Monday night’s game, which went as expected, leaves us with a 10-4 week, and finally gets us back to 60% on the season, 42-28.    It results in only tiny adjustments to the rankings, which are now as follows:

             I was thinking about how to express these rankings as winning percentages, or expected winning percentages.   Here’s how I think it can be done. 

            First, we have to express the rating as a points for/points against ratio.   An average NFL team scores 22.2155 points per game (this year), so let’s use that as a starting point.  

            In football, unlike in baseball, basketball and English literature, the defense scores points, and the offense allows points.   Occasionally those points are scored directly by the defense or against the offense, but even when they don’t score them directly, the defense gets turnovers that lead to field position that lead to points; there’s a pretty close relationship.   It is harder to say for sure what is “defense” and what is “offense” in football than in the other major sports. 

            But if a team scores at 110.00 in this system, that means they are ten points better than an average team.   Let’s assume that’s equally spread between offense and defense.   We take the 22.2155 starting point, add 5.00 for the offense, subtract 5.00 for the defense, and that makes a ratio of 27.2 to 17.2.    Tennessee scores at 110.2; that works out to 27.3 versus 17.1.

            In fact, the average score of a Tennessee game this year is 25.7 to 12.4, but. . .more on that in a moment.   Using the numbers 27.3 and 17.1. . ..

            Daryl Morey is now the GM of the Houston Rockets, but years ago, when he was working at STATS, Inc., he calculated that the “Pythagorean Exponent” in football was 2.37.  If we assume that a team scores 27.4 points per game and allows 17.0 and that the Pythagorean exponent is 2.37, that makes an expected winning percentage of .752.    These are the expected winning percentages for the 32 NFL teams this year:

             Tennessee, the best team in football so far this year, would figure to go about 12-4 against an average schedule, whereas Kansas City would figure to finish about 3-13. 

            Tennessee, however, is 7-0; we’re not really figuring they are going to go 5-4 the rest of the way.   Cincinnati is 0-8; it seems a little unlikely that they will go 4-4 the rest of the way to climb up to their .255 expected winning percentage (4-12). 

            So far our process doesn’t deal with three things:  Strength of Schedule, Scoring Volume, and Close Game performance.  Let’s take those on one at a time:

Strength of Schedule

            The Strength of Schedule analysis here was a real eye-opener.  I had always assumed that, because of the NFL’s parity scheduling program, the teams that were the best last year would in fact play the toughest schedules this year.   Maybe they usually do, I don’t know.

            But you know who has played the softest schedule in the NFL this year?   The weakest schedule of any NFL team this year has been for the New England Patriots—and the third-weakest is for the Giants!  The Super Bowl teams from last year, far from having to confront powerhouse schedules, have had the road paved for them.

            This is the “average strength of opponent” for all NFL teams so far this year:

             First of all, it appears that the “parity scheduling program”, at least based on this analysis of the games this year, doesn’t really do a whole lot.   The standard deviation of schedule strength is 2.4.  I would suspect that, if you just scheduled games at random, it would be about the same.

            And second, it doesn’t really appear that, at least this season, the parity scheduling program has accomplished its purpose.

            Well, not to overstate the case. . .. the goal of the program is that, in the NFL, the good teams will play the good teams more often, and the weaker teams will play weaker schedules.  That has happened, in general.   If you look at the eight teams that have played the most difficult schedules this year, eight of them are good teams, defining “good” as “a ranking above 100.000”.    If you look at the eleven teams that have played the weakest schedules, seven of them are bad teams (below 100.000), and four of those are really bad—Kansas City, San Francisco, Seattle and Oakland.  It is true, in general, that the good teams have played good schedules.   It’s just that the program has failed spectacularly with regard to the two Super Bowl teams from last year, which are over-performing this year in large part because they’ve had cream puff schedules.  The Giants have played three good teams—Pittsburgh, Washington and Cleveland—but they also have played four really awful teams (St. Louis, Seattle, Cincinnati and San Francisco.) 

            Perhaps this is already well known. . I don’t know.  I don’t pay that much attention to the NFL analytical chatter; I’m just trying to catch up here.   Anyway, back to Tennessee.   Tennessee is 7-0, whereas we would expect a team of their caliber to be 5-2 at this time, because (1) they have played a weak schedule, three points worse than average on average.   They’re 10.2 points better than average (110.2), but they’re 12.9 points better than their opponents, which increases their expected winning percentage from .752 to .805. 

            Not running down the whole list, but this also increases the expected winning percentage of New England from .404 to .561, while decreasing the expected winning percentage of the Colts from .564 to .462.   I would assume that these extreme numbers will moderate over the second half of the season, but we’ll check it out. 

Scoring Volume

            I have been assuming that every team is operating in an atmosphere of 44-45 points a game (22.2 for each team), but in fact of course this is untrue; some teams play high-scoring games, other teams not so much.    If your team is ten points a game better than their opponents, that’s a much bigger deal if they’re playing in a context of 34 points a game than if they’re playing in a context of 55. 

            Tennessee has scored 25.7 points per game and allowed 12.4, but they’ve played one more home game than road game; if you adjust for that these numbers become 25.6   and 12.6.   The general point is that they’re not playing in a context of 44 points a game, like an average NFL team, but in a context of 38 points a game.  This makes their 13-point edge over their opponents bigger than it appeared before.   Adjusting for this (2) increases their expected winning percentage from .804 to .844.

Home/Road Differential

            Further adjusting that because the Titans have played one more home game than road game, this increases their expected winning percentage from .844 to .849.

Close Game Performance

          So we would expect the Titans, based on their margin of superiority, their close-to-the vest style and the advantage of having an extra home game, to be 6-1 at this point.   They’ve had a couple of close games and won them both, so they’re 7-0. 

            Looking ahead, the Titans schedule appears to be slightly tougher. . an average schedule strength for remaining games of 100.9, plus an adjustment for the extra road game making it 101.0. 

            New England’s remaining schedule, on the other hand, remains fairly week.   The average rank of their nine remaining opponents is 99.3. 

Predictions for Week Nine

            Jets at Buffalo                                       Buffalo by 7

            Detroit at Chicago                                Chicago by 17

            Jacksonville at Cincinnati                       Jags by 7

            Baltimore at Cleveland                         Cleveland by 1

            Tampa Bay at KC                                Tampa Bay by 15

            Houston at Minnesota                           Minnesota by 7

            Arizona at St. Louis                              Arizona by 11

            Green Bay at Tennessee                       Tennessee by 9

            Miami at Denver                                   Miami by 2

            Dallas at the Giants                               Giants by 7

            Atlanta at Oakland                                Atlanta by 11

            Philadelphia at Seattle                           Philly by 11

            New England at Indianapolis                 Colts by 9

            Our predictions are consistent with the current Las Vegas lines for every game except Miami at Denver.   And honestly, if I was just picking, I’d pick Denver in that one, too, figuring that Denver probably isn’t as bad as the way they have played the last couple of weeks, and that the home field advantage may be larger than we are allowing for.