Season
Résumé Rankings After Week 10
1. Florida 8-1 (1st)
2. Alabama 9-0 (4th)
3. Notre Dame 9-0 (2nd)
4. Kansas State 9-0 (3rd)
5. Ohio State 10-0 (5th)
6. Oregon 9-0 (8th)
7. Georgia 8-1 (6th)
8. Oregon State 7-1 (9th)
9. South Carolina 7-2
(NR)
10. LSU 7-2 (7th)
Out:
Florida State (10th).
Comments:
Before we get to this week’s rankings, I need to explain the final changes I
made to the formula a few weeks ago. I say “final changes” because I really do
think I’m done tweaking the formula at least for this season. I’ve used it now
for the last 3 weeks and I think it’s good enough to be useful. So here’s the
process.
The
Season Résumé Formula
As you know, a team’s
season résumé grade is based solely on the results on the field. There are 3
components to the formula: strength of opponent; location; and margin. Strength
of opponent is the most important factor. That’s where the majority of a team’s
“points” will come from. Location and margin are more like bonuses. In
addition, strength of opponent determines the value of location and margin, as
you’ll see.
Placing
Teams on the Strength of Opponent Scale
The first thing I do
each week is go through all 124 FBS teams and place each team on my “Strength
of Opponent” scale. I finally settled on a scale which is divided into 7
levels. The levels are as follows, from most value to least value: Best; Great;
Good; Average; Decent; Poor; Worst. The titles of each level may seem vague but
they aren’t really important. You could just as easily look at it as Level 7
through Level 1, but I prefer giving each level a title.
Placing all the teams
on the scale first is important. This helps me to keep things in perspective
and not be skewed by things like how a team is doing relative to their
expectations. A team with a good record in one of the weaker conferences may be
having a “good year,” while a team at the bottom of the standings in a BCS
conferences might be said to be having a “bad year,” but that doesn’t tell us
which team is a tougher opponent. For example, Kentucky is 1-9 and is probably
the worst team in the SEC, while Kent State is 8-1 and is one of the best teams
in the MAC. However, when the two teams squared off on the field this season
Kentucky won 47-14. I think it’s fair to say that Kentucky is the tougher
opponent, even if they aren’t very tough compared to their conference.
The point is that it’s
important to keep things on a national perspective and placing all the teams on
the scale first helps me to do that. Not to mention it makes the whole process
of calculating each team’s season résumé grade much faster and keeps me from
making errors.
The
Strength of Opponent Scale
One of the weaknesses
of my system is that the key component is more or less a traditional version of
“power rankings” based on how strong I think each team is. But to be honest, I
don’t know of any alternative.
Where a team falls on
my strength of opponent scale determines how valuable it is to beat that team
and how costly it is to lose to that team. The “Win” scale mirrors the “Loss”
scale, so that a win over a team in the top level is worth the most and a loss
against a team on that level is the least costly. Conversely, a win over a team
on the lowest level is worth the least and a loss to a team on that level is
the most costly. The scale is as follows:
Best (Win: 6) (Loss: 0)
Great (Win: 5) (Loss:
-1)
Good (Win: 4) (Loss:
-2)
Average (Win: 3) (Loss:
-3)
Decent (Win: 2) (Loss:
-4)
Poor (Win: 1) (Loss:
-5)
Worst (Win: 0) (Loss:
-6)
Location
Scale
Strength of opponent
impacts the value of location. In earlier versions of my formula I realized
that my system did not account for the fact that some road victories were more
impressive than others. My new formula attempts to correct that issue. The
location scale has two components: location of the game (win/loss at home, on
the road, or on a neutral field) and strength of opponent.
