Iowa State University

Inside Iowa State
October 9, 1998

Odds on favorite

by Skip Derra

Numbers don't lie, but they can surprise and astonish. Take these numbers: Iowa State 27, Iowa 9. Surprising to many, but astonishing to Hal Stern, professor of statistics.

"Twenty-eight point underdogs win roughly 4 percent of the time. So that was a major upset," Stern said, referring to the Cyclones' Sept. 12 football victory over the Hawkeyes. "When you factor in the 18-point victory margin, it was truly shocking."

Stern was not surprised, on the other hand, that St. Louis Cardinal Mark McGwire broke Roger Maris' record of 61 home runs in a season. McGwire, he noted, had been averaging a homer in every nine at bats for the past four years (the highest average ever). It was only a matter of putting together an injury-free season and the record would fall.

But what was unexpected was that Chicago Cub Sammy Sosa (one homer every 15 at bats) also broke the Maris mark this year.

"I'm not sure anyone could've predicted Sosa's home run total," said Stern, who uses his expertise in statistics to interpret and predict various sporting and gaming events, as both a statistician and a fan.

"What I like about all sports is the uncertainty in the event," Stern said. "Baseball is full of examples of this, and it has a long history of keeping track of records, numbers and averages. It's a wonderful place to play."

Stern discovered a love of numbers during his formative years in Bayside, N.Y., and his involvement in sports.

"I was the kid who kept the batting averages for the Little League team," he said of his sports statistics apprenticeship.

Throughout his career, Stern's love of sports fueled his passion for numbers. For example, as a doctoral student at Stanford, he sought out and studied under Tom Cover, a statistics and electrical engineering professor with an interest in gambling. Cover showed Stern how to apply statistics to the games people play.

When he teaches statistics or carries out research, Stern is quick to make connections between statistical models that can be used in sports and other areas to which they might apply. For example, the same model used to rate football teams can be used in the selective breeding of animals or in studying disease incidence rates. As another example, the normal distribution used in models of weight, agricultural yields or rainfall also is relevant in studying the probability of an upset in football.

This dual use of statistical models led one of Stern's former students to develop a rating system for chess players and another to pin down a statistically sound strategy for throwing darts (aim for triple 19).

"I don't want a student to do a Ph.D. in sports and statistics," Stern said. "I want him or her to do a Ph.D. in statistics that develops methods that can be applied to a large number of areas. If it turns out it can be applied to sports, then so much the better."

For two years, Stern had a regular column (he now is the editor) in Chance magazine, a publication of the main organization of professional statisticians. The quarterly column, "A Statistician Reads the Sports Page," detailed the hidden statistics in several different games. Stern showed, for example, why you shouldn't kick a field goal when it's fourth down and short; why there are no streaks (statistically verifiable at least) in sports; how the home team has an advantage in a best of seven baseball series; and how rare it is when a 16-seed beats a 1-seed in a basketball tournament (see "pointers" below).

While his love of numbers and sports go together like tailgating and football games, Stern is quick to point out the difference between professional statistics and recreational statistics.

"Sports are full of small 's' statistics," he said. "These are the field statistics that track a player's performance. The field of statistics, or the big 'S' statistics, is about drawing conclusions from data."

Last year Stern and Mark Glickman of Boston University developed a statistical model for determining National Football League wins. The model, which accounted for changes in a team's strength during the season (injuries, team psychology and different players), outperformed the experts who set the Las Vegas point spreads.

"The question for me is how well we can predict," he said, "not because I want to become a professional gambler. But to really understand something means you can make predictions about it."

On deck for Stern is a book on sports and statistics. But he wants this book to become more than a hardcover trivia companion for couch potatoes.

"I want to take sports nuts and subliminally teach them about statistics," Stern said. "I want to teach them how to be more intelligent consumers of statistics."

And with good reason. We all are inundated with statistics, Stern said. One week you might read a story about a substance that causes cancer and the next week that substance is found to be good for your heart.

"How do we filter it all out?" he asked. "I don't think we can create 100 million adults who are capable of doing statistical analyses. But I think we can help them make good choices."

A Few (Statistically-Valid) Pointers

Probability, according to Hal Stern, may seem counter- intuitive to people at times. Yet in every game played, probability and statistics are used by coaches, managers, odds makers and fantasy league players to make sound choices. Here are some statistically valid insights into several games:

Go for it

If it's fourth down and short and you must decide whether to kick a field goal or go for the first down, Stern says, "Go for it!" The odds are greater (65 to 70 percent success) in trying for a first down than in kicking a 40-yard field goal (50 percent). You also have the possibility of scoring six rather than three points.

Aim for triple 19

When throwing darts, the fastest way to accumulate points is to avoid the bullseye (50 points) and go for the triple 19s, unless, of course, you are a good dart thrower. Then try for triple 20. But the 19 is bordered by a 3 and 7. The 20 is bordered by 1 and 5. You can make up points on the darts that don't hit the 19 to compensate for the 3-point difference between triple 20 and triple 19.

On any given day

The uncertainty of sports makes it possible on any given day for team A to beat team B, regardless of ability. But in the NCAA basketball tourney, the odds of a 1-seed losing to a 16-seed is 100 to 1. Stern said thus far, only 52 1-seed vs. 16-seed games have been played in the men's tournament, which explains why there haven't been any mega-upsets. But it's just a matter of time.

Baseball's Holy Grail

With a new mark in baseball for home runs, many people will turn their attention to what could be the unreachable statistic in baseball: Joe DiMaggio's 56-game hitting streak. Stern said that this "streak" is extremely rare, making it just about unreachable in today's highly skilled game. "The DiMaggio streak is sufficiently unusual that it shouldn't have happened yet in baseball," he said.

What streak?

Speaking of streaks, Stern said -- but doesn't necessarily believe -- they don't exist statistically in sports. Any streak can be explained by ordinary variations in a player's performance. Getting a hot hand in basketball or hitting the ball well in baseball may be more of a "feel" by a player in a groove than a statistically verifiable phenomenon.

Iowa State homepage

Inside Iowa State,, University Relations
Copyright © 1998, Iowa State University, all rights reserved
Revised 10/8/98