February 15, 2007
Monday-morning quarterbacking is a science after all
by Mike Leonard
If the Indianapolis Colts weren't the best team in the National Football League in the 2006-07 season, few people would argue that they weren't at least among the top few.
Indiana University scientist Chuck Bower thinks they're mighty good. "Indianapolis has one of the most efficient offenses in the league — and maybe in the history of football," the astrophysics researcher said this week.
Plug the Colts' season-long playcalling into the Zeus football analysis program, however, and you'll see Indianapolis in 25th place out of 32 NFL teams.
That doesn't surprise Bower, who spent 5 1/2 years developing Zeus with partners Frank Frigo, of Louisville, and Bo Durickovic, of Washington, D.C.
Immediately after Indianapolis won the Super Bowl, Zeus showed that the Colts' conservative play-calling kept the Chicago Bears in the game until the end, despite the Colts' almost 2-1 advantage in yards gained, first downs and time of possession.
Zeus flagged the Colts for six critical call errors in the Super Bowl. The worst was when the Colts led 19-14 with the ball on the one-yard-line on fourth down. Bower contended that "even the worst offensive team in the NFL is a favorite to score a TD" in that situation. The Colts kicked a field goal instead.
Bower laughed at and agreed with the suggestion that football coaches are almost always too conservative. "I think it's part of the culture," he said. In the example of the Colts, it means that when Indianapolis plays conservatively, it doesn't take advantage of its greatest strength, its formidable offensive process.
"They're amazing," Bower said. "At crunch time, when they're not playing conservatively, they almost always march right down the field. No one can stop them."
That kind of analysis is part of the reason why Bower and his collaborators developed Zeus. Using the massive amount of statistics gathered annually by the NFL and data on teams and individual players, the computer program comes up with statistical probabilities that are undeniably accurate.
A Zeus recommendation isn't a guarantee that any particular play will definitely work. Nor could it replace the knowledge and experience a coach and his staff bring to the game. It is, rather, another tool and another measurement for coaches to consider.
"We believe that, on average, using Zeus is worth about one win over the course of a season," Bower said.
Zeus can even be used to put a "player position value" on an individual, which could help teams make critical personnel decisions when they're trying to stay under the NFL's salary cap.
The problem facing Bower and his partners in the Louisville-based company, Endgame Technologies LLC, is that so far, no one has purchased Zeus.
"We've met with nine teams and talked to many more," he said. "They almost always say they're interested but they don't know what they'd do with it."
The program will run on a decent laptop computer, so utility isn't an issue.
It is pricey, Bower acknowledged, saying only that the cost is in the "high six figures."
"That's really just the annual salary of an average player," he said. "Gaining one win a season is worth a lot more than that."
It might just be a matter of time before a team, or a lot of teams, invest in the computer program. ESPN.com has published two recent articles about results Zeus has kicked out. The latest, published this week, observes that the Dallas Cowboys may have lost more than a coach when Bill Parcells resigned from his head coaching position. Zeus showed the Cowboys to be the smartest play-calling team in the NFL last year.
In December, Esquire magazine pointed out the brilliance of the Zeus software and asked Bower and his partners to highlight the three most surprising conclusions that the deep statistical analysis showed.
• It's almost always a better decision to try for a first down on fourth and short yardage than to punt. An obvious exception is when your team is in very poor field position.
• Onside kicks should be used more frequently. A conservative figure of 25 percent recovery rate by the kicking team suggests that this tactic is used less than half as often as it could be.
• In kicking, strength trumps accuracy. Kickoff distance is more valuable to a team than marginally better field goal accuracy.