Sunday was the day all the NFL fans went home angry.   The home teams in the NFL in Week Twelve went 5-11, and it wasn’t that all the good teams were on the road; a bunch of NFL teams just lost, at home, to teams that weren’t any better than they were.  San Diego, Denver, Jackson, Miami, Tennessee, Cleveland. . .all lost, at home, to teams that they could reasonably have beaten in a road game.

            Did you ever think about this issue, from a game design standpoint, of what is the perfect home winning percentage?   You would probably want the home team to win more than 50% of the time, from the standpoint of the league, because when people put out the money to actually go to the game, you want them to enjoy the experience, and the fans enjoy the experience more if the home team wins.   On the other hand, you don’t want the home field winning percentage to get too far from .500, because:

            1)  Games become un-interesting when they are predictable, and

            2)  A huge home field advantage would detract from the credibility of the contest.  

            If the home team won 70% of the time, then the home-field advantage--.400—would be much larger than the quality differential.   Games would be determined much more by who was at home than by who was better.   That would start to stink pretty quickly.

1)  Predictions from last week

            We were 9-7 on the Week Twelve, making us 82-47-1 on the season.  

2)  Predictions for this week

 3)  Updated Rankings

            The New York Giants have finally reached the top spot in our rankings. . . .meaning that our system has finally caught up to the rest of the world, which has generally regarded the Giants all season as the best team in the NFL. 

 4)  Team Scoring and Allowing Tendencies

            I have been trying to puzzle out a method to predict the number of points scored in an NFL game.    It is essentially this:  that if Arizona is playing San Francisco and Arizona has a Scoring and Allowing Tendency (S&A) of “8” and San Francisco has a Scoring and Allowing Tendency of “6”, then there should be 48 points scored in the game.

            I have been trying to derive these numbers by using the schedule, and “backing out” the points for each team each week.   In the NFL this year the average team has scored 22.5142 points per game.   This makes 45.0284 points per game for the two teams.   The square root of 45.0284 is 6.7103.    If we assume that each team has a Scoring and Allowing tendency (S&A) of 6.7103, we conclude that the average game should have 45.0284 points.

            Therefore—this is the theory; doesn’t quite work—let’s take the Arizona/New York Jets game early in the season, the score of which was 35-56.   Looking at if from Arizona’s standpoint, if the opposition’s S&A number is 6.7103 and the total points scored in the game is 91, then the S&A number for Arizona would be 91 divided by 6.7103, which is 13.56. 

            We derive in this way the S&A tendencies for each team in each game, and take the average of those numbers.  Arizona’s average is 7.912.   What I wanted to do was to take the 7.912 number, plug it back into the system, and refine it by repeating the process.   This doesn’t work; it just causes everyone to go back to 6.7103.  Unlike the rankings system, the S&A tendencies system allows the teams to move all the way to their targets in the first try, which causes them to move back to point zero in the second repetition of the process.

            What I did next was to use, instead of the average of the S&A tendencies derived from all of the games, that average plus .067103, divided by 1.01.   This has the effect of preventing the system from jumping back and forth over the goalposts, and allows the teams, through many repetitions, to center on a number which is very close to the square root of the points per game scored in that team’s games by the two teams combined.   That is, an average Arizona game this season has had 53.09 points.   This method will give us a figure, for Arizona, which is the square root of 53.09 (7.286. . ..let’s call it 7.3), modified slightly by the extent to which Arizona’s opponents have tended to score and allow a larger-than-average number of points. 

            But 7.3 cannot be the S&A number for Arizona, or, to get to 53.09 points per game, their opponents would also have to have the same S&A number (7.3 * 7.3 = 53.29).   Arizona’s number must be that number which, when multiplied by the S&A tendencies of Arizona’s opponents, produces 53.09.   That number is 7.906.    Thus, Arizona’s S&A number is 7.906. 

            These are the S&A (Scoring and Allowing Tendency) numbers for the 32 NFL teams:

             The highest-scoring game this week figures to be Denver at the New York Jets (57 points), followed by Arizona at Philadelphia (54 points.)   The lowest-scoring game figures to be Baltimore at Cincinnati (35 points), followed by Pittsburgh at New England (36).   If you’re betting the overs and unders you can get what I know about that from the predicted scores up above. 

