Guest Post: Power Rankings Round-Up, Week 16

Guest Post: Power Rankings Round-Up, Week 16


Today's guest post comes to you from Kreg Asay, who does wonderful stuff with Power Rankings each and every week.

This week's published Power Rankings proved to be an interesting test of the various systems the sites use.  I watched the placement of two teams in particular (LAG with 3 wins, and RSL with 2 losses); for RSL a small drop could indicate that they still consider RSL to be a quality team, and these losses are small stumbles in the marathon; a large drop would make me question their method of determining positions. For LAG, a small rise would indicate the site is still wary of the poor start to the season, while a large jump would make me want to question their methods.

One of the sites I've used in the averages has apparently ceased their Power Rankings; has not published a ranking since week 11.  That leaves 6 sites used to create the averages; so on to the results.

The most consistent ranking continues to be SBNation's, which is not surprising since it is also done by averaging the rankings of a number of writers.  Most of the teams remained fairly close to their placement from the previous week, the biggest movers were LA jumping from 14th to 9th, NYRB from 6th to 3rd, SS falling from 5th to 7th, SKC dropping from 3rd to 5th, and CR moving from 10th to 12th.

For RSL's case, all sites had them listed as #1 last week.  The sites with largest drop was ESPN with RSL at #5 and at #4.  Pretty much all sites had LAG making a huge jump across the board rising 5 places in all but two rankings.  Those other two were which had LAG move 9 places from 15th to 6th and ESPN moving LAG 4 places from 15th to 11th.  Curious that ESPN is so willing to drop RSL 4 places for poor results in 2 games, but move LAG up 4 for 3 positive results...  One other team had a large jump - SJE rising 4 places from 5th to 1st; the only other large movers on their list was Chicago rising 3 positions from 10th to 7th.  All others moved 2 places or less.

The results from all 6 sites leave us with the following Weekly Average Power Ranking:

San Jose, 1.57
Real Salt Lake, 2.71
New York, 2.86
DC United, 3.29
S. Kansas City, 4.57
Vancouver, 6.57
Chicago, 7.57
Seattle, 9
LA Galaxy, 9.57
Columbus, 9.71
Houston, 9.86
Colorado, 12.43
Montreal, 13
New England, 13.43
Chivas, 14.71
Portland, 15.57
Phil. Union, 17.29
Dallas, 18
Toronto, 18.71

Overall Power Ranking from start of season:

Real Salt Lake, 2.02
S. Kansas City, 3.06
Seattle, 4.08
San Jose, 4.47
New York, 6.67
Houston, 6.75
Vancouver, 7.26
DC United, 8.43
Colorado, 9.46
Chicago, 9.87
LA Galaxy, 10.28
Dallas, 11.44
Columbus, 13.23
New England, 14.07
Portland, 14.39
Chivas, 14.67
Montreal, 15.35
Phil. Union, 16.53
Toronto, 17.91

For the last 4 weeks I've tested an alternate method of determining a team's Power Ranking using the published Castrol Index.  Adding the listed points for each player on a team results in the following Ranking for week 14:

Castrol Total Team Points

San Jose, 9911
Seattle, 9309
DC United, 9306
Montreal, 8703
Columbus, 8470
Real Salt Lake, 8469
Portland, 8272
LA Galaxy, 8246
Dallas, 8220
New York, 8167
S. Kansas City, 7998
Chivas, 7904
New England, 7649
Vancouver, 7621
Toronto, 7611
Houston, 7496
Phil. Union, 7149
Colorado, 7116
Chicago, 7000

There are several problems that show this ranking isn't completely accurate.  For one, the team that is placed last in total points is 15th here, and other teams that are mid or upper table are at the bottom.

To try and correct that problem I factored in how many players each team has used over the season to create an Average Points per Player on each team.  The result changes the table's look a bit.

Castrol Average Points per Player
DC United, 443.1
Real Salt Lake, 423.5
San Jose, 413.0
New York, 408.4
Seattle, 404.7
S. Kansas City, 380.9
Montreal, 378.4
Portland, 359.7
Houston, 357.0
Chivas, 343.7
Dallas, 342.5
Columbus, 338.8
Chicago, 333.3
New England, 332.6
Vancouver, 331.3
Toronto, 330.9
Colorado, 309.4
Phil. Union, 297.9
LA Galaxy, 294.5

LA has used the most players during the season (28), while NY and RSL have used the least (20).  The top of the table closer resembles the scoreline, while some of the lowest ranking teams (LA, CR, VWc) are mid-table.  The 4 week test of this ranking system shows that the Castrol numbers don't accurately indicate that a team will rank highly based on player performance.