. It teaches value-at-risk, hedge ratios, and options Greeks with real formulas and worked numbers, but it never once tells a reader when to enter or exit a trade.
The one idea
Edwards worked as a risk manager at a hedge fund and later at Bank of America, and the book reads like the training manual he wished he'd had. Its actual subject is value-at-risk (VaR): a single number, built from a confidence level and a time horizon, that converts every kind of position (stocks, futures, options, credit) into one comparable measure of how much money could be lost. A firm sets soft and hard VaR limits per trader and per desk the same way a speed limit works: it does not tell you where to drive, only how fast you are allowed to go.
The book spends its middle chapters showing how that one number gets built and abused. It walks through estimating volatility four different ways (simple historical, exponentially weighted, GARCH, and options-implied), combining positions into a portfolio VaR that depends on a shaky correlation estimate, and backtesting the result against real P&L to see if the model is honest. It closes with the mechanics that sit downstream of VaR at a real desk: hedging with a minimum-variance ratio, pricing the option Greeks that make VaR's assumptions break down, and calculating the credit-value adjustment a bank charges a counterparty for default risk.
None of this is a trading system. There is no chapter on chart patterns, no entry rule, no stop-loss method. The 'technique' in the subtitle refers to institutional risk technique: how a firm measures, limits, and reports the size of what its traders are doing, not how an individual trader decides what to buy.
What it actually teaches
The book's central, quantifiable process is building and checking a value-at-risk number, which it lays out step by step in Chapter 6.
SET THE HORIZON AND CONFIDENCE LEVEL
Pick a time window (commonly one day or one week) and a probability (commonly 90, 95, or 99 percent). A '$1 million, 95 percent one-day VaR' means a 5 percent chance of losing more than $1 million in the next 24 hours. Longer horizons need more data and get harder to calculate reliably.
ESTIMATE VOLATILITY
Choose a method: equally weighted historical volatility (the book's default, using 3 months to 2 years of data, minimum 30 to 50 samples), exponentially weighted volatility (a decay factor lambda, typically 0.94 to 0.98, so a value at lambda 0.94 loses 99 percent of its weight after about 75 days), a GARCH(1,1) model, or implied volatility pulled from options prices.
CONVERT TO A DOLLAR THRESHOLD
Multiply the volatility by the z-score for the chosen confidence level (1.645 for 95 percent, 2.326 for 99 percent, from the normal distribution) and by the position size. Worked example from the book: a $200 million gold portfolio with 15 percent annual volatility gives a daily volatility of 0.945 percent; at 95 percent confidence that is 1.554 percent, so the one-day VaR is $3.11 million.
COMBINE ACROSS POSITIONS
Portfolio VaR is not the sum of individual VaRs. The formula is VaR_portfolio squared equals VaR_A squared plus VaR_B squared plus two times correlation times VaR_A times VaR_B. The whole calculation lives or dies on the correlation estimate, which the book itself flags as unstable and prone to breaking down exactly when it matters most.
BACKTEST THE NUMBER
Compare the VaR threshold to actual daily results. A 95 percent VaR should be breached about 1 day in 20. The book's own NYMEX crude oil example found 5.58 percent of days breaching the threshold against an expected 5 percent, meaning the model was under-calling risk and needed recalibration.
What it looks like Monday morning
This is not a book that changes what a retail trader does on Monday morning, because it was never written for one. A reader who works, or wants to work, on a trading desk or in a risk function would come away able to build a parametric VaR calculation from scratch, size a hedge with the minimum-variance ratio instead of guessing a 1:1 match, and explain why a firm's VaR number can look calm the day before it blows up.
For anyone managing their own account, the practical takeaway is smaller: position size should scale with volatility, not with a flat dollar or share count, and a hedge that is not sized to the correlation between the two assets can add risk instead of removing it. Everything else in the book operates one level up, at the desk or fund level, not the individual trade.
| Confidence level | Probability loss exceeds VaR | Standard deviations |
|---|---|---|
| 90% | 10% | 1.282 |
| 95% | 5% | 1.645 |
| 99% | 1% | 2.326 |
These are the multipliers parametric VaR uses to convert a volatility estimate into a dollar threshold; a 95 percent one-day VaR is 1.645 times the one-day standard deviation of P&L.
Reducing risk is not always a smart business decision.Davis W. Edwards, Risk Management in Trading
Where it fails
- It is shelved as day trading but written for institutions. There is no chapter on entries, exits, chart setups, or position sizing for an individual account anywhere in the book. Every example is a trading desk, a hedge fund, or a bank managing a book of many traders' positions against firm-wide limits. A retail day trader picking this up expecting a method will find none.
- It assumes a math background most traders don't have. Chapter 3 (Financial Mathematics) covers calculus, variance-covariance matrices, and stochastic processes as prerequisites for the risk chapters that follow. Without that foundation, the core VaR chapter's matrix notation and the Black-Scholes derivation in the options chapter are close to unreadable.
- The book's own case studies show VaR failing at the moment it mattered. The Barings collapse case study in Chapter 6 notes that Nick Leeson's position would have shown a 95 percent monthly VaR of about $835 million, larger than the firm's entire equity capital, meaning the number existed but nobody was aggregating and enforcing it. VaR as a limit only works if someone is actually watching it in real time, which the book admits is an organizational problem, not a math problem.
- The framework it teaches was already being superseded when it was written. The book flags in its own closing VaR section that expected shortfall 'may end up replacing VaR at some point in the future' because expected shortfall is sub-additive and VaR is not. Basel's post-2016 Fundamental Review of the Trading Book pushed banks toward expected shortfall for regulatory capital, which means the institutional audience this book targets has since moved partway off the exact metric the book spends its longest chapter on. There is only one edition (2014), so none of this has been updated.
- About an eighth of the book is quiz questions and an answer key. Each chapter ends with a 'Test Your Knowledge' multiple-choice section, and the Answer Key runs from page 269 to 299, roughly 30 pages of a 320-page book. Useful for a student studying for a risk-management interview, dead weight for someone reading to extract a technique.
Who it is for
Buy it if
You are studying for, or working in, a trading-desk or risk-management role and need the vocabulary (VaR, minimum-variance hedge ratio, Greeks, CVA) explained with real formulas and worked numbers rather than jargon. It also works as a refresher for anyone who has to read a bank's or fund's risk report and wants to know what the numbers actually mean.
Skip it if
You trade your own account and want a method: entries, exits, stop placement, or how much to risk per trade. This book never addresses individual trading decisions. It also is not a good first finance book; it assumes calculus and statistics going in.
As an Amazon Associate I earn from qualifying purchases. It costs you nothing extra, and it does not change what gets recommended or what the verdict says.