Tweet
Seattle (+4) 23 Pittsburgh 24
The pointspread in the Super Bowl is determined by public perception
more than a normal NFL game because more
amateur money is going to be placed on this game than on any other game
all season. The public deemed Pittsburgh’s
playoff road wins at Cincinnati, Indianapolis, and Denver as more
impressive than Seattle’s two home wins over
Washington and Carolina and thus Pittsburgh is an undeserving 4 point
favorite in this game instead of the 2 or 2 ½
point favorite that they should be. I don’t want to diminish
Pittsburgh’s playoff accomplishments, as beating the
Colts and Broncos on the road is truly a great achievement, but two
games is an awfully small sample size and the
Steelers simply aren’t much better than Seattle when you look at each
team over the course of the season (and even
in the playoffs only). Seattle is getting criticized for playing an
easier schedule than the Steelers and that is
absolutely true, but my math model compensates each team’s statistics
for the level of the opponents that they faced
and those compensated numbers are pretty close. Seattle’s offense was
one of the very best in the NFL this season,
as their balanced attack would averag 5.79 yards per play against an
average NFL defense. That number was derived by
compensating Seattle’s offensive numbers by the defensive numbers of
their opponents (and the level of their
opponent’s opponents) while excluding games against the Colts in week
16 (Indy didn’t play their starters) and
against Green Bay in week 17 (Seattle rested many of their starters).
Seattle’s compensated yards per rush is 4.70
ypr and quarterback Matt Hasselback’s compensated yards per pass play
(including sacks) is 6.91 yppp. The league
average for yards per play is 5.13 yppl (I take kneel downs and
quarterback spikes out of my stats, which is why
that number is different from the official stats you may see), so
Seattle’s offense is 0.66 yppl better than
average. Pittsburgh’s defense would allow 3.66 ypr, 5.35 yppp, and
4.60 yppl to an average offensive team and a
team’s compensated defensive numbers are adjusted depending on
opposing quarterbacks faced. For instance, my ratings
adjust for facing backup Jon Kitna for all but 2 plays in the
Steelers’ first playoff game against Cincinnati
instead of facing All-Pro Carson Palmer. Pittsburgh’s defense is 0.53
yppl better than average, so the Seahawks’
attack has a slim 0.13 yppl advantage against Pittsburgh’s defense.
Seattle has proven themselves against good
defensive teams, averaging 5.5 yppl in games against Jacksonville,
Washington twice, and Carolina – the 4 games that
they played against teams that allowed less than 5.0 yppl for the
season (Green Bay also allowed less than 5.0 yppl,
but Seattle sat their starters in that game). Those teams would combine
to allow 4.8 yppl to an average attack, so
the Seahawks were 0.7 yppl better than average against good defensive
teams, which is the same that they were
overall this season.
Pittsburgh’s offense was without quarterback Ben Roethlisberger in 4
games this season (weeks 6, 9, 10, and 11), and
the Steelers’ attack with Roethlisberger would average 5.67 yppl
against an average NFL defense. The rushing attack
is a bit below average (4.01 ypr compensated; NFL average is 4.13 ypr
when you exclude kneel downs as I do in each
game), but Roethlisberger averaged an impressive 7.90 yppp this season
and would average 7.81 yppp against an
average defensive team. Seattle is better defensively than most people
think, as the Seahawks would allow just 3.76
ypr, 5.64 yppp, and 4.81 yppl to an average NFL team, which isn’t
much worse than Pittsburgh’s compensated defensive
numbers. The Steelers are 0.54 yppl better than average offensively and
Seattle is 0.32 yppl better than average
defensively, so Pittsburgh has a 0.22 yppl advantage when they have the
ball. Pittsburgh faced 4 better than average
defensive teams with Roethlisberger at quarterback (New England in week
3 when the Pats had their secondary intact,
San Diego, Baltimore, and Chicago) and the Steelers averaged 5.1 yppl
in those 4 games against teams that would
allow 4.7 yppl to an average team, so Pittsburgh wasn’t quite as good
offensively against good defensive teams as
they were overall.
