Sunday, February 01, 2009

No Computer Was Ever Hung in Effigy

There is a small but vigorous branch of the scientific community dedicated to the proposition that pro football coaches ought to get fired more often.

About five years ago, an economist wrote a study stating that teams would be better off going for it on fourth and goal and fourth and short inside the red zone instead of kicking field goals. The math was impressive. The psychology, not so much.

Yours truly ran that one past Bill Belichick at a press conference. Belichick, of course, had seen the professor's work, and had enough respect for science not to sneer. He did, however, note that three birds (points) in the hand were worth six in the bush of probabilities, and that the economist may not have factored in the morale elements of a successful goal-line stand.

NFL Films picked up on the article shortly thereafter, and the other coaches they interviewed did not delve into football theory. They had a personal take on the consequences of coming up on the wrong side of the egghead's equations, to wit, a coach who takes points off the scoreboard and loses a game moves a big step closer to losing his job.

The prof's theory remains just that. Teams take the points unless they have NO alternative, and will be doing so long after we're all watching games on the Celestial HDTV. But in their own way, scientists are as persistent as coaches. Today's "New York Times" contains their next salvo in their war on the fuddy-duddys wearing headsets and ugly Reebok apparel.

An astrophysicist and a backgammon champion combined their mathematical skills to create a computer program modestly called Zeus to evaluate game decisions made by NFL coaches in the 2008 season. The program's conclusion: Coaches are wimps, gutless wonders who fail to see how the biggest gambles in the game are actually the safest bets they can make.

Their very first conclusion was that coaches should almost always go for it on 4th and 1. I'm glad they got that brainstorm published in the Times, because that means Tom Coughlin probably read it.

Coughlin went for it on 4th and 1 twice against the Eagles in the fourth quarter of their playoff game last month. The Giants were stuffed both times, and lost. The exigencies of time management and being down ten points made going for it an easy call, but the fact remains that after the second 4th and 1 failure, New York was down 13 points.

This shows us what Zeus cannot. The immediate downside risk of failure on 4th down is so dire, coaches almost always prefer the 35-40 yard shift in field position of a punt. Even Belichick, who goes for it on 4th and short more than most coaches, makes sure he places that bet when the Pats are in opposition territory, that is, when he is playing with house money.

Zeus's second conclusion "not all points are created equal" is both true and false. Touchdowns are far preferable to field goals. But a goose egg in the red zone is a catastrophe unless one is giving the opponent the ball on about their own six-inch line. A successful goal line stand is one of the most exhilarating experiences a football team can have. It makes them feel invincible. This is not a mindset the opposing coach wants to encourage.

Zeus bright idea three is to use the onside kick more often. This is hilarilously foolish. Mathematics does not account for the element of surprise, but onside kicks do. The coach who used this idea in a game likely would have success with it. The next game? No.

I will give the scientist and the gambler credit for intellectual honesty. They published a chart of their findings for each team, and made a point of drawing attention to its most obvious factor. The percentage of "correct" decisions made by coaches according to their formula had no relationship to the success of the teams involved in terms of wins and losses.

The six teams who had the highest percentage of game decisions rated as the right move by Zeus were, in order, the Jaguars, Texans, Rams, Bengals, Cowboys and Chiefs. Combined 2008 record: 30-65-1

The six pusillanimous teams who ignored the computer's sage advice and most frequently did mathematically ill-advised things like punting were, from 27th to 32nd were, in order, the Titans, Dolphins, Broncos, Falcons, Eagles and Panthers. Combined 2008 record: 64-31-1.

The average fan, not to mention the average coach, might leap to the conclusion that doing the OPPOSITE of what Zeus recommends is the best way to formulate in-game decisions. But science has the answer. It's insight? "A coach's ability to make decision is independent of player talent."

No it isn't. A coach's decisions are, or ought to be, ENTIRELY dependent on player talent. What can my guys do best and worst? What can their guys do? As Bill Parcells has said many times, "the coach's job is to put his players in the best position to win the game." To do that, he must start with the players themselves, not formulas. Belichick's special genius is his abhorrence of all formulas, recognizing that in a game of violent chaos, the requirements of victory shift week to week, quarter to quarter, and play to play.

Logic, if not mathematical logic, suggests that the teams who took the most computer-recommended gambles did so because their coaches knew they weren't good enough to win without gambling, and that the conservative teams played that way because their coaches knew that all things being equal, their teams were more likely to win, and that the point of game management was not to give away any inequalities or big breaks.

Fans and sportswriters have been urging football coaches to take more chances since the legalization of the forward pass, and coaches have just as stubbornly refused to take their advice. Now machines have joined the chorus. Science is no more likely to change coaches' minds than were generations of boos, and according to the evidence, it shouldn't.

Here's what baffles me. The world's entire economy is on the skids because a bunch of mathematically gifted people came up with the bright idea that financial risk could be quantified, and that the quantification indicated investors ought to take on much, much more risk than in the past. Now we're supposed to take that idea into football?

Football coaches as a class are among the most risk-averse managers in our society. You'd think we'd be honoring them for it.


Post a Comment

<< Home