Don R. Smith, vice president of the Pro Football Hall of Fame, invented the formula under which Joe Montana is the top-rated passer of all time.
Accordingly, Smith wasn’t surprised by Montana’s big winning performance for the San Francisco 49ers last Sunday against the Raiders.
But as an expert on statistics, he says that sports fans often misunderstand the meaning of the Smith formula, which pro football uses to measure passers each week and at the end of each year.
Question: What is it that the fans misunderstand?
Answer: People don’t always realize that this is a system for grading passers, not quarterbacks. It isn’t a quarterback rating system. Montana could be the best quarterback in football, but statisticians can’t prove that.
Q: What can they prove?
A: Our objective is simply to give a statistical picture of passing. The formula, which the National Football League has used for several years, compares the passing skills of NFL quarterbacks. That is, it compares the most important of these skills, not all of them.
Q: Why not all?
A: Whose fault is it when a pass is dropped? You can’t really measure such things as dropped passes. But you don’t have to. There are several other good ways to compare passers, and all you need is a logical, meaningful system that isn’t too burdensome.
Q: And you have it with this system?
A: We think so. The formula can be used to rate all kinds of passers--former passers, active passers, bomb throwers, play-action types--against one standard.
The standard is based on the passing skills needed in four significant categories.
Q: What are the four?
A: Completion percentage, interception percentage, touchdown percentage, and average gain per attempted pass. We grade each passer in each of these categories. Then we add up his four grades to get the total--his rating.
Q: And when you do this, Montana comes up No. 1?
A: Yes, going into the 1985 season, Montana was first (with a 92.7 rating), followed by Roger Staubach (83.4), Danny White (82.7), Sonny Jurgensen (82.6), Len Dawson (82.6), Ken Anderson (82) and Dan Fouts (81.2). These are the only seven passers with lifetime ratings above 81 based on at least 1,500 career pass attempts.
Q: Some football fans say your rating system may be logical enough for its purpose, but that it’s too complicated. What is your reaction to this complaint?
A: Our formula is as simple as you can make it and still do an effective job. The problem is that the quarterback position is the hardest to measure statistically in sports. Running backs can be graded on total yards gained. Receivers on number of catches. Outfielders on hits and errors. There’s no comparable easy, satisfactory way to measure passers.
Q: There are, however, other ways. Didn’t the NFL use two or three different systems in former years?
A: Yes. During the early days of pro football, passers were graded on one accomplishment only--yards gained. Then, for a while, the criterion was completion percentage. And not too long ago, it was average yards gained per attempted pass.
Q: Any of these three is much simpler than the present formula.
A: Sure, the older, one-category methods were all simpler. But they were also unfair and too incomplete.
Q: Why is a yards-gained method unfair?
A: It’s unfair to the passers on running teams, whose coaches would rather run the ball than throw it.
Q: What’s wrong with a completion-percentage system?
A: Some teams have coaches who like to throw mostly short passes. Their quarterbacks may have a high completion average without necessarily being more effective passers. Measuring them only on percentage of completions is unfair to long passers and medium-range passers, among others.
Q: The average-gain system--comparing average yards made per pass attempt--seems both simple enough and quite meaningful. It takes into account total yards as produced by either long passes or short passes.
A: Yes, but it doesn’t take into account such things as interceptions and touchdowns. The object in football is not just to come down the field with the ball but to get it into the end zone. That’s why the average-gain system is incomplete. When you only measure average gains, a 10-yard touchdown is no more important than 10 yards at midfield. And an interception is no worse than an incomplete.
Q: Some interceptions--some deflections, for example--aren’t the passer’s fault. Why should every interception be counted?
A: There’s no sound way to differentiate between kinds of interceptions. The assumption is that most are probably the passer’s fault. Besides, an interception happens to be a measurable statistic. It’s a performance stat--and stats don’t tell why. They just tell what.
Q: Now, if Montana leads all passers with a 92.7 rating, how far above average is he?
A: A statistically average passer is one with a 66.7 rating.
Q: How does a man get a 66.7?
A: The formula we use is based on performance standards. For example, the standard for completions is 50%. If an NFL quarterback completes half his passes, his completion grade is 1.0. That’s average.
Q: What’s the standard for the three other categories?
A: For touchdown passes, it’s 5% of a quarterback’s pass attempts. If he hits that exactly, his touchdown grade is 1.0. If exactly 5.5% of his attempts are intercepted, his interception grade is 1.0. If his passes gain an average of precisely seven yards per attempt, his average-gain grade is 1.0.
Q: Next, you say, you add a passer’s four grades to get a composite grade.
A: That’s right. And if they add up to 4.0, that’s average. Then to make public understanding easier, we use a conversion table. A 4.0 rating on our computers is converted to a publicly announced rating of 66.7
Q: What total converts to 100?
A: A composite 6.0 grade would convert to a rating of 100.
Q: Can any NFL passer score 100?
A: It can be done for a game or two, or sometimes up to a year, but no one with 1,500 pass attempts has ever done it.
Q: How did Montana hit 92.7?
A: In his first six years in the league, Montana had 2,077 pass attempts. He completed 63.7% for a completion grade of 1.685. He threw 106 touchdowns for a touchdown percentage of 5.1 and a touchdown grade of 1.020. He threw 54 interceptions for a very low interception percentage, 2.6, and an excellent interception grade, 1.725. And his average gain was 7.52 yards for a grade of 1.130. His four grades total 5.560--which converts to a rating of 92.7.
Q: How long does it take to work out a rating system of this kind?
A: This one began evolving in 1960 when Pete Rozelle became commissioner. As the years went by, a lot of people suggested the criteria--touchdowns, interceptions, and so on.
Q: What did you suggest?
A: After I became involved about 15 years ago, we changed the touchdown category from total touchdowns to the present percentage system. This enabled us to work out the formula based on 1.0 for an average grade in each category on up to a maximum of 2.3 and down to zero. The maximum rating is 158.3. The minimum rating--no matter how ineffective you are--is zero. We don’t use minus figures. I thought of the final formula on an airplane one day.
Q: When was that?
A: A dozen years or so ago. We were flying over Kansas.
Q: Some folks say you’re still up in the air.
A: I don’t hear that from 49er fans. They think we’re doing something right with a formula that makes Montana No. 1.