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.