Is there anything WRONG with the N.F.L's system?? Well,I
guess you could say not,depending on your biases ;-) - for it IS biased.
Let's look at what biases there are in it....
The original rationale behind the creation of the system
was that a "perfect" passer would get a rating of 100. In fact,the
theoretical maximum is actually 158.3,and several Quarter-backs of
recent years have surpassed 100,but the original intent was that 100
would represent the peak of achievement (in contrast to the "average
standard" of 66.6).
What figures - spread evenly across the 4 categories - would
give a rating of 100?
They are 60% completions,9 yards/attempt,7.5% TD's,and 3.5%
Int's. Each of these would contribute 25 points,leading to a total of
100.
Of course,you don't have to have these EXACT numbers to get
a rating of 100,and this is where the biases start slipping in.
In 1992,the average figures were 57.5%,6.86,3.85%
touchdowns,and 3.87% Int's. These figures translate to a rating of 75.3.
The thing that stands out immediately,is not just
that these figures are smaller than the "ideal",but how MUCH smaller they
are in some cases. i.e. the average figure for completions is 74% of that
used in the formula,whereas the average for T.D's is only 32% of the formula
figure.
This clearly shows that some figures are more achieveable than
others - hence,each category isn't WORTH an equal amount (even though that
would be the apparent intention,since all explanations of the system revolve
around "to get 25 points in THIS category.....").
To see how it's biased,let's consider an "improved"
Quarter-back. Let's say that in 1992 his completion percentage was "average".
i.e. 57.5%. Let's also say that he then improved by 10% in 1993,to 63.25%
(1.10*57.5=63.25). What happens to his N.F.L. rating?
It turns out that his N.F.L. rating would improve from 75.3
to 80.1. This is an increase of nearly 6.4%. This is the first hint of
bias. If there was no bias (towards completion percentage in this case),you
would expect an increase in his rating of 10% (or 2.5% - 1/4 of 10% - since
there are 4 categories).
Further bias is evident once you carry out the same scenario
on OTHER categories. A 10% improvement in his yards/attempt,gives only a 1.9%
increase in his rating. A 10% improvement in touchdown percentage gives a
2.1% increase in his rating,and similarly a 3.8% increase to his rating
results from a 10% improvement in his intercept percentage.
This makes the order of importance; completion percentage
(6.4%); intercept percentage (3.8%); touchdown percentage (2.1%); and yards
per attempt (1.9%).
There is further bias in the sense that the formula uses
static figures,and the evolution of the game isn't taken into account. In
particular,the modern game features MUCH better completion and touchdown
percentages (on an individual basis in the case of touchdowns) than
previously.
As evidence of this,of the 20 Quarter-backs who comprised
the top-10's for these 2 categories in 1992,14 of them were active players.
In contrast,of the 20 Quarter-backs who comprised the top-10's in 1992 for
intercept percentage and yards per attempt,only THREE were active players.
As FURTHER evidence of this evolution,consider the
following list of average N.F.L. ratings....
year att. com. yds/com TD int. NFL rating
1932 1044 35.63% 14.25 4.02% 9.39% 27.2
1940 2254 42.95% 14.24 4.44% 9.89% 38.6
1950 4307 46.62% 14.47 5.11% 7.96% 52.9
1960 4114 50.17% 14.42 5.37% 6.66% 64.2
1970 9796 51.12% 13.16 4.36% 5.21% 65.6
1980 13705 56.18% 12.46 4.42% 4.57% 73.7
1990 13516 56.02% 12.51 4.25% 3.55% 77.3
2000 16342 58.11% 11.60 3.88% 3.25% 78.0
This is why the all-time top-20 Passers,as rated by the
N.F.L. system,is so DOMINATED by modern players. In 1992,this list included
Ken O'Brien and Dave Krieg,but not Sammy Baugh or Johnny Unitas. What gives?
Something had to be done! :-)
Next topic
P.P.R. (and N.F.L. Passer rating) online calculator
P.P.R. mailing-list
Back to Pedder Passer Rating main page