Review: MLS Power Rankings 2014

Montreal in last place? I guess we'll see, but doesn't seem like they've earned that.

 

Looks like the MLSSoccer.com writing staff is the next on Hinchey's bad list...

BTW, Pablo is not the interim head coach. He's the coach that's currently heading the team, but he is not head coach, interim or otherwise.

 

Big D Soccer does a Dallas-centric recap of all the rankings here:

http://www.bigdsoccer.com/2014/3/19/5523164/rankings-recap-fc-dallas-week-three

Some of the variations are quite interesting. MLS' rankings have TFC 11th. Sports Illustrated has SKC 4th.

For the record, I do a set of rankings for 3rd Degree here:

http://soccerblog.dallasnews.com/2014/03/mls-week-3-power-rankings.html/

 

ESPN
2014 MLS Power Rankings: MARCH 31

 

DMN - http://soccerblog.dallasnews.com/2014/04/mls-power-rankings-week-4.html/

Big D Soccer Rankings Recap (Dallas centric) - http://www.bigdsoccer.com/2014/4/2/5570956/rankings-recap-fc-dallas-is-up-up-and-away

 

I took a few minutes to put together a bit more detail on how we come up with our rankings for 3rd Degree, but I also put together a rankings composite.

For your consideration - Defending the Thesis - http://soccerblog.dallasnews.com/20...the-process-of-ranking-mls-teams-weekly.html/

 

As usual I immediately scrolled down to the very bottom. For once I didn't find DC United!

 

Are the posters who maintain rankings models going to post their results this season? I really like the power rankings threads they produce.

@vividox @aletheist @Jough were some that I remember. Apologies to those I can't recall.

 

I occasionally added some as well. Statistics based rankings models tend to do notoriously badly at the beginning of a season. I don't even begin to have any faith in mine until every team has played at least 5 games.

 

3rd Degree/Dallas Morning News - http://soccerblog.dallasnews.com/2014/04/mls-rankings-week-5.html/

Big D Soccer Rankings Review (reminder Dallas-centric, but links to most of the major rankings) - http://soccerblog.dallasnews.com/2014/04/mls-rankings-week-5.html/

 

Sure. I just wanted to get the call out there, and make sure you guys knew there was at least one person who appreciated seeing those once they're available.

 

LOL I did this too

 

Agreed. I don't post the raw numbers for that reason. I use them to identify outliers (FC Dallas had an inordinate amount of home games, etc.) but who is played and how the games go is more important than numbers at this point.

 

Looks like the wrong link for Big D Soccer. Here is the correct one:
http://www.bigdsoccer.com/2014/4/9/5594604/rankings-recap-can-fc-dallas-go-any-higher

 

Oops! Thanks for the correction.

 

Duane Rollins is doing some advanced stats analysis over on www.Canadiansoccernews.com

This week he did a team ranking based off Total Shot Ratio(TSR) as well as a measure of "Luck" using PDO(shot percentage over save percentage). Both these values have proven quite accurate over the course of the season.

http://www.canadiansoccernews.com/c...th-part-II-The-fully-objective-power-rankings!

It's worth reading, and is part of a wider series called "Jock Math"

It suggests that SLC and Dallas are both underperforming, and their success comes form temporary luck more than sustainable performance.

Alternately, Montreal is playing very well, but has been particularly "unlucky" thus far. Ditto Portland, and thus both teams should bounce back over the course of the season.

 

Ask and you shall receive! As @aperfectring already noted, it takes at least a few games for a system that only uses data from the current season, like my ELO+ ratings, to generate meaningful results. In fact, the numbers will continue to be quite "bouncy" for a while yet. A few other reminders:

• A perfectly average team has a rating of 100; a 150-rated team is roughly 50% better than average, while a 50-rated team is about 50% worse than average.
• All teams begin the season with a rating of 100; there is no carry-over from one year to the next.
• Home-field advantage is calibrated to the average actual performance of home teams so far this season, and is updated with each new result.
• The weight assigned to each outcome is based on the average number of games played per team; so it is the same for all rating calculations at any point in time, but changes with each new result.
• There is no "bonus" for winning by multiple goals.
• Points projections come from adding actual points earned to date to expected points for remaining games, which are calculated using current ratings.

Here are the 2014 MLS ELO+ ratings and points predictions for 04/14.

Code:

Rank Team ELO+    Rank Team Points
 1   COL   144     1   COL   69.5
 2   RSL   127     2   SKC   62.1
 3   LAG   126     3   RSL   60.5
 4   SKC   126     4   LAG   59.5
 5   SEA   123     5   TOR   58.2
 6   DAL   119     6   CLB   57.7
 7   TOR   115     7   SEA   57.5
 8   CLB   114     8   DAL   57.0
 9   DCU   108     9   DCU   50.3
10   NER    97    10   PHI   44.0
11   PHI    96    11   NER   43.3
12   VAN    88    12   VAN   38.7
13   CHI    84    13   CHI   35.3
14   CHV    78    14   HOU   33.3
15   HOU    73    15   CHV   31.9
16   NYR    71    16   NYR   29.4
17   POR    71    17   MTL   27.8
18   MTL    70    18   POR   27.7
19   SJE    70    19   SJE   27.4

The average number of games played per team is now 5.47 (K=36.5), and so far home teams have a (rather anemic) winning percentage of 0.558 (H=12). COL is at the top mainly because their last two games were both road wins.

 

Even with that I did not see us at the top coming. ESPN's weekly rankings also put us at the top. I'm not used to this...

 

I'm not used to this either. Except "this" in my case is finding teams other than mine at the bottom.

 

Here's my numbers, they are both, at their heart based on the ratio of goals scored and goals allowed.
HFA - adjusts for the remaining schedule, home field advantage, and includes the points earned so far
EAP - Expected performance over a whole season if playing against an average opponent

Team	EAP	HFA
FCD	66.9	60.0
CLB	60.9	57.5
RSL	62.3	55.0
COL	60.9	54.6
LAG	66.4	54.4
SKC	51.0	51.9
SEA	54.2	51.2
TOR	45.8	50.5
VAN	54.6	48.4
PHI	46.6	46.3
DCU	41.0	44.8
HOU	42.4	44.6
CHI	43.4	42.4
NER	28.7	38.2
SJE	35.9	37.2
CHV	32.4	36.4
POR	36.3	36.1
NYR	31.4	35.7
MON	31.4	34.9