Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 (Tek) 9/28/84 based on 9/17/84; site tekchips.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!vax135!cornell!uw-beaver!tektronix!tekcrl!tekchips!stevev
From: stevev@tekchips.UUCP (Steve Vegdahl)
Newsgroups: net.sport.football
Subject: RE: NFL Quarterback ratings
Message-ID: <341@tekchips.UUCP>
Date: Mon, 4-Nov-85 13:21:41 EST
Article-I.D.: tekchips.341
Posted: Mon Nov  4 13:21:41 1985
Date-Received: Thu, 7-Nov-85 03:45:19 EST
References: <345@drutx.UUCP>
Organization: Tektronix, Beaverton OR
Lines: 72

> 
> > Does anybody out there in netland know how quarterback ratings are determined?
> > I have looked through several books and have been unable to find the OFFICIAL
> > NFL method for calculating these ratings.
> 
> 	The closest method I heard of (to the OFFICIAL method) appeared in
> The Sporting News, in a fall 1980 issue.  The formula is:
> 
> 		(5/6)*(CMP + 5*AVG + 4*TD + 2.5 - 5*INT)

> The rating must also be:    0 <= rating <= 150
>  
> 					
> 							Dave Van Handel

I also got some info from an article that appeared in The Sporting News
several years ago, and incorporated it into a computer program I wrote.
The following is based on the program, as I no longer have the original
article (I think it was fall of '78 or '79).

It is based on 4 factors mentioned above:
   - completion percentage (completions/attempts)
   - average gain per attempt (yardage/attempts)
   - touchdown percentage (touchdowns/attempts)
   - interception percentage (interceptions/attempts)

The formula (which is an extension of the one above) is
   5/6 *
	(2.5
	 + 100 * ((completions/attempts MIN .775) MAX .3)
	 + 5 * ((yardage/attempts MIN 12.5) MAX 3)
	 + 400 * (touchdowns/attempts MIN .119)
	 - 500 * (interceptions/attempts MIN .095))

The reason for the max's and min's may be account range of the NFL's
charts.  In addition, my experience was that it was necessary to round
the everything to the nearest tenth of a percent (or something similar)
in order to get the numbers to be consistent with the "official" ratings.
Again, this account for the NFL's using charts.

This formula was introduced a number of years ago (early seventies?).
Prior to that time, the quarterbacks were ranked according to several
criteria (I believe these same four), and their rankings were summed.
Thus, if Montana were ranked 2, 1, 5 and 3 in the categories respectively,
his "rating" would be 11 (2+1+5+3); obviously, the low score would win.
This had the blatant misfeature that two QB's who were ranked 1 and 2
in 1966, might be ranked 2 and 1 in 1977, even if each had identical
statistics to his previous year; the stats of other QBs would come
into play.

Because of the MIN's and MAX's, the highest possible rating is something
like 157.  I believe that some of the constants were chosen so that
the worst possible rating is exactly zero.  A rating of over 100 indicates
an exceptionally good year.  It is often the case that no QB has a ranking
over 100; I believe that it is not unusual for the highest ranked QB
in a conference to be in the 80's.

		Steve Vegdahl
		Computer Research Lab.
		Tektronix, Inc.
		Beaverton, Oregon

P.S.  If anyone has access to archives of The Sporting News, and are
interested in looking it up, I remember that this information was
in an article that was discussing the rankings of college QBs who, at
that time at least, were ranked only by completion percentage.
Steve Dils (of Stanford) had his picture somewhere near the article,
I believe because he at that time had the highest completion percentage
of major college QBs.  That would likely put it in Dils' senior year
(possibly junior year) which was sometime around 1978.  I'm sure that
it occured during the football season; my guess is around the beginning
of December.