Week 13 Power Rankings

Calculated Power Rankings, Unit Strength, Strength of Schedule, and Estimated Final Record following week 10 of the NFL season. I use an engineering method of iteratively solving for best fit to the data (scores) to create the ranks and related applications that corrects for opponents played.

Everything is based on correcting the scoring offense and defense of each team for opponents, by calculating the expected scoring against an average opponent, instead of averaging the points scored against the actual opponents played. This allows all teams are evaluated on a level plane. Power is defined as scoring offense minus scoring defense, the list sorted by the net difference.

I run a full season calculation, and a most recent 4 weeks calculation and blend them (4:1) to acknowledge change and teams that are hot/cold.

Power Rank:

The Saints continue to hold their place at the top after reclaiming it last week, however Indy has closed the gap at the top. NO and Indy have clearly separated from the pack.

The Pats are propped up by the 59-0 beat down on the Titans, Titans are low because of it. I need to find a good non-arbitrary way to deal with the NE-Tn type situation, though I think that it will take care of itself in next year’s version. NE is declining and Tn improving (very low/high in the change indicator), however coming from such high highs and low lows, it hasn’t yet shown up that much in the rankings.

San Diego, Green Bay, and Philly have closed the gap on Minnesota. There could be a pretty big shakeup at the top of the order next week, with how close the top clusters are. The big slightly above average cluster of a few weeks ago has really broken up as teams has risen and fallen.

The strong trend of the year of the offense continues, a complete reversal of last season and early this season. 8 of the top 10 teams have top 10 offenses, whereas only 5 of the top 10 teams have top 10 defenses.

The Chargers continue their strong change showing again, which has been the case weekly for most of the season. They have ridden that wave all the way top 5 and could very easily make a push for the top 3 with a strong showing on Sunday.

Net = Projected net scoring agaisnt an average opponent (PPG)
Off = Projected offensive scoring against an average defense (PPG)
O Rk = Rank of offense
Def = Projected scoring by an average offense against the defense (PPG)
D Rk = Rank of defense
Chg = Difference between data run of most recent 4 weeks and full season (hot/cold) (PPG)
C Rk = Rank of change.

Strength of Schedule:

Instead of using points, or quality opponents, etc, the method that I chose to rank the strength of schedule is the average power rank of opponent played. The current power rank is used and each week the full season recalculated.

One trend that should be readily apparent is many of the best teams are on the bottom, and worst on the top. Good teams don’t play themselves and bad teams don’t play themselves, which gives a lopsided impression.

Seas = Average power rank of all opponents all season
Prev = Average power rank of teams played thus far
P Rk = Ranked difficulty of past schedule
Fut = Average power rank of teams yet to be played
F Rk = Ranked difficulty of future schedule.

Estimated Final Record:

I use the power rank, current record, and future opponents to estimate the final record for the team. The net scoring against an average opponent for one team is subtracted from another to determine who is favored and by how many points.

If a team is favored by 14+ points, I assumed that they win 100% of the time.
If a team is favored between 7 and 14 points, I assumed that they win 80% of the time.
If a teams is favored between 0 and 7 points, I assumed that they win 60% of the time.
If a team is not favored between 0 and 7 points, I assumed that they win 40% of the time.
If a team is not favored between 7 and 14 points, I assumed that they win 20% of the time.

No change in order this week from last, including the projected wild card winners. Indy joins New Orleans this week with the prediction of perfection, New Orleans was first projected at 16-0 last week.

Power Ranking Performance:

Week 1 record – 14-2
Week 2 record – 11-5
Week 3 record – 12-4
Week 4 record – 11-3
Week 5 record – 7-7
Week 6 record – 9-5
Week 7 record – 9-4
Week 8 record – 7-6
Week 9 record – 8-5
Week 10 record – 8-7
Week 11 record – 12-4
Week 12 record – 12-4
Wekk 13 record – 10-6
Total – 130-62
Missed Last Week – NE-Mia, Pit-Oak, Dal-NYG, SF-Sea, Mn-Az, Balt-GB

A couple of big upsets this week, lacking some sort of momentum based predicting tool those will never be predicted.

Sorry about the yellow tint, I didn’t notice when I put the graphics together.



Tags: , , , , , , , , , , , , , , , , , , , ,

2 Responses to “Week 13 Power Rankings”

  1. Kevin Says:

    Do you weigh the more recent games over past games? That could possibly handle momentum…

  2. waldo56 Says:

    I do weight more recent games more strongly (the most recent 4 games count 20% more than the rest of the season), however I don’t think that it is a strong enough weight to capture momentum. To capture momentum really requires a slowly changing baseline and a rapidly changing oscillator. Something such as 66% weight on the most recent game and 33% on the one previous, and no consideration for other games on the oscillator. How that relates to the slowly changing baseline (which is what is largely captured here) can give a clue to the next data point. It is definitely possible to code it, and is something that I plan on adding for next season’s run. The manner in which I set up this season’s modeling program is inefficient and pretty much has maxed out it’s capability without major structural changes. This years addition has no way of attaching a date to a game, I have two separate matrices of scores, one is color coded, and each week I go and erase all the scores from 5 weeks ago (via the colors) and add the new ones. The next version (in the works) ties the schedule directly into the scores, making it possible to code things like an oscillator.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: