LINTHICUM, MD, March 8 - A Yale operations researcher who studies AIDS prevention will turn to a lighter subject when he delivers a scholarly paper on playing March Madness office pools at the spring convention of the Institute for Operations Research and the Management Sciences (INFORMS®).
"We decided we'd like to win once in a while," explains Edward H. Kaplan, the William N. and Marie A. Beach Professor of Management Sciences at the Yale School of Management, who wrote the paper with co-author Stanley J. Garstka, Deputy Dean of the Yale School of Management. He laments, "We have unsuccessfully participated in such pools for more than a decade."
Operations research is noted for the strong tools it provides decision makers. In tests, the Yale model achieved overall prediction accuracy of about 58%, potentially offering those who use the model an edge on their office mates.
Billions and Billions of Possible Outcomes
"The idea is to select a slate of 63 predicted game winners to maximize the expected points you gain in an office pool, subject to the constraints of consistency in predictions - I can't simultaneously pick Duke to lose in the first round and win in the second - and the point rules of the pool," he explains.
In the NCAA tournament, there are a staggering 9.22 billion billion possible tournament outcomes, Professor Kaplan explains. Yet if one filled out a slate of predictions completely at random subject only to a consistency in prediction constraint, one could expect to predict the winners correctly about a third of the time.
Professor Kaplan and Deputy Dean Garstka explain that the object in an office pool is to maximize total pool points, where different points are awarded for different correct winning predictions. Their model studies the structure of single elimination tournaments and shows how to efficiently calculate the mean and variance of the number of correctly predicted wins - or, more generally, the total points earned in an office pool - for a given slate of predicted winners. They then show how to determine optimal office pool predictions that maximize the expected number of points earned in the pool.
Considering probability models for predicting game winners based on regular season performance, professional sports rankings, and Las Vegas betting odds, they compared their predictions with actual results of past NCAA and NIT tournaments. The model achieved overall prediction accuracy of about 58% using this historical data.
Professor Kaplan and Deputy Dean Garstka note that their models outperform the simple strategy of picking the seeds for a more sophisticated point structure, but do not surpass picking the seeds when the goal is to predict as many game winners as possible.
Operations Research Models Have Broad Applications
Interestingly enough, says Professor Kaplan, a similar viewpoint to that espoused in the paper can be applied to saving lives of those at risk of contracting AIDS. "It is the breadth of problems that can be addressed by operations research that is appealing," he explains. "Playing the office pool is fun, but preventing HIV infections is obviously more serious and much, much more important."
Professor Kaplan will present the paper, "March Madness & the Office Pool," on Sunday, May 2 in Salt Lake City at the national convention of INFORMS. At the convention, he will also present the Omega Rho Distinguished Lecture: Policy Modeling for Better Decisions - The Case of HIV Prevention.
Professor Kaplan has contributed numerous studies and articles to the literature on AIDS prevention. In 1993, INFORMS presented Professor Kaplan and the New Haven Health Department - AIDS Division with the prestigious Franz Edelman Award for a paper, "Let the Needles Do the Talking! Evaluating the New Haven Needle Exchange." Professor Kaplan also received the 1994 Frederick W. Lanchester Prize for twelve papers on AIDS modeling.
The Institute for Operations Research and the Management Sciences (INFORMS®) is an international scientific society with 12,000 members, including Nobel Prize laureates, dedicated to applying scientific methods to help improve decision-making, management, and operations. Members of INFORMS work in business, government, and academia. They are represented in fields as diverse as airlines, health care, law enforcement, the military, the stock market, and telecommunications. The INFORMS website is at http://www.informs.org .
The Yale School of Management Website is http://www.yale.edu/som . Contact at Yale SOM is Karin A. Nobile, Director, Media Relations, (203) 432-6010 or karin.nobile@yale.edu .