Monday, August 08, 2005

MLS: One step forward, one step back?



No, the title has nothing to do with the quality of play. I'm talking about preformance from year to year. I've always been interested in statistics, as you can tell, and one of the best sites is Baseball Prospectus. One of the ideas I've read there is about how teams that make a big jump up or down almost always come back to earth the next year (see the 2001 White Sox or 2002 Mariners). I thought I would take a look to see if it applies in MLS.

Through 2004, there have been 98 MLS seasons. Minus the first and last seasons for each team, and that's 74 seasons to take a look at. The first thing I did was eliminate the shootout and consider all of those games to be draws. Since the number of games is different for each year, I used points per game. I then measured the change from year to year.

An example of what I did:

Metrostars adjusted PPG

1996: 1.28
1997: 1.16
1998: 1.22
1999: 0.63
2000: 1.69
2001: 1.62
2002: 1.25
2003: 1.40
2004: 1.33

The first value is the change from the previous year, and the second is the change the next year:

1997: -0.13, 0.06
1998: 0.06, -0.59
1999: -0.59, 1.06
2000: 1.06, -0.07
2001: -0.07, -0.37
2002: -0.37, 0.15
2003: 0.15, -0.07

I did this for all 74 seasons, and tried to look for a pattern. Here's my results:

The number on the left is the change in ppg of any given season, and the number on the right is the ppg of the next season. The middle is the number of teams in that range. You have in groups of .10, .20, .25, and .50.

0.50 7 -0.28
0.40 2 -0.89
0.30 3 -0.07
0.20 5 0.09
0.10 11 -0.14
0.00 11 -0.15
-0.01 7 -0.10
-0.10 5 0.31
-0.20 6 0.02
-0.30 7 0.29
-0.40 5 0.08
-0.50 5 0.45



0.20 17 -0.21
0.00 22 -0.11
-0.01 13 0.06
-0.20 22 0.21



0.26 12 -0.33
0.00 27 -0.08
-0.01 17 0.04
-0.26 18 0.26



0.50 7 -0.28
0.00 32 -0.15
-0.01 30 0.11
-0.50 5 0.45


The next data is whether or not teams had a pos or neg PPG change.

pos 36 -0.18
zero 3 0.21
neg 35 0.15


This takes into account the number of years in a row ppg increased or decreased. That is, positive or negative years in a row, not the rate increasing. Only 23 times did a team go the same way two years in a row. DC has actually increased the past 4 years in a row.

u3 3 0.00
u2 10 -0.19
u1 36 -0.18



d1 35 0.15
d2 10 0.33

It seems that overall, teams are likely to increase after a decrease and vice versa. Of course, the teams who miss the playoffs have an advantage, due to the "You Suck Allocation." Let's seperate the teams into playoff and non-playoff teams, and see how they did the next season:

PO 53 -0.13
miss 21 0.30

Now, let's break down each group and see what happens:

Playoff teams


pos 31 -0.24
neg 20 0.01



Missed

pos 5 0.16
neg 15 0.35

The difference is pretty consistant whether you make the playoffs or not, although not making the playoffs overrides anything. Let's see how many years you make or miss the playoffs in a row has to do with anything:

Playoff teams


1 12 -0.12
2+ 41 -0.13




Missed


1 14 0.30
2+ 7 0.31

Doesn't appear to have made a difference.

Of course, the ultimate use of stats is to predict how teams will do the following season. PO2 is the breakdown of whether or not Playoff/missed teams were sporting pos. or neg. ppg change.

2004 points are listed first, then the other columns are what the 2005 point total is predicted to be based on the different ways of looking at the data. Of course, this takes into account the 2 extra games this year.


04 pts 0.10 0.20 0.25 .50
+/- U/D PO PO2
Dallas 36 10 32 28 34 33 33 48 43
Columbus 49 50 46 42 48 46 46 48 45
Kansas City 49 55 46 50 48 46 46 48 45
Los Angeles 43 49 39 43 41 40 40 42 38
DC United 42 40 41 42 40 39 45 41 37
Colorado 41 39 40 41 39 38 38 39 36
Metrostars 40 39 45 44 46 48 48 38 43
New England 33 38 42 44 39 40 40 31 35
San Jose 38 43 47 49 44 45 45 36 41
Chicago 33 50 42 44 50 40 40 45 46

I don't know how accurate these are. Maybe these totals should be inflated due to expansion. Dallas is severely underrated, probably because of the incredibly bad 2003 season. And of course, there is limited data to work with. It would be interested to try it with other leagues, although promotion and relegation could be a problem. Well, this is just a starting point so if you have any thoughts or suggestions, let me know. Peace.

(bigsoccer thread)

Comments on "MLS: One step forward, one step back?"

 

Blogger D said ... (10:17 AM, August 08, 2005) : 

Interesting work, and I'm glad to see that someone is taking a statistical approach to analyzing Soccer in terms of long term performance. That being said, please do not turn into Dayn Perry and start posting your, um, fantasies, here :)

 

post a comment