1)  Ranting about Monday Night 

            I enjoy Tony Kornheiser very much, but did anybody but me think it was weird for him to assume there was some sort of moral imperative for the Bears fans to boo the home team because they were behind at half-time?    Since when is booing a social obligation? Actually, I like Jaworski, too, although I was on his case last week.   One thing I like about him is that he’s an ex-athlete, of course, but he sort of looks like an aging accountant who couldn’t get a date until grad school.   He wears unfashionable glasses and tucks his head into his coat as if afraid somebody was going to slap him at any moment.   It’s really kind of charming, in contrast with all the big, none-too-bright ex-jocks with their Armani suits and $200 haircuts, smiling broadly and trying urgently to convey the message that “I’ve still got it.”  

2)  Predictions from last week

            Not too good.   We were 9-7 on the week, making us 124-69-1 on the season.   Actually, I think we beat most of the experts, most of whom I think were 8-8.    Not many people called Cincinnati over Cleveland, Houston losing in Oakland, Seattle over the Jets or Washington upsetting Philadelphia.   Quite a few people did pick Atlanta over Minnesota and San Diego over Tampa Bay, both of which we missed, but then we were right on Baltimore over Dallas and Tennessee over Pittsburgh, which most people missed.  In last week’s summary we named Pittsburgh as the best team in the league, the most consistent team in the league and the hottest team in the league, but picked Tennessee to beat them because of the Home Field advantage.  

            The consequence of the upsets was that there was a lot of shuffling in our rankings—more than there had been in several weeks.   A lot of teams had their best or worst weeks of the season.  

3)  Predictions for this week

            The home teams should dominate this weekend.

             There are three games this week that come in as ties, which I think is two more than we have had the rest of the year.   By the work process, Redskins/49ers comes in at 16-16, Bears/Texans at 25-25 and Giants/Vikings at 23-23.   I moved the 49ers ahead of the Redskins on the theory that

            1)  the 49ers are technically ahead (by .04), and

            2)  the Home Field advantage at this time of year, particularly after a cross-country flight, could be larger than we are allowing for.

            I moved the Bears ahead of the Texans on the theory that the Bears have more to play for, and, while I was at it, changed the score from 26-25 to 27-23 to make it look more like a football score.

            I gave the Vikings a point over the Giants, also on the theory that the Vikings have more to play for. 

4)  Updated Rankings

            Tennessee takes back over as our best team in the league, not because they beat Pittsburgh, but because they beat Pittsburgh decisively. 

             This is Tennessee’s seventh week atop the rankings.  We did our first rankings of the season after Week Three, when some teams had played only two games. ..obviously not enough to do a solid ranking.   At that time we showed Dallas as the best team in football.   Since then:

            Week Three     Dallas

            Week Four       Dallas

            Week Five       Tampa Bay

            Week Six         Tennessee

            Week Seven     Tennessee

            Week Eight      Tennessee

            Week Nine       Pittsburgh

            Week Ten        Tennessee

            Week Eleven    Tennessee

            Week Twelve   Tennessee

            Thirteen            Giants

            Fourteen           Giants

            Fifteen              Giants

            Sixteen             Steelers

            Seventeen         Tennessee

5)  Team Scoring and Allowing Tendencies

            OK, it is embarrassing to admit this, but I have been making a very basic mistake in how I calculated these things.   I wrote a formula wrong and copied the formula hundreds of times and never put enough time into double-checking what I was doing.   Because of this I wasn’t doing what I thought I was doing, or what I was intending to do.  

            The Scoring and Allowing Tendency calculation works this way.   We start with the assumption that every team’s scoring and allowing tendency is some arbitrary number. …let’s say 5. 

            We then look at the points scored in an actual game.  Carolina at the Giants, last week; Giants 34, Carolina 28.   62 points.  

            From Carolina’s standpoint, we say “there are 62 points in the game; the Giants’ S&A number is 5, therefore Carolina’s must be 12.4 (62 divided by 5).” 

            If we do this for every game of Carolina’s season, we get S&A estimates for Carolina, for the 15 games, of 10.0, 7.4, 6.0, 6.6, 6.8, 6.0, 7.4, 10.2, 4.6, 10.6, 14.6, 13.2, 12.2, 8.0 and 12.4.  This is an average of 9.067.   Therefore, in the second round of calculations, we assume that Carolina’s S&A number is 9.067.

            Of course, in the second round, all of their opponents also have different start values; nobody stays at 5.0.    In the second round, the Giants are assumed to have an S&A number of 9.093.   62 divided 9.0933 is 6.8 (6.8181818).

            The average for all the Carolina games, in the second round, is 4.88951; thus, Carolina has a second-round output of 4.88951.   This becomes the input number for the third round.

            Except it doesn’t; the system tends to go back and forth on us, jumping from 5 to 9 to 5 to 9 through hundreds of rounds.   To prevent this from happening, we take the average of these two figures—4.88951 and 9.0667—to be the input number for the third round.

            That’s what I thought I was doing in these calculations, and, as I acknowledged pretty much every week, it wasn’t working great, except one week it worked well. . .last week, I guess.   But actually, I was making a calculation error.  I divided the points scored in the game not by the OPPONENTS’ previous S&A number, but by the team’s OWN S&A number, thus leaving the opponents out of the cycle entirely.   Because we assume that the starting points are the same for all teams, you get the same results in the first round of calculations one way as the other; I guess that’s why I didn’t spot the error.  Frankly, I don’t know how I was able to come as close to the correct figures as I did.  

            Trailblazer kept trying patiently to explain to me that I was making this mistake, but I could never understand what he was saying.   He was describing a mistake in my ACTUAL process, but I was trying to relate it to what I THOUGHT I was doing, what I intended to do.   Also, I am impatient and easily befuddled.   I didn’t get it.

            Anyway, correcting this mistake, predictably, makes the process work a lot better.   Most of the teams come out with similar numbers, but last week we showed Washington with an S&A number of 5.36, Pittsburgh at 5.32—the two most defense-oriented teams in the league.   This week we have Pittsburgh at 5.60, Washington at 5.04—still the two most defense-oriented teams in the league, but a very different relationship between them.     Pittsburgh’s opponents this year have an average S&A number of 6.46; Washington’s an average of 6.68.   When we factor that out, Washington winds up with a lower figure. 

            These are the current S&A tendencies for the 32 NFL teams:

             The results from last week were typical of the season. . ..pretty good at the extremes, not at all good in the middle.   There were 7 games that we predicted would have scores in the thirties, a total of 247 predicted points in those games.  The actual points in those games was 316.   That’s not close.   For the league, I predicted 634 points.   The actual total was 706.  That’s not close. 

            Frankly, I don’t know that our predictions for point totals are ever going to be great, but perhaps we can do a little better.  

6)  Small Study       

            Defense wins.   Fans love offense and are fascinated by offensive stats, but coaches know is it defense that wins.  

            Is that true?

            No, it isn’t.   The eight most defense-oriented teams in the league (that is, the eight teams least inclined to score or allow points) have an average performance rank of 99.95.  

             There is no obvious relationship between defensive orientation and performance.  

7)  NFL Team Temperatures

             Our Team Temperature system doesn’t work great either, but that’s another story.  I’ll try another method when I get a little time.  These are the current team temperatures for the 32 NFL teams:

8)  Good Week, Bad Week

            Cincinnati, New England, New Orleans and San Diego had their best weeks of the season in Week Sixteen, while Arizona, Cleveland, Detroit and Tampa Bay had their worst.