|
|
|
|
 |
Daily Speculations The Web Site of Victor Niederhoffer & Laurel Kenner Dedicated to the scientific method, free markets, deflating ballyhoo, creating value, and laughter; a forum for us to use our meager abilities to make the world of specinvestments a better place. |
Write to us at:
(address is not clickable)
05/09/2004
The Lady Tasting Tea by David Salsburg
Book Review by Victor Niederhoffer
The Lady Tasting Tea, by David Salsburg, is an excellent book describing the attempt to understand the variation that exists in the world of science. The conditions before they started relating to such common areas of inquiry such as how to increase crop yields was "a mess of confusion and vast troves of useless data."
A sophisticated argument to prove that there did not exist a single uniform quantity in an experiment but a distribution of outcomes which on successive tests would show regression to the mean is given. He looks at the observations on an individual (or stock) in successive periods and notes a regression to mean. And then he devolves a tremendous insight. "Why there has to be a regression. IT HAS TO BE SO."
If only spec investors considering market phenomena would look at the scatter they observe from their own counting of differences at various times and after various events and say the same thing regarding observed differences. "Why there has to be a difference because of randomness." His approach to explaining why regression has to occur involving a successive path of more and more extreme individuals in successive generations without regression is a bit different from Stigler's approach to the problem, namely that any observation consists of a skill part plus an ephemeral one, with a good rule of thumb or starting hypothesis being that they are both equal. Thus, an observation that 10 away from the mean might contain a skill part that is persistent of 5 and a random or ephemeral part of 5. This led to Pearson's revolutionary idea that we do not look upon experimental results as carefully measured numbers, but as examples of a scatter of numbers, a distribution of numbers. This distribution can be written as a mathematical formula that tells us the probability that an observed number will be a given value. But what value that number actually takes in a specific experiment is unpredictable. The results of individual experiments are random, in the sense that they are unpredictable.
The idea that "whatever we measure is really part of a random
scatter" is central to all science and in particular to spec
investors. If only they would realize that when they observe
differences from the base level after some seasonal or chart or
system or Fed event, such differences must occur and are part
of the random scatter all experiments will generate. If only
investors would then try to tease the real from the random out
of their results, their world would be a much more rational,
calm and profitable place. The book and the realization are
highly recommended.
-- Victor Niederhoffer