clock menu more-arrow no yes

Filed under:

Looking Statistically at Team Defense

New, 2 comments

Today I decided to use the continuation of the first round of the MLS Playoffs as a point to test a few of my theories about soccer defense, how to rate it and hopefully a method to create a with/without you system that rates a player as +X% above team and +Y% above league.

My first test of this is to try and figure out which stats matter, and I decided to test it first with the correlation between shots against, goals against and points on the table. My initial hope was to be able to look at the USL Seattle Sounders and how they compared in these stats to MLS on average and at 15% worse than average.

Here's the table

Team
Games
Total Shots
Shots/Game
SA Rank
Goals
Goals/Game
GAA Rank
Points
GD
San Jose Earthquakes
30
124
4.13
1
38
1.27
4
33
-6
LA Galaxy
30
187
6.23
13
62
2.07
14
33
-7
Toronto FC
30
191
6.37
14
43
1.43
10
35
-9
FC Dallas
30
175
5.83
12
41
1.37
8
36
4
Colorado Rapids
30
162
5.40
11
45
1.50
11
37
-1
DC United
30
152
5.07
4
51
1.70
13
37
-8
Real Salt Lake
30
140
4.67
2
39
1.30
5
39
1
Red Bull NY
30
157
5.23
7
48
1.60
12
39
-6
KC Wizards
30
162
5.40
10
39
1.30
6
42
-2
CD Chivas USA
30
146
4.87
3
41
1.37
7
43
-1
NE Revolution
30
160
5.33
8
43
1.43
9
43
-3
Chicago Fire
30
156
5.20
6
33
1.10
2
46
11
Houston Dynamo
30
156
5.20
5
32
1.07
1
51
13
Colombus Crew
30
161
5.37
9
36
1.20
3
57
14
USL Seattle Sounders
37
287
7.76
41
1.11
           
MLS Average
30
173
5.77
39
1.30

The correlation between Goals Against and Points on the Table is a decent .61, there was no significant correlation between Shots Against and Goals, nor Shots and Points.

Dropping the USL Sounders, MLS Average and the 15% worse than Average from the Correlation runs. This actually increased the correlation between shots against and goals against to .43 which means I might be onto confirming the theory that there is a relationship between shots allowed and goals allowed. It seems intuitive, but having the stats confirm it is pretty nice.

The limited but existent correlation between Shots Against and Goals leads me to believe that I can get something done using shots against allowed by the team in games started by a defensive player as a way to rating the individual players.

The much stronger correlation between Goals Against and Points on the table is a primary reason why we have seen Seattle Sounders FC start there major acquisitions with Kasey Keller, and why the San Jose Earthquakes traded away allocation dollars to acquire one of the league's top keepers in history.

In fact if you drop just the Quakes from the MLS numbers the correlation between goals against and points increases to .71. Their lack of offense was obviously the reason for their lack of success.

While this is just an initial look at defense statistics in MLS I'm hoping with access to a broader database (wake up MLS!) or my own dedication during the offseason (building it myself) I can find ways that we can rate an outfield players defensive contributions (shots against), offensive contributions (goals+assissts) and find that those numbers correate to goal differential for the team. Goal Differential has a fairly strong correlation of .84.

The theory which I hope to test over the next few months would mean that we can judge the absence of a player due to injury or national team call-up and the effect that would have on the team. It would also be a method to test the impact of trades, transfers and free agent decisions.

Oh, one other nugget tucked in there as pertains to the likely upcoming player announcements for SSFC - if SSFC has a defense only 15% worse than the league average but has an average offensive team they would make the playoffs. Also, it should be noted that though Eylander faced two more shots a game than the average MLS keeper he gave up .39 goals less per game. That means that if he is needed during the double game weeks (US Open Cup) during the season he would give up two less goals than the average MLS keeper.

That's an extraordinary backup. What other team could rest their major keeper and count on the back to provide better than average keeping? For a team that wants to make a CCL run during Ljungberg's time with the club (two year contract) they MUST make a strong play at the US Open Cup in year one so that in year two they are in the CCL. This means having depth where their players are oldest (Keeper) and at CAM (Ljungberg) because in year one the team will play at least 35 matches in 30 weeks, and in year two that number should go to 40+ in 30 weeks just due to Open Cup and CCL run.

Lastly, it should be noted that Graham and Jackson have already been recognized as being at least marginal MLS quality defenders, and that Levesque is also of at least marginal MLS quality and he is capable of playing as an UpBack or CDM similar to the rover role that he has deployed with the USL club as recently as against Montreal in the playoffs. I say at least due to the fact that it turns out that MLS quality players have been signing USL contracts to get more money and playing time. Chemistry on the defensive line matters a lot, and keeping these three would be ideal, for if they are only 15% worse than the league average, and Keller is at least league average this would be a top 8 defensive club next year.