Red-Blooded Risk

Aaron Brown has worked as a risk manager at trading firms and banks, and this book combines a personal history of quantitative risk management with an argument about what risk management is for. It is unconventional in form, with digressions on the history of gambling, on money, and on the culture of the people who built the field.

His central position is that risk management exists to enable risk-taking rather than to reduce it. A firm that manages risk well can take more of it, in the places where it is paid to, because it knows what it holds. He is critical of the compliance-oriented risk function that produces reports nobody uses, and of the belief that value at risk failed in 2008. His argument on value at risk is that the number was never meant to describe tail losses; it is a measure of whether the firm's models are working, and a breach frequency higher than the stated confidence level tells you the model is wrong, which is the information it was designed to produce.

The technical content covers the Kelly criterion and optimal bet sizing, the distinction between money and risk as separate quantities to manage, the use of risk limits as a control on model error rather than on loss, and the behaviour of markets with fat tails. Brown writes informally and the structure wanders. The book is best read for its arguments about what risk numbers mean and how a risk function should relate to a trading business, which are rarely stated as clearly elsewhere.