6)  NFL Team Temperatures

            These, at the moment, are the hottest teams in the NFL:

             And these are the coldest:

             Whether this means anything is debatable.   Green Bay and Tennessee, which were listed among the hot teams last week, nonetheless played their worst games of the season this weekend, thus dropping off the hot list, while Oakland—third-coldest team last week at 41º—went to Denver and put a whuppin’ on ‘em.    But. ..I didn’t claim that it means anything.   We perceive teams, and players, as being “hot” and “cold”.   Whether they actually are is something we can argue about another time.

The Good, the Bad, the Ugly

            Twelve teams—nearly half the league—had either their best game of the year this weekend or their worst.    Baltimore’s 36-7 shellacking of Philadelphia was Baltimore’s best game of the year and Philadelphia’s worst.   Buffalo’s 54-31 win in KC was Buffalo’s best game of the year, at least by our formula.   Minnesota defeating Jacksonville 30-12 in Jacksonville was the Viking’s best game of the year and the Jag’s worst.   New Orleans’ 51-29 romp over Green Bay was the Saints’ best game and the Packers’ worst.    The Jets’ 34-13 win at Tennessee was the Jets’ best game and the Titans’ worst.   Oakland’s 31-10 win in Denver was the Raiders’ best game and the Broncos’ worst.   Miami’s 28-48 loss to New England, at home, was their worst game of the year.  

            In our system, the extent to which one team over-performs is always equal to the extent to which the other team under-performs.   When Oakland beats Denver, we don’t know whether this happened because Oakland played better than usual or Denver played worse than usual, so we assume it is half of one and half of the other—thus, Oakland’s over-performance is equal to Denver’s under-performance.   It’s not a conclusion, it’s an assumption.   It tends to lead to the conclusion that whatever is the best game of the year for one team must have been the worst for the other—not always, but often.   In any case, it was a big week for bests and worsts.   St. Louis and Cleveland also had their worst games since Week One, and Kansas City missed their season-worst output score by a thin margin.

Consistency

            The most consistent team in the NFL this year, in terms of playing at the same level week-in and week-out, has been Minnesota.   Sunday’s game at Jacksonville was only the second game this year the Vikings have won or lost by more than 7 points, the other one being a 20-10 win over Carolina—plus their schedule has been very consistent.   The Vikings have played the toughest schedule in the NFC, but that can best be explained in this way:  that the fact that they haven’t had to play most of the league’s best teams is outweighed by the fact that they also haven’t gotten to play the worst teams.   The worst teams in the NFL are much further below the norm than the best teams are above the norm, so if you don’t play either—and the Vikings basically haven’t played either—you’re coming out behind on the deal.   The Vikings haven’t played the Giants, the Titans, the Steelers, or the Ravens, and they also haven’t played Kansas City, St. Louis, Oakland, San Francisco or Cincinnati.  It’s actually kind of remarkable; they have played almost exclusively the middle-of-the-pack opponents.

            The league’s least consistent teams have been Denver and New England.   These are the standard deviations of Game Output Scores for the 32 NFL teams:

 Strength of Schedule

            The team which has played the most difficult schedule in the NFL, by far, is Cincinnati.    The average score of a Bengals’ game this year is 13.5 to 25.1—a loss by 11.6 points.   7.1 of those 11.6 points are explained by the Bengals not being a very good team—but 4.3 points are explained by the fact that they’ve had to play the Steelers twice, the Giants, Titans, Cowboys, Jets and Eagles.   The other 0.2 points are explained by Cincinnati having had one more road game than home game.

            Minnesota, in a sense, has a bigger beef than Cincinnati, because the Vikings are a pretty decent team, but have played a lot of close games, so losing a couple of points a game to the schedule-makers is really a bigger problem for them than it is for Cincinnati.  Buffalo has still played the league’s softest schedule:

 Vegas

            I’ve been getting a lot of questions about how my system compares to the betting odds, but can you see why I don’t want to get into that?   I’ve got too many punts in the air anyway.   I’m trying to learn how to figure the quality of the teams, how often the best team wins, what the expected winning percentage is for each team, each team’s consistency, strength of schedule, who’s hot and who’s not, home field advantages, how to make the best guess about how many points will be scored in a game.  I don’t really know anything about the NFL; I’m just trying to figure out how to piece all of this together.   I don’t need to complicate the process by feeding another information source into what is already a barely manageable process.