Quantitative Risk Management: A Practical Guide to Financial Risk, + Website

7.1 Project Implementation

Risk measurement is an applied science and, as such, we need to take the theoretical ideas and actually make them work. Make them work on computer systems, with complex and messy data, used by people with varying degrees of sophistication and knowledge.

Risk projects are as much about boring data and IT infrastructure as about fancy quantitative techniques. In building or implementing a risk management project, roughly 80 percent of the effort and investment is in data and IT infrastructure, and only 20 percent in sophisticated quantitative techniques. I cannot overemphasize the importance of data and the IT infrastructure required to store and manipulate the data. The bottom line is that if you don't know what is in the portfolio, it is hard to do any sophisticated analysis on the portfolio. For market risk, but credit risk in particular, good records of positions and counter parties is critical, and these data must be in a form that can be used.

Data

Data are always a big issue. Obtaining and using data is often more challenging than anticipated. Good quality and timely data, however, form the bedrock of any risk project.

Data can roughly be separated into external and internal data. External data are items such as history on market risk factors or security characteristics. (Is that bond the trader bought this morning maturing on the 15th or 31st of August?) We need to collect, clean, warehouse, update, and distribute these data.

Internal data are sometimes ...