SI Vault
 
DOING IT BY THE NUMBERS
Joe Marshall
January 14, 1974
Statistician Bud Goode uses one of the world's biggest computers to pick football games. In 1973 the IBM 360-91 had the Dolphins by seven in the Super Bowl. This year it likes them by nine
Decrease font Decrease font
Enlarge font Enlarge font
January 14, 1974

Doing It By The Numbers

Statistician Bud Goode uses one of the world's biggest computers to pick football games. In 1973 the IBM 360-91 had the Dolphins by seven in the Super Bowl. This year it likes them by nine

View CoverRead All Articles View This Issue

LEGEND:

MIA

MIN

LEFT LEG

36.21

38.43

RUSHES PER GAME

LEFT SHOE

5.00

4.20

YARDS PER RUSH

RIGHT LEG

36.50

32.14

OPPONENT RUSHES

RIGHT SHOE

3.90

4.40

0PP. YARDS PER RUSH

LEFT ARM

5.88

5.93

YARDS PER PASS

RIGHT ARM

3.53

4.65

0PP. YARDS PER PASS

PATCH-HELMET

1.14

1.21

FUMBLES LOST

STARS-HELMET

0.57

1.07

FUMBLES RECOVERED

LEFT EYE

0.86

0.64

PASS INTERCEPTIONS

RIGHT EYE

1.50

1.50

0PP. INTERCEPTIONS

HELMET

0.93

2.29

QB TACKLED

NOSE

3.21

2.14

TACKLED OPPONENT QB

LEFT TOE

1.79

1 .50

FIELD GOALS PER GAME

RIGHT TOE

1.07

1.50

OPPONENT FIELD GOALS

JERSEY NUMBER

13.60

5.00

TOTAL STRENGTH

Meet Maxim (Bud) Goode, resident of Studio City, state of California, continent of North America, planet Earth—a Hollywood press agent loose in a Twilight Zone of numbers. In defiance of all oddsmakers he says Miami will beat Minnesota in the Super Bowl by nine—count 'em nine—points.

Don't scoff. Goode (it is pronounced goody) correctly predicted the winner in 75% of last year's NFL games. Early in the season he forecast that Cincinnati would make the playoffs, an expectation based on the Bengals upsetting both the Vikings and the Browns, which they did. In the last three years he has picked 17 of the 20 winners in postseason playoffs, and in the last two years he has beaten the line 10 of 13 times in the playoffs.

That degree of accuracy surely suggests supernatural assistance. In fact, the forces of Goode consist of IBM 360-91, one of the world's largest computers, which he calls "Cal." Electronically speaking, Cal says that Miami will beat the spread—seven points—on Sunday in Houston. So put your money on the Dolphins, and while you're at it place a bet for Goode, who prefers to kick a gift horse in the mouth. He is, alas, philosophically opposed to gambling.

In making its prediction the computer points to an unexpected Viking weakness, the defensive line. Goode is not surprised. Last year only two teams got to the quarterback fewer times than Minnesota and Goode was saying, "Color the Purple People Eaters puce." This year the Vikes improved their sacks by 43% but only rose to 16th. Miami, on the other hand, tied for second in sacks. Color the Dolphins royal aqua.

Perhaps Minnesota's lowly ranking in sacks reflects nothing more than Coach Bud Grant's preseason intention to concentrate on stopping the run. But if that was their aim, the Vikings failed miserably, yielding 4.4 yards per rush. Only Chicago, Philadelphia and New England gave up more. Miami was tied for seventh in this category, yielding less than four yards an attempt.

To make matters worse for Minnesota, Miami was second in the league in 1973 with an average of five yards per rush. " Larry Csonka provides the power and Mercury Morris has more moves than a water bed," says Goode, citing Csonka's third straight 1,000-yard season and Morris' gaudy, league-leading 6.4 yards a carry. Furthermore, the line that opened holes for those two allowed fewer sacks than any other in the NFL, while the Vikings' front gave only mediocre protection to Fran Tarkenton.

In passing, however, the Dolphins and Vikings are almost identical. However, Minnesota's John Gilliam, who has averaged over 21 yards a reception the past two years, statistically upstages Miami's Paul Warfield.

The computer does reveal subtle indications that Minnesota's defensive line weaknesses are misleading. For instance, although the Vikings allowed fewer points than any other team in the league except, of course, Miami, they gave up quite a few field goals. This suggests that they grew increasingly stingy as the goal line neared. Given Garo Yepremian's range and accuracy. Minnesota will have to get stubborner sooner. The Vikings did hold Dallas to 3.6 yards a carry in the NFC Championship game, but the loss of Calvin Hill, the conference's second leading rusher, may well have contributed to that.

And, after all, can anyone stop Miami? One of the most startling facts the computer showed is that Miami ranked dead last in total offensive plays yet led the NFL in average gain per play and tied for fifth in scoring. Apparently the only thing that can put a halt to the Dolphin offense is the end zone.

In order to predict the outcome of Super Bowl VIII Bud Goode used two programs from something called the Biomedical Statistical Package, a set of computer programs developed at UCLA for the application of statistical methods in medical research. With the aid of the computer he created a correlation matrix, a table that essentially showed the relationships between all football statistics. Hidden in that matrix was a pattern of statistical factors, such as rushing, passing, interceptions, field-goal ability and speed, which account for the outcome of any game. To determine these factors Goode had to do—get ready now—a factor analysis, a program designed to reveal the hidden pattern of primary factors in a body of data. Each factor is totally independent so that strength in one does not mean strength in any other. Using the one or two statistics that related most strongly with each factor, Goode came up with a list of 14 important statistics. These he put into a second program, a multiple-regression analysis, which assigned a weight to each statistic for the purpose of maximizing predictive accuracy. A 100% correlation with points was not possible but. Goode says, the 14 statistics account for 96% of scoring, which means he can come close to predicting or—more accurately—explaining the outcome of any season. Last year, for instance, he missed the total points the Redskins allowed for the whole season by one and the points they scored by 13. The ability to explain statistically the outcome of the regular season makes Goode's computer a sound tool to predict the outcome of the season's culminating event. Last year the computer picked the Super Bowl right on the nose—Miami by seven—although because of a calculation mistake Goode released the figure as eight.

Continue Story
1 2 3 4