I project 5.32 yppl for Pittsburgh in this game and 5.23 yppl for
Seattle by using the compensated numbers for each
team and matching one team’s offensive numbers with the other
team’s defensive numbers. Projected turnovers are only
slightly in favor of Pittsburgh and special teams are slightly in favor
of Seattle. Overall, my math model favors
Pittsburgh by only 0.7 points. Some may argue that Pittsburgh is
playing better now than they played during most of
the season, but you may be surprised to find out that Seattle has
actually played better the last two weeks than
Pittsburgh has. The Steelers averaged 5.13 yppl in their playoff wins
at Indianapolis and at Denver (I left out
their playoff game against Cincinnati because of Palmer’s injury in
that game) and allowed 5.52 yppl, for a
difference of -0.39 yppl. The Colts and Broncos would combine to
out-gain an average NFL opponent at home by an
average of 1.23 yppl, so Pittsburgh was in fact 0.84 yppl better than
an average team in those two games from the
line of scrimmage. Seattle had a yppl differential of +0.99 yppl in
their two playoff victories over Washington and
Carolina (5.38 yppl on offense and allowed 4.39 yppl) and the Redskins
and Panthers would combine to out-gain an
average opponent by 0.24 yppl on the road, so Seattle was 1.23 yppl
better than an average team from the line of
scrimmage in their two playoff wins, which is considerably better than
Pittsburgh’s compensated performance the last
two weeks. So, those of you making a case for the Steelers based on
their playoff results should actually be
favoring Seattle in this game if playoff performance is your criteria
for Super Bowl success – which it shouldn’t be
(a full season of games is much more predictive than just two games).
Another way of calculating a fair pointspread involves using each
team’s individual game ratings and creating a
matrix using those ratings. For instance, Pittsburgh opened the season
with a 34-7 home win over Tennessee. That
game was at home (home field advantage this season was 3.5 points) and
Tennessee was 9.0 points worse than average
this season so the Steelers’ game rating in that week 1 win was +14.5
points (they won by 27 points, - 3.5 for being
at home, and -9 for Tennessee’s average rating equals +14.5 points).
I calculated a game rating for each of the
Steelers’ games and for each of Seattle’s games. I tossed out the 4
games for Pittsburgh that Roethlisberger didn’t
play in and the playoff game against Cincinnati when Palmer was knocked
out on the second play. For Seattle, I
didn’t include their week 16 game against the Colts, who rested their
starters, or the week 17 game against the
Packers when Seattle rested some of their starters. What I have
remaining is 14 game ratings for Pittsburgh and 16
game ratings for Seattle. If I compare each of Pittsburgh’s game
ratings against each of Seattle’s game ratings I
get a 14 by 16 matrix with 224 possible results - with each cell
representing a projected margin of victory/defeat.
I can use that matrix to determine what the median cell is (what the
spread should be) and how many cells have
Pittsburgh with an advantage. The matrix has a median cell of
Pittsburgh by 2 points and has the Steelers with an
advantage in 55.6% of the cells, which would represent their chance of
winning straight up. I have found that any
method of prediction using scoring margins is not as accurate as
predictions using yardage, since turnovers have
such an influence on scoring margins and past turnovers have a
relatively low correlation to future turnovers.
However, these teams project about the same in turnovers, and the
matrix is a good way to dampen the affect of
outliers (i.e. really good or really bad games that would skew the
season numbers), so I thought the use of a matrix
would be very insightful and pretty accurate in this case. The number
that is most important in this game is the
55.6% chance of winning that Pittsburgh has based on the matrix. Using
that percentage, and a standard distribution
of NFL margins of victory, gives Seattle a profitable 59.2% chance of
covering at +4 points (57.8% at +3 ½, 60.2% at
+4 ½). Pittsburgh may be worthy of being called the favorite in this
game, but the line should not be more than -2
or -2 ½ points and Seattle is certainly the percentage play in this
game. I also like the fact that Super Bowl
underdogs with the same or more number of victories (including the
playoffs), are 8-2-1 ATS since Super Bowl 15. I
will consider Seattle a Strong Opinion in this game and I have no
opinion on the total.
Super Bowl Propositions
The only Super Bowl proposition I see with any value is Seattle’s
Matt Hasselbeck to go UNDER in passing yards (the
line ranges from 234 ½ yards to 240 ½ yards). My math model projects
191 net passing yards for Seattle and 14 sack
yards, which gives Hasselbeck 205 gross passing yards. I’ll consider
Hasselbeck Under in gross passing yards a
Strong Opinion.