Again, the “Win” scale
mirrors the “Loss” scale. No location points are ever given for home wins,
regardless of opponent, and no negative location points are ever assigned for
losses on the road. Negative location points are given for home losses and
positive location points are given for road victories. For games on neutral
fields, some losses receive negative points and some wins receive positive
points. Here’s the breakdown:
Home Win (no location
points regardless of opponent)
Home Loss (Best: 0;
Great: 0; Good: -0.25; AVG: -0.5; Decent: -1; Poor: 1.5; Worst: -2)
Road Win (Best: 2;
Great: 1.5; Good: 1; AVG: 0.5; Decent: 0.25; Poor: 0; Worst: 0)
Road Loss (no negative
location points regardless of opponent)
Neutral Win (Best: 1;
Great: 0.5; Good: 0.25; AVG: 0; Decent: 0; Poor: 0; Worst: 0)
Neutral Loss (Best: 0;
Great: 0; Good: 0; AVG: 0; Decent: -0.25; Poor: -0.5; Worst: -1)
Margin
Scale
Strength of opponent
impacts the value of margin. In earlier versions of the formula I found that
blowout victories were really skewing things because my system made no
distinction between a blowout win over a terrible team and a blowout win over a
good team. My new formula attempts to solve this problem. The margin scale has
two components: size of margin [win/loss by 1 score (1-8 pts), 2 scores (9-16
pts), 3 scores (17-24 pts), or 4+ scores (25+ pts)] and strength of opponent.
Again, the “Win” scale
mirrors the “Loss” scale. To make things easier, I labeled each level of margin
(from 1 score to 4+ score games) A, B, C, and D with A being 1-score, B being
2-score, C being 3-score, and D being 4-score. Here’s the breakdown:
(Win) Worst: A (0); B
(0); C (0); D (0.1)
(Win) Poor: A (0); B
(0); C (0.1); D (0.2)
(Win) Decent: A (0); B
(0.1); C (0.2); D (0.3)
(Win) Average: A (0); B
(0.2); C (0.3); D (0.4)
(Win) Good: A (0); B (0.3);
C (0.4); D (0.5)
(Win) Great: A (0); B
(0.4); C (0.5); D (0.6)
(Win) Best: A (0); B
(0.5); C (0.6); D (0.7)
(Loss) Worst: A (0); B
(-0.5); C (-0.6); D (-0.7)
(Loss) Poor: A (0); B
(-0.4); C (-0.5); D (-0.6)
(Loss) Decent: A (0); B
(-0.3); C (-0.4); D (-0.5)
(Loss) Average: A (0);
B (-0.2); C (-0.3); D (-0.4)
(Loss) Good: A (0); B (-0.1);
C (-0.2); D (-0.3)
(Loss) Great: A (0); B
(0); C (-0.1); D (-0.2)
(Loss) Best: A (0); B
(0); C (0); D (-0.1)
Dealing
with FCS Opponents
I treat games against
FCS opponents different from all other games. Rather than placing FCS opponents
on my strength of opponent scale, I have chosen to separate FCS teams into a
class that doesn’t even qualify for the lowest level. This may not be the best
way of doing things, as we all realize that some FCS teams are better than the
very worst FBS teams, but I think it’s an acceptable way to deal with the
problem.
In my system, a win
against an FCS team automatically receives zero points regardless of the team,
the location, or the margin of victory. Basically I ignore games against FCS
teams. I say “basically” because there are a few exceptions. A win over an FCS
team by 1-8 points will earn a team -0.3 points. A loss to an FCS team is -10
points.
That may seem a little
silly and it might seem to go completely against the spirit of the system for
there to be a 9.7 point difference between an overtime win and an overtime loss
against an FCS team. However, any fan who has seen their team lose to an FCS
opponent knows that there is a gargantuan difference between beating an FCS
team by 1-point on a blocked PAT and losing to an FCS team by 1-point on a
blocked PAT.
I actually considered
making a loss to an FCS team an automatic disqualification from the season résumé
rankings, but I figured that -10 points would be punishment enough.
The
Season Résumé Grade
Once I’ve placed every
team on the strength of opponent scale I go through each candidate’s schedule
and calculate a grade for each one of their games using my formula (strength of
opponent + location + margin). I add up the game grades to come up with the
team’s season résumé grade.
Flaws
and Limitations
Just for good measure,
I would like to point out some of the system’s more significant flaws and
limitations.
