Peter Cotton, PhD

ML Score
Intech Investment Management LLC | Senior Vice President. Chief Data Scientist.
Is Claude Shannon the true creator of the Kelly criterion? The attached article by Graham Giller begins with this most intriguing possibility: namely that Shannon wished to publish work related to optimized gambling, but needed Kelly to be the fall guy in the event of ... how shall I put this ... puritanical predispositions regarding probability and its applications? "However, Dr. Z told me that he and Thorp secretly suspected that the paper was written by Shannon himself, and that Bell Labs at the time of publication (the 1950’s) persuaded Kelly to provide cover for an analysis of such an immoral topic."If this came to us via some other route it might be easier to dismiss, but Dr Z. and Thorp would have made for quite the witness list in an alternative McCarthyian universe - an American version of the cultural revolution I envisage where everything and everyone related to gambling is cleansed. Dr Z. was a handle adopted by the late Prof William Ziemba, when writing on matters regarding racing and wagering. And the picture Graham includes reminds us of one of the more outlandish exploits of Ed Thorp, detailed in Thomas Bass' book "The Newtonian Casino" - surely mandatory reading for anyone with a passing interest in mechanics, chaos, or just gambling chicanery. I digress, but Bass' original title for Thorpe and Shannon's "immoral" adventure was "The Eudaemonic Pie". There was an ambition to create a scientific commune whose members might be freed of all material distractions by roulette profits. I'd ask you to reflect on the net benefit, or otherwise, of the use of one type of label for contingent claims versus another (gambling/investing). It is a distinction so fine that most formal attempts just give up and equate it to the sign of a mean of expected return. I doubt having one chapter on positive returns and one on negative will help students of finance, though.