One look at Bill Reese and you have to ask yourself, What kind of a football expert is this? He is a bookish, 70-year-old grandfather who's as likely to curl up in the kitchen with his computerized chessboard as he is to tune in to The NFL Today. He has never played football—never played any varsity sport, for that matter. Oh, he won the campus chess championship at Buffalo State College for four years, beginning when he was a 15-year-old freshman there in the 1930s. And if you count flying fighter planes against the Japanese in World War II as a sport, then Reese has some athletic credentials. But when the war ended, he left all that behind to pursue his doctorate in clinical psychology. Since then he has spent nearly four decades practicing psychology and running a mental health clinic in Williamsburg, Va.
And picking football games.
Reese has been picking college and pro football games every season since the fall of 1946. He has averaged more than 1,000 picks a season. Each forecast is painstakingly calculated at the kitchen table on the pages of a dime-store spiral notebook, which is then stored in the basement of his home at the end of one of Williamsburg's tree-shaded lanes. Since 1961 those picks have been printed in the sports section of Williamsburg's Virginia Gazette.
Plenty of people have relied on Reese when making wagers. Professional gamblers have offered him cash for his system, and no wonder. He says his success rate is "about 77 percent for colleges, 70 percent for the pros." But Reese is not a betting man, and his system is not for sale. Not that anyone with less than a roomful of personal computers could understand it even if he did hand it over.
Those mountains of notebooks are jammed with tens of thousands of intricate numerical codes, figures that indicate the psychological—yes, psychological—status of more than 650 college teams, plus all the pros. Reese is the first to admit he knows little about Denver's defensive schemes or the ability of New Orleans's offensive line. He doesn't need to: Playbooks and passing stats are not what his system is about. "What this is about," says Reese, running a hand over his precious notebooks, "is changes, adjustments, improvements—learning. I guess what you would call me is a people understander."
Which is what brought Reese to football back in 1946, when he was a student in Ohio State's doctoral program in psychology. Among his classes was an advanced statistics course taught by Herbert Toops, who played a role in the early days of UNIVAC, the world's first commercially available large computer. "Toops was brilliant, really a brain," says Reese, who was one of only four students in Toops's class. While his three classmates focused on more traditional projects, Reese started thinking about football and group learning—about how people learn to perform physical skills as a unit, as a group, as a team. "There is psychology to it," says Reese of team sports. "These are not machines. They are individuals, learning to perform together, to become cohesive."
Team sports such as baseball and basketball, Reese says, can be dominated by an individual talent—an overpowering pitcher, an extraordinary center. According to him, those sports also are given to sporadic, often unexplainable streaks. Football, however, with its regular weekly rhythm, seemed the most apt model for a study. The game was prone to patterns—and rich with data.
Reese found his original data in 1946 by charting the results of the previous three seasons for more than 100 college football teams. With those numbers in hand, he studied certain factors in each week's game, which he referred to as a "test." Did a team seem to test better away or at home? How did it perform the week after a loss? How did it perform the week after a humiliating loss? How did it perform after a win? Did it show a tendency toward a letdown after a strong performance? What were the answers to those same questions for the team it was playing in the coming week? And how did a team do the last time it met a particular team?
None of these questions or their answers, notes Reese, were startling or new. "It's what you do with them that makes the difference," he says.
What Reese did was to assign different weights to each of the factors and then draw a graph—"a standard learning curve," Reese calls it. Next he created the key to his system, his "slide rule," an intricate formula into which he plugs numbers based on a team's performance each week. The numbers spit out by Reese's slide rule are charted on his learning curve. The higher a team's place on that curve—and the ranking often shifts dramatically from week to week—the higher its "power rating," as Reese calls it. One team's power rating may be higher than another's, but that does not mean the system will necessarily pick the one with the higher rating to win. It is the direction of the teams within the system—up or down the curve and at what rate—that determines the favorites on a particular weekend. "It's a question of who is learning and how well," says Reese. "This is an ongoing process, a continuum."