Market Risk: Bitemporal Model – Part 1


Modelling a trade and the events that occur on a trade is always an interesting exercise. It’s become clear over the years however that a Bi-Temporal model is the only real way to go. This model is the lifecycle of the trade based on events and actions. The reasoning behind the need for bi-temporal model is quite simple. If you imagine an IRS trade, it gets created (New Trade $100m), amended, followed after sometime by a partial unwind ($50m), and amendment, another partial unwind ($40m).

Historically if the first unwind is found to be incorrectly entered for some reason (maybe it should have be $60m), the values and calculation from that point onwards will be wrong. The fix for this is usually to amend the trade (current view) to resolve the issue. However what you actually want to do is go back to the event/action that had the error and amend and then reply any appropriate events to ensure the current trade view is correct. This leads to the need for a bi-temporal model that offers a lifecycle of events and actions on a trade. Essentially a trade has a system and effect timeline.

What is interesting is that FpML provides us with a very good model for the product of a trade. This fits nicely since a trade from a bi-temporal viewpoint is really a trade id which has a product attached.

Below is a screen show of an XLS that provide an FX and IRA trade, coupled with the calculations for working out PV, and hence PV01. It ignores Stubs (a calculation period is not of a ‘regular’ length e.g. if you have a 5 year 2m swap and the calculation periods are six months, then there will be a stub), holidays and there is no compounding to keep things simple. It also assume its a FRA with payment being made at the end of the periods. Pay is a negative value, as the payment is going out, receive is positive. Future Value (FV) and Present Value (PV) are displayed for each cash flow. 365 DCB is assumed for the yield curve, again to make things simple. 6×12 etc provide the rate from 6mths to 12mths. With the calculation of NPV, we can bump the curve and hence calculate PV01 would could then possible be displayed on the traders desktop (in this scenario) as USD Parallel Risk – bucketed by currency. If we were to calculate the partial PV01, we’d bucket by the yield curve time intervals, and then display accordingly.

~ by mdavey on March 5, 2010.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

 
%d bloggers like this: