My Computer Is My Bookie

Before you begin reading, go ask your roommate if you can borrow his pocket calculator. You'll probably need it as you peruse this column, because Mark Zbikowski and Tim Matthews aren't your average pigskin prophets. While members of the Cube sit in silent revery and appeal to divine inspiration when it comes to making predictions, Zbikowski and Matthews use the Science Center computer and a chain of statistical formulas to predict the outcomes of college games across the nation.

Both prognosticators played high school football but turned to the cerebral side of the game when they came to Harvard. Matthews broke his arm playing high school ball back home in Middlesboro, Kentucky; Zbikowski tore up his knee while serving as captain of his schoolboy 11 in Detroit, and both have become reconciled to playing House football.

The two met while serving as terminal supervisors at the Science Center and have been comparing notes ever since. Both use a system known as the "least-squares regression model" to get their point spreads. The mustachioed Zbikowski spent his sophomore year ironing out the kinks in his program, and now that he's a senior is making predictions for the second year.

Zbikowski put in a "man-month" tinkering with his program and then went to Lew Law, the director of the Science Center, who liked the idea and gave Zbikowski an independent computer account.

Zbikowski's model is based on a numerical ability ranking for each team, which he refines after each week's play. I'd write the formula out for you but the typists don't have Greek letters at their disposal. Zbikowski ranks 153 college teams, including Ball State, McNeese State and Fullerton State, and gives point spreads for their games. Texas is at the top of his poll, as the Longhorns boast an ability rating of 35.87, while Harvard is ranked 120th with a rating of -1.37 and Davidson is in the 153rd spot, with -24.53

Both Zbikowski and Matthews are right about 75 per cent of the time when it comes to predicting the winning team, and beat the spread around 60 per cent of the time. Matthews began using the computer to make his predictions as a freshman, at the promptng of his father, a math professor at Lincoln Memorial University, who had been making football predictions for 25 years.

Matthews recalls that his father originally used a less accurate methods known as the "modified simplex method," which is a linear programming technique, because he lacked access to a computer.

When Matthews arrived in Cambridge, however, the situation changed and his father began feeding him reams of data to run through the computer.

Last year, Matthews' father began trying to market his predictions by advertising in "Football News," a weekly sports tabloid. Matthews pere has yet to cash in big on his prognosticating powers but this season "Football News" has begun to print young Matthews' selections every week under the heading, "Matthews' grid ratings."

Some professional prognosticators such as Danny Sheridan claim psychic powers. A novel technique of predicting by using biorhythms is also gaining a foothold in the trade, but Matthews's advise to would-be soothsayers is: "You have to have a good math background, know what statistical methods are available and how to use them, and a high-speed computer." ZBIKOWSKI  The Top Ten  MATTHEWS 1. Texas    1. Texas 2. Ohio St.    2. Ohio St. 3. Notre Dame    3. Notre Dame 4. Kentucky    4. Kentucky 5. Penn St.    5. Michigan 6. Nebraska    6. Pittsburgh 8. Alabama    7. Alabama 9. Oklahoma    8. Arkansas 10. N. Carolina    9. Penn St. 1. Penn by 2  Ivy Predictions  1. Harvard by 4 2. Brown by 4    2. Brown by 1 3. Yale by 3    3. Yale by 3 4. Columbia by 2    4. Cornell by 3