The pointspread in the Super Bowl is determined by public perception
more than a normal NFL game because more
amateur money is going to be placed on this game than on any other game
all season. The public deemed Pittsburgh’s
playoff road wins at Cincinnati, Indianapolis, and Denver as more
impressive than Seattle’s two home wins over
Washington and Carolina and thus Pittsburgh is an undeserving 4 point
favorite in this game instead of the 2 or 2 ½
point favorite that they should be. I don’t want to diminish
Pittsburgh’s playoff accomplishments, as beating the
Colts and Broncos on the road is truly a great achievement, but two
games is an awfully small sample size and the
Steelers simply aren’t much better than Seattle when you look at each
team over the course of the season (and even
in the playoffs only). Seattle is getting criticized for playing an
easier schedule than the Steelers and that is
absolutely true, but my math model compensates each team’s statistics
for the level of the opponents that they faced
and those compensated numbers are pretty close. Seattle’s offense was
one of the very best in the NFL this season,
as their balanced attack would averag 5.79 yards per play against an
average NFL defense. That number was derived by
compensating Seattle’s offensive numbers by the defensive numbers of
their opponents (and the level of their
opponent’s opponents) while excluding games against the Colts in week
16 (Indy didn’t play their starters) and
against Green Bay in week 17 (Seattle rested many of their starters).
Seattle’s compensated yards per rush is 4.70
ypr and quarterback Matt Hasselback’s compensated yards per pass play
(including sacks) is 6.91 yppp. The league
average for yards per play is 5.13 yppl (I take kneel downs and
quarterback spikes out of my stats, which is why
that number is different from the official stats you may see), so
Seattle’s offense is 0.66 yppl better than
average. Pittsburgh’s defense would allow 3.66 ypr, 5.35 yppp, and
4.60 yppl to an average offensive team and a
team’s compensated defensive numbers are adjusted depending on
opposing quarterbacks faced. For instance, my ratings
adjust for facing backup Jon Kitna for all but 2 plays in the
Steelers’ first playoff game against Cincinnati
instead of facing All-Pro Carson Palmer. Pittsburgh’s defense is 0.53
yppl better than average, so the Seahawks’
attack has a slim 0.13 yppl advantage against Pittsburgh’s defense.
Seattle has proven themselves against good
defensive teams, averaging 5.5 yppl in games against Jacksonville,
Washington twice, and Carolina – the 4 games that
they played against teams that allowed less than 5.0 yppl for the
season (Green Bay also allowed less than 5.0 yppl,
but Seattle sat their starters in that game). Those teams would combine
to allow 4.8 yppl to an average attack, so
the Seahawks were 0.7 yppl better than average against good defensive
teams, which is the same that they were
overall this season.
Pittsburgh’s offense was without quarterback Ben Roethlisberger in 4
games this season (weeks 6, 9, 10, and 11), and
the Steelers’ attack with Roethlisberger would average 5.67 yppl
against an average NFL defense. The rushing attack
is a bit below average (4.01 ypr compensated; NFL average is 4.13 ypr
when you exclude kneel downs as I do in each
game), but Roethlisberger averaged an impressive 7.90 yppp this season
and would average 7.81 yppp against an
average defensive team. Seattle is better defensively than most people
think, as the Seahawks would allow just 3.76
ypr, 5.64 yppp, and 4.81 yppl to an average NFL team, which isn’t
much worse than Pittsburgh’s compensated defensive
numbers. The Steelers are 0.54 yppl better than average offensively and
Seattle is 0.32 yppl better than average
defensively, so Pittsburgh has a 0.22 yppl advantage when they have the
ball. Pittsburgh faced 4 better than average
defensive teams with Roethlisberger at quarterback (New England in week
3 when the Pats had their secondary intact,
San Diego, Baltimore, and Chicago) and the Steelers averaged 5.1 yppl
in those 4 games against teams that would
allow 4.7 yppl to an average team, so Pittsburgh wasn’t quite as good
offensively against good defensive teams as
they were overall.
