Global Cooling

Wall Street and the BCS

As our national economy is awash in bailout news, from Citibank to AIG to General Motors, it is worth noting that this is not a new phenomenon.  Recall September 1998, the Russian Debt Crisis and the spectacular demise of Long Term Capital Management, the hedge fund run by former Salomon Brothers Vice Chairman and Head of Bond Trading John Meriwether, along with 1997 Nobel Laureates Myron Scholes and Robert Merton.  The fund, in spite of its blue ribbon management team, blew up fantastically and was eventually bailed out by the New York Federal Reserve Bank to the tune of some $3.6 billion.  And what was at the root of the fund’s collapse?  The trading strategy employed by these geniuses was highly dependent upon a computerized algorithm, a “model,” to determine value and evaluate risk. 

Fast forward to the present and ask yourself just what has allowed so many bright minds on Wall Street fall victim to the sub-prime debacle?  How can institutions such as Fannie Mae, Freddie Mac, Bear Stearns, Lehman Brothers, Merrill Lynch, Wachovia, Washington Mutual, AIG, Citibank and Goldman Sachs, institutions managed by the crème de la crème of top tier executives and whom annually attract the best and the brightest from our nation’s finest business schools, have allowed themselves to be entrapped in a house of cards?  You will no doubt find, when searching for answers, that “financial engineering” is high on the list of sins. 

We have become overly enamored with the computational abilities of the modern day semiconductor and spreadsheet and, as a result, have attempted to apply mathematics and physics to fields far beyond their scope in search of solutions to the social sciences; which are much more social than science.  Put simply, computer powered mathematical models, regardless of their complexity, do a very poor job of accounting for systems that contain numerous interrelated variables, especially those of the human behavior kind.

Enter the Bowl Championship Series.  While choosing two teams to play for all the marbles via a beauty contest is bad enough, interjecting a computerized formula based component and inferring that this portion is scientific is simply madness.  The computer rankings, despite the assertions of their creators, are pseudo-science at best and failed alchemy at worst.  The six computer models that mean life and death to college football fans across the country cannot even accurately pick winners against the spread on individual games, yet we allow these same algorithms decide who is going to what bowl.  

Mathematical formulas are highly efficient in solving equations with limited variables and in situations where logic and reason always prevail over emotion: Science.  It is quite hard to theorize that on any given fall Saturday, some fifty FBS College Football fields across the country are populated with twenty-two 18 to 20 year olds, a gaggle of middle aged coaches and a set of referees who are all making cold, hard, unemotional scientific decisions at every moment of every game.  As of yet, we have not developed an algorithm to account for the human element.  You need look no further than Wall Street for confirmation.

Like LTCM, College Football is putting too much faith in the model.  This year, the model has even gone so far as to override an outcome already determined on the field (UT vs. OU.)  If football were science, models could be produced that would pick winners every week and someone would become quite wealthy as a result.  Football is not science.  The ONLY way to determine a champion is on the field. 

*Strength of Schedule is taken from NCAA Statistics and is calculated vs. other FBS teams only.  Games vs. Non-FBS teams lowers the total number of opponent games, overweighting other games.  Excluding the Oklahoma game vs. Chattanooga (1-11), OU’s opponents are a combined 70-42.  Throwing out Texas’ worst opponent, UTEP (5-7), UT’s opponents are a combined 71-42.  This is contrary to what Bob Stoops would like you to believe.