1)    Predictions from last week

            We were right on three out of four predictions last week, making us 138-75-1 on the season.    

2)  Updated Rankings

            Baltimore, with an impressive win at Miami, has re-emerged as our top team.  Bold Face for the teams that are still alive:

             Basically, the best teams in the NFL are still alive.  Plus Arizona. 

3)   Brief Observation, Redux

             Last week I had an observation about Detroit:

             The Lions are locked in history as the first 0-16 team, but they were really not that bad.     By my math, they’re not actually even the worst team in the league. . .  They played a very tough schedule—39 points better than average, cumulative—and they were outscored by 249 points.   That’s bad, but it’s really not 0-16 bad; it’s more like 3-13 bad.   

             To this a reader responded (slightly edited):

             I disagree about the Lions expected record based on their point differential (-249).  If you look at teams whose differentials averaged about half that, either positive or negative. . .New England, Pittsburgh, Baltimore, Tennessee and the Giants averaged +128 points and averaged 11.5 wins.   Cincinnati, Cleveland, Denver, Oakland, Kansas City and Seattle averaged -121 points and averaged 4.5 wins.   So -249 should be just one win, and probabilistically, one expected win will be winless about 37% of the time.

            I was just commenting based on looking at the scores, but having studied the issue more carefully, I stand by my statement.   I think the there are three problems with the reader’s calculations.

            1.   An average NFL team this year scored and allowed 705 points.  The Lions scored and allowed 785.   249 points have less expected negative impact in a context of 785 points than in a context of 705, and the difference is significant. 

            Putting it in a baseball context, a baseball team that was outscored 477-228 would probably have a winning percentage about .186—but a baseball team that was outscored 517-268 would probably have a winning percentage about .212.

            2.  The reader assumes that one can make a straight-line inference of points to games—that is to say, that the point differential between 1.0 wins and 8.0 wins should be the same as the point differential between 4.5 wins and 11.5 wins.   In fact, of course, it is NOT a straight-line relationship.  The lines bend as you approach zero.

            In baseball, a team that outscored its opponents 938 to 600 should probably win about 115 games (115-47); a team that was outscored 600 to 938 should probably lose 115, so these 676 runs (338, twice) result in a swing of 68 wins (from 115 to 47). 

            But if you apply the same relationship from the center toward the margin, it doesn’t work.   Assuming the center is 769 runs (769-769. . .the same run total as 938-600), a team that was outscored by 676 runs would have an assumed run ratio of 431-1107.    By the reader’s logic this should produce 13 wins (81 minus 68).   In reality, it would be more likely to produce 21 to 22 wins.

            3.   The reader has made no adjustment for the strength of schedule that was specifically sited as one of my reasons.   I had calculated (based on regular season games) that the Lions schedule was 39 points tougher than average.   With the games of this weekend that drops to 36 points, but still, the Lions were more like 213 points below average than 249.  

In the NFL the “Pythagorean Exponent” is, I think, 2.37. ..that’s from memory, don’t shoot me if it’s wrong.   A football team which is outscored 517-268 would have an expected winning percentage about .174—that is, an expectation of two to three wins.   But a team that was outscored 481-268 would have an expected winning percentage of .200—that is, an expectation of three to four wins (3.2).

So, unless I’m missing something, my intuitive statement that the Lions were more like a 3-13 team then an 0-16 team actually appears to have been slightly conservative.  

4)  Research—Effects of the Bye Week

            How much does a team gain from having a bye week?

            We have to know that for this week, of course, because all of the home teams this weekend have also had a week off.  How large is the Bye Week advantage?

            I looked at that issue in this way.  The Pittsburgh Steelers this year had an overall ranking of 110.1.  The week BEFORE they had a bye week, they beat Jacksonville 26-21 for a game output score of 108.0.   The week AFTER their bye week, they beat Cincinnati 38-10 for a game output score of 117.1.   Thus, the Steelers were 7.0 points better after their bye week than overall, and 9.1 points better than in their previous game.

            This chart gives the same data for all 32 NFL teams:

             Twenty of the 32 NFL teams did have better-than-average games after their bye week, but the average gains were very small—less than half of a point.

            I am uncertain what exactly to make of this data.   It appears, based on this very limited study, that the effect of the bye week is essentially to restore teams to normal—to restore them to their “full health” level of ability.

            I looked at the issue of whether teams derived a larger “boost” from a bye week later in the season, but the opposite was true; in fact, the teams which had bye weeks early in the season received almost all of the advantage, and the teams that had bye weeks late didn’t seem to get much out of it.  

            There is a wide array of possible way to interpret this data, but I decided to interpret in this way:  the home field advantage in the NFL this year, including the playoff games and throwing out the game in London, has been 2.4 points, perhaps larger late in the season.  I decided to make the home field advantage for this weekend 4 points, since all of the home teams this week had bye weeks last week.  A small but meaningful advantage.  

5)  Arizona at Carolina 

            Carolina, 33-21.

            Should be the highest-scoring game of the weekend.

6)  San Diego at Pittsburgh

             San Diego is not a .500 team; they’re a 10-6 team that lost too many close games.   Their overall rank is 104.6—five points above average—and they are coming off a string of games that score at 113.4, 113.4 and 106.9.  

            I think San Diego is a legitimate playoff team, but Pittsburgh is a better team, at home, coming of a bye week, and San Diego is flying across the country West to East.   I don’t expect it to be real close.  

             Pittsburgh 26, San Diego 16

7)  Philadelphia at New York (the Giants)

             The Giants haven’t played especially well for several weeks and have slumped in the rankings to where they are now only 0.1 points ahead of Philadelphia.   Some readers have tried to downplay this by pointing out that the Giants clinched their division early and have been coasting, but there is another interpretation:  the Giants really haven’t played well since Plaxico Burress shot himself in the butt.   And Philadelphia to New Jersey, you know. . ..it ain’t exactly a trip to the Orient.    

            I’m picking the Giants, but I think it’s a close game:

            New York 27, Philadelphia 21

8)  Baltimore at Tennessee

             And this game is almost too close to call.  As I’ve mentioned before, I’ve been rooting for Baltimore—but I also have been rooting for Tennessee.  It seems like all the other top teams this year have taken turns at being everybody’s pick for the Super Bowl—the Cowboys were early, the Giants have been, the Steelers, Indianapolis.  I rooted for Tennessee because I never thought they got the respect they deserved.

            But Baltimore, in my view, is probably the best team in football right now.   If Baltimore was at home, if Baltimore was coming off the bye week, I’d be picking the Ravens by a touchdown or more.   With Tennessee at home and rested, I have to pick them to win—but I see the game basically as a tossup, and most likely as a defensive struggle.

            Tennessee 18, Baltimore 17