I project 5.32 yppl for Pittsburgh in this game and 5.23 yppl for
Seattle by using the compensated numbers for each
team and matching one team’s offensive numbers with the other
team’s defensive numbers. Projected turnovers are only
slightly in favor of Pittsburgh and special teams are slightly in favor
of Seattle. Overall, my math model favors
Pittsburgh by only 0.7 points. Some may argue that Pittsburgh is
playing better now than they played during most of
the season, but you may be surprised to find out that Seattle has
actually played better the last two weeks than
Pittsburgh has. The Steelers averaged 5.13 yppl in their playoff wins
at Indianapolis and at Denver (I left out
their playoff game against Cincinnati because of Palmer’s injury in
that game) and allowed 5.52 yppl, for a
difference of -0.39 yppl. The Colts and Broncos would combine to
out-gain an average NFL opponent at home by an
average of 1.23 yppl, so Pittsburgh was in fact 0.84 yppl better than
an average team in those two games from the
line of scrimmage. Seattle had a yppl differential of +0.99 yppl in
their two playoff victories over Washington and
Carolina (5.38 yppl on offense and allowed 4.39 yppl) and the Redskins
and Panthers would combine to out-gain an
average opponent by 0.24 yppl on the road, so Seattle was 1.23 yppl
better than an average team from the line of
scrimmage in their two playoff wins, which is considerably better than
Pittsburgh’s compensated performance the last
two weeks. So, those of you making a case for the Steelers based on
their playoff results should actually be
favoring Seattle in this game if playoff performance is your criteria
for Super Bowl success – which it shouldn’t be
(a full season of games is much more predictive than just two games).
Another way of calculating a fair pointspread involves using each
team’s individual game ratings and creating a
matrix using those ratings. For instance, Pittsburgh opened the season
with a 34-7 home win over Tennessee. That
game was at home (home field advantage this season was 3.5 points) and
Tennessee was 9.0 points worse than average
this season so the Steelers’ game rating in that week 1 win was +14.5
points (they won by 27 points, - 3.5 for being
at home, and -9 for Tennessee’s average rating equals +14.5 points).
I calculated a game rating for each of the
Steelers’ games and for each of Seattle’s games. I tossed out the 4
games for Pittsburgh that Roethlisberger didn’t
play in and the playoff game against Cincinnati when Palmer was knocked
out on the second play. For Seattle, I
didn’t include their week 16 game against the Colts, who rested their
starters, or the week 17 game against the
Packers when Seattle rested some of their starters. What I have
remaining is 14 game ratings for Pittsburgh and 16
game ratings for Seattle. If I compare each of Pittsburgh’s game
ratings against each of Seattle’s game ratings I
get a 14 by 16 matrix with 224 possible results - with each cell
representing a projected margin of victory/defeat.
I can use that matrix to determine what the median cell is (what the
spread should be) and how many cells have
Pittsburgh with an advantage. The matrix has a median cell of
Pittsburgh by 2 points and has the Steelers with an
advantage in 55.6% of the cells, which would represent their chance of
winning straight up. I have found that any
method of prediction using scoring margins is not as accurate as
predictions using yardage, since turnovers have
such an influence on scoring margins and past turnovers have a
relatively low correlation to future turnovers.
However, these teams project about the same in turnovers, and the
matrix is a good way to dampen the affect of
outliers (i.e. really good or really bad games that would skew the
season numbers), so I thought the use of a matrix
would be very insightful and pretty accurate in this case. The number
that is most important in this game is the
55.6% chance of winning that Pittsburgh has based on the matrix. Using
that percentage, and a standard distribution
of NFL margins of victory, gives Seattle a profitable 59.2% chance of
covering at +4 points (57.8% at +3 ½, 60.2% at
+4 ½). Pittsburgh may be worthy of being called the favorite in this
game, but the line should not be more than -2
or -2 ½ points and Seattle is certainly the percentage play in this
game. I also like the fact that Super Bowl
underdogs with the same or more number of victories (including the
playoffs), are 8-2-1 ATS since Super Bowl 15. I
will consider Seattle a Strong Opinion in this game and I have no
opinion on the total.
Super Bowl Propositions
The only Super Bowl proposition I see with any value is Seattle’s
Matt Hasselbeck to go UNDER in passing yards (the
line ranges from 234 ½ yards to 240 ½ yards). My math model projects
191 net passing yards for Seattle and 14 sack
yards, which gives Hasselbeck 205 gross passing yards. I’ll consider
Hasselbeck Under in gross passing yards a
Strong Opinion.
MY MEAT IN THE HOT DESERT.......
Comment