I was reading an interview with Orrin Pilkey and Linda Jarvis-Pilkey, and found the following question both fascinating and relevant: 

Q: You've written numerous books on coastal hazards and how we should respond to them. Why did you want to write this book about the abuse of mathematic models?

Orrin H. Pilkey and Linda Pilkey-Jarvis: For more than twenty-five years we have monitored beach nourishment projects around the United States. In order to secure federal funding and justify the enormous costs of these projects, anyone undertaking one must make a prediction of how long the sand will last on the replenished beach. The predictions are based on mathematical models that are said to be sophisticated and state of the art, and yet are consistently, dramatically wrong-always in an optimistic direction. In the rare instances when communities questioned the models after the predictions of a long healthy replenished beach clearly failed, the answer typically was that an unusual and unexpected storm caused the error. Well, the occurrence of storms at any beach is neither unusual nor unexpected. Eventually we became interested in how models were used in other fields. When you start looking into it, you find that a lot of global and local decisions are made based on modeling the environment. There are some fascinating (and discouraging) stories of model misuse and misplaced trust in models in the book.

Kim Zussman comments:

This relates to the question of "what to study?", and the fact that well-described numerical analysis can be used to convince in most arguments.

For example, say you are feeling quite bullish. At any moment, there are scores of patterns which could be described (what happened this hour, morning, day, overnight, week, month, in reference to prior periods, events, other markets, valuation metrics, etc). Since your job is convince yourself to buy, your objective/multiple-hypothesis approach combines with carpe trade'em urgency to screen up bullish-testing patterns.

Of course you may find, to your surprise, that most of the tests are bearish. In which case you can argue that the run of bearish historical results is ready to be broken.





Speak your mind


Resources & Links