Obviously, I rely
heavily on the strength of opponent scale, which is essentially my power
rankings extended to include all 124 teams in the FBS. Because the levels are
divided into integers of 1 (and because location and margin bonuses are based
off the strength of opponent ranking), moving a team up or down just one level
can make a huge difference when calculating a team’s season résumé grade. The
fact that beating a team may well lead to that opponent falling on the strength
of opponent scale is a paradox that is tough to escape.
Another problem is that
even though the margin bonuses are divided by integers of just 0.1 points,
there are still times where a meaningless score on the field can have a
significant impact on a team’s grade. And I will admit that my system of
dividing margins into 1-score, 2-score, 3-score, and 4+-score games isn’t
perfect. I think we’d all agree that on the field there is a big difference
between a 9-point lead and a 16-point lead, but in my system there is no
difference. Conversely, the difference between a 23-point lead and a 25-point
lead on the field is not usually great, but in my system it’s the difference
between a “C” value and a “D” value for margin bonuses.
It also seems to me at
times that I am not rewarding teams enough for not losing. Many of the top
teams have only lost to other great teams, so they don’t get punished much for
the loss. Consequently, it sometimes seems like a team without a loss is not at
any advantage over teams with 1 or even 2 losses. However, I do think this
makes sense in a way because most of the time teams with no losses have not
played as many tough games as teams with 1 or 2 losses have.
Eventually I may tweak
the system slightly so that even a 1-point loss to the top team in the country
results in at least a tiny negative number, like -0.1 instead of 0. I may also
tweak the system in the future to include negative scores for “bad wins,” such
as a triple overtime win at home against Pitt for example.
That actually brings me
to another limitation. My system does not account for flukes and luck and bad
calls and the like. A Hail Mary pass into the end zone that bounces off 4
players to give a team a 1-point win will look the same as a win when the other
team scores a meaningless touchdown with 5 seconds left to make it an 8-point
final margin.
Another key issue is
that each team’s place on the strength of opponent scale is spread equally throughout
the season. Thus injuries and suspensions which can have a major impact on a
team from one week to the next are not accounted for. For example, South
Carolina will play the remainder of the year without RB Marcus Lattimore. That
means that the teams playing South Carolina over the next few weeks will go up
against an opponent that is simply not as formidable as the opponent that teams
faced during the first 9 weeks of the season. My system has no way to account
for that. This may actually be the single biggest flaw of the system.
The
Season Résumé Rankings
I know this is far from
a perfect system but I’m actually pretty proud of it. This week’s rankings are
easily the best and most relevant in the history of my Season Résumé Rankings
Blog. While I may make some changes in the future, for the rest of this year
the formula is set. It’s going to be fun to see how things play out.
This
Week’s Rankings
Okay, in the unlikely
event that anyone has actually made it this far (and when you consider that it’s
highly unlikely that anyone actually started to read this entry in the first
place, we’re talking about a really, really unlikely event), it’s time to go
over this week’s edition of the Season Résumé Rankings.
This week I finally
expanded the pool of candidates for the rankings to include teams with
2-losses. Following play last weekend there were still 6 undefeated teams, 11
teams with 1-loss, and 16 teams with 2-losses for a total of 33 eligible
candidates. After calculating the season résumé scores for all 33 teams I was
able to rank them. The top 10 listed above was the result.
8 of the 10 spots in
the rankings experienced change this week. I only switched out 1 team, with FSU
falling out of the rankings from the #10 spot. This was primarily a result of
the pool of candidates being expanded to include 2-loss teams, which allowed
South Carolina to jump back into the rankings at #9.
3 teams moved up in the
rankings this week, with 2 of those teams climbing more than 1 spot. Both
Oregon and Alabama rose 2 spots in the rankings this week. The Tide moved from
#4 to #2, while the Ducks jumped from #8 to #6.
4 teams dropped in the
rankings while still remaining in the top 10. LSU was the only team to fall
more than 1 spot this week, dropping 3 spots in the rankings from #7 to #10.
I was somewhat
surprised to see that 1-loss Florida came out at #1 in these rankings for a
second straight week despite the fact that 6 teams remain undefeated.
No comments:
Post a Comment