Monday, March 2, 2009

History of Value at Risk or V@R

(c) Raymond May 2008

It was 1989, Angel Court, City of London the swaps trading floor for JP Morgan; Michael Einhoven recently replacing Connie Vostad as group head asked me into his office. With his feet on his desk and picking his nose he complained that he didn't understand all the risks that the group was running, was it not possible to have a single number that described these risks, rather than spread this, gamma that.

In the prior three years I had completed two large system developments using JP Morgan's mainframe infrastructure, the first captured all the Swap deal data and was used to operational support the business and produce accrual accounting results. The second supported mark-to-market accounting, gap analysis and position descriptions. The second system was way before its time, and found no takers, it was many years before any of these models became standard.

So I embarked on a project to describe the risks in the swaps business as a single number. Using a research document produced by Salomon Brothers that stated that all changes in the yield curve could be described in three parameters - as a parallel change in rates, a steepening and a humping. Based on this we generated our first simple number based on the square root of the squares of three simple observations - a 10 basis point parallel shift - a 25 basis points steepening and - a small 5 basis point humping in the middle.

I also began to collect in a spreadsheet all daily pricing parameters. Gradually as this data was collected, day by day options for calculating a more accurate representation increased. I asked our research group in New York if they could provide me with any help. Next thing I knew I was sharing my desk with Dr. Gustav. Together we generated a model to describe a position in terms of the two standard derivations loss based on historical rates. Very simply this stated " the amount of money that the portfolio's loss or gain due to market movements would not exceed 19 out of 20 days" - bingo we had the number! the Value at Risk (note we initially we set it at 95% confidence, the market has since moved to 99%)

I finally had this number which we dubbed Value at Risk in place at the end of 1991. I had developed a spreadsheet which all portfolio managers in the Swaps Group now used. This took a starting position from the mainframe in the morning, added all new deals and calculated the portfolio - inter-day position, daily profit and loss and final Value at Risk.

At the start of 1992 I left London to join the Trading desk in New York where my first task was to introduce my position spreadsheet to our US dollar operations. By 1994 I was the lead trader for US Dollars and one day I was summoned by the current top man Peter Hancock. Peter complained that he had a Value at risk for each currency but he had no overall Value at risk, and he needed one as soon as possible.

I took myself off the desk for a week and constructed a new spreadsheet which expanding the model for all currencies. The spreadsheet automatically extracted data from all the individual spreadsheets without the users even knowing and I was able to run a total picture by 4pm in New York. The results were presented daily at a risk management meeting which would be dubbed the 4:15 Meeting. The spreadsheet I developed would be called the "Ray May Spreadsheet" and would remain in place until eventually being replaced at the end of the 90's by a 40 million dollar project named "Pyramid".

JP Morgan Research Group released the Value at Risk model and the associated data to the general market as Risk Matrix in 1994 as part of a campaign to persuade Congress that derivatives where not a risk to the system following P&G and Gibson Greeting Inc large derivative losses.

No comments: