- error prone
- not databases
Earlier this year I was asked to develop an attribution model for a client that employs option writing along side its long stock positions. It's a global firm, so the model has to also be sensitive to currency movement, though since the client doesn't employ a hedging strategy, naive multi-currency attribution works.
I completed the assignment, and delivered the spreadsheet. Because of its complexity, the firm will most likely not employ it, which I can fully understand.
I do see the exercise as a way to prototype a model, so as to gain confidence in the approach, formulas, assumptions, etc. From such an effort a programmed system can be built.
This exercise will also likely result in an article for The Journal of Performance Measurement(R), to explain the model as well as to further support the use of spreadsheets to prototype models, but not to turn into "systems."