Developing CONCACAF coefficients, Part 2

Two weeks ago I wrote a post on my attempts to develop a formula for calculating club and domestic league coefficients in CONCACAF.  The primary application of this formula is the allocation of slots in future editions of the CONCACAF Champions League by ranking the clubs and their respective leagues.  It is possible to look at the coefficient formula as a club ranking, but that is NOT the original intention.

In the process of developing this formula I now have a nice database of competition data.  It took a while to teach myself some basic database programming so that I could calculate the statistics I needed, but now that I have them I can move forward.

Version 1.0

The first version of the club coefficients looked like this:

Club Coefficient = 0.25*(Points in Preliminary Round + Bonus Points in Preliminary Round) + 1.0*(Points in Champions Cup/League + Bonus Points in Champions Cup/League)

Bonus points are given for advancing to next round of the preliminary rounds (+1) or the Champions' Cup (+2).  I use results from the previous five editions of the CONCACAF Champions Cup and the preliminary competitions in Central America and the Caribbean, according to the European league season (e.g. 2005/06 - 2005 preliminary competitions for the 2006 CCC).  Two points are given for a win, and one for a draw.

These coefficients work well and have given results along the lines of what I expected.  The top ten clubs were drawn from Mexico, Costa Rica, and the USA.  There's a substantial gap between those clubs that have been regular participants in the Champions' Cup and those that have occasionally appeared, and another large gap between the 26 sides that have appeared in the Champions' Cup and the rest that have never advanced beyond the preliminary rounds.  Some teams were ranked higher than I expected (e.g. DC United in the top three despite not having won a Champions' Cup), most likely because of the lack of weighting by the respective national leagues.

Version 2.0

The next version develops the national league coefficient which will be used to refine the club coefficient.  The national league coefficient looks a lot like the club coefficient described above:

League Coefficient = [0.25*(Points in Preliminary Round + Bonus Points in Preliminary Round) + 1.0*(Points in Champions Cup/League + Bonus Points in Champions Cup/League)] / (Number of Teams)

So in a given season the league coefficient is the points won by all the domestic league's teams in the CONCACAF club competition divided by the number of teams that participated from that league in that season.

The revised club coefficient incorporates the country coefficient and the points earned by the club in the Champions' Cup.  This change makes the coefficient formulae similar to the UEFA coefficient.

Club Coefficient = 0.50*(League Coefficient) + (Points in Champions' Cup/League)

To create the club coefficient accurately, the league coefficient and the points earned in the Champions Cup/League must be from the same season.  The results from the previous five seasons are added together to give the CONCACAF coefficient.

Results

According to my coefficient formula, here are the top 10 national league rankings.  Puerto Rico's results are a bit misleading because only one team is being considered (the Islanders); perhaps that will change with their new domestic league.

Country	League Coefficient
Mexico	11.500
Costa Rica	8.708
USA	6.400
Jamaica	3.025
Trinidad & Tobago	2.625
Guatemala	2.508
Honduras	2.450
El Salvador	0.975
Suriname	0.700
Puerto Rico	0.700

And here are the top 10 club "rankings".  I know that it will be seen as that but it is strictly based on performance at CONCACAF club competitions, weighted by national league performance at such competitions. 

Club	Country	Total
CF Pachuca	Mexico	46.750
Deportivo Saprissa	Costa Rica	44.771
Club América	Mexico	38.750
CF Monterrey	Mexico	37.750
UNAM	Mexico	36.750
Toluca FC	Mexico	35.750
CD Guadalajara	Mexico	35.750
LD Alajuelense	Costa Rica	33.771
DC United	USA	31.000
CF Atlante	Mexico	30.750

(I tried to include coefficients from previous years, but my blog columns are too narrow to fit a table that wide.)

In contrast to the results of the previous coefficient formula, there is only one team from MLS in the top 10 club "rankings", and that team is now ninth on the list instead of third.  Pachuca is now top of the list, followed by Saprissa, which is what you would expect since those have been the most successful teams in recent editions of the competition.  Positions 11-20 are dominated by clubs from MLS and Costa Rica, and 21-30 by clubs from Guatemala, Honduras, Trinidad & Tobago and Jamaica.

I'll post the raw data later today.  Comments are welcome.  You're welcome to use the data and revise the coefficient formulae as you like, but I only ask that you reference me and this post.  If CONCACAF would like to use these formulae for future editions of the Champions League, send me an email and let's talk.

UPDATE: 

Download CCL_NatlCoefficients.ods

Here's the spreadsheet containing the raw data and my implementation of the coefficient formulae.  Sheet 1 contains the league coefficients, and Sheet 2 the raw data and club coefficients.  It's in OpenOffice Calc format.  Unfortunately I don't know how to convert ODS files to Excel; if anyone has any suggestions please let me know.

UPDATE #2 (9/22/08):

I just figured out how to save the spreadsheet in Excel format, thanks to that wonderful command known as "Save as".  Here's the spreadsheet; I don't know if the formulae have been retained properly, so if there are any problems please inform me.

Download CCL_NatlCoefficients_Excel.xls

Comments

[PhantomTollbooth]

Any chance that you'll post teams 11-30 to you blog for those of us who can't open the spreadsheet?

[Edgar]

Very nice work, Howard! But it's an exercise in futility, I'm afraid. I can't see CONCACAF giving more spots to Costa Rica at the expense of the US. The money is in MLF and MLS.

[Howard Hamilton]

PhantomToolbooth: I just uploaded an Excel version of my spreadsheet. I can try to upload the teams from 11-30 depending on how much of a pain it ends up becoming. I had to write in the top 10 teams by hand because a cut-and-paste operation wreaked havoc on the formatting.

Edgar, you hit on a very good point. There are too many financial considerations to allow MLS to have less than four slots in the Champions League. CONCACAF's definitely not the only confederation that doesn't allocate slots on a strictly sporting basis -- Copa Libertadores operates the same way. I proposed these formulae because I wanted to see what a CONCACAF coefficient would look like and how it might be applied. I think CONCACAF would take a look at it and put it on the shelf.

[Germaine Thomas]

What money? Almost no one showed up in the preliminary rounds. Besides, MLS is still losing money to this day.

More importantly, many view this tournament as a joke because it is quite obvious that it is designed to be a SuperLiga with other teams sprinkled in it. There is no merit and we clearly deserved it. We are only 2nd to Mexico in CONCACAF Club achievements and we have had 5 finalists since 2002. How many finalists has USA had despite the apparent favoritism to awards them automatic spots in the CCC?

If Costa Rica doesn't get the 3 spots we deserve next year, we should just not enter this crap tournaments at all and if they make us, we simply will send F teams. There is no sense in traveling thousands of miles just so CONCACAF can tell the referees to kick our clubs out because MLS & MFL is not winning. I can handle losses. Pachuca beat us fair and square this year. Robbing is one thing. CONCACAF tried doing so in 2004 when Saprissa & Alajuelense played Chicago & Monterrey respectively. That time we managed to beat the referees.

The same with Saprissa vs Monterrey & UNAM Pumas. Once again, we managed to beat the referees.

The same in 2006. This time, the referees beat us.

If CONCACAF is so focused on having a south of the border final every time, STOP FORCING US TO ATTEND! We can simply focus on our league and making trips to South America for friendlies.

Our 2 clubs have had better performances in this CCL then the 4 MLS clubs so far. As a matter of fact, MLS has yet to win a game. Next year, CONCACAF better make it Mexico 4, Costa Rica 3, MLS 3, etc etc.

We are getting tired of this BS.

[Chris]

Why not just use the same formula that Uefa uses? duh. Here are the rankings from CCL to date using that formula.

CAN 8
MEX 4.5
PAN 4.5
HON 4.5
CAR 3
CRC 2.5
SLV 1.5
USA 1.25
GUA 1
BLZ 0
NCA 0

Impact De Montreal 10.667
Cruz Azul 7.5
Marathon 7.5
Pumas 7.5
Tauro FC 7.5
PR Islanders 7
Deportivo Saprissa 5.834
Atlante 5.5
Santos Laguna 5.5
Olympia 4.5
San Francisco FC 4.5
Houston Dynamo 4.417
Joe Public 4
Luis Angel Firpo 3.5
Club Municipial 2.33
DC United 1.417
Harbor View FC 1
Alajuelense 0.834
Isidro Metapan 0.5
Chivas USA 0.417
NE Revolution 0.417
Deportivo Jalapa 0.333
Hankook Verdes 0
Real Esteli 0

[Howard]

Please read my first post on the issue: http://www.hexagonalblog.com/2008/09/developing-concacaf-clubcountry-coefficients.html

If you read it, you will find that I did indeed use the UEFA coefficient formula, but divided the points gained in the preliminary competitions (UNCAF, CFU) by 4 since the representatives from USA and Mexico received byes to the final eight of the old Champions' Cup and didn't have to play a preliminary round.

I haven't done calculations for this year's Champions League, but the formula will be identical to UEFA's as USA and Mexico representatives now play a preliminary round.

I haven't worked through all of your numbers, but just from inspection your Canada coefficient is incorrect, which means that your Impact de Montreal coefficient is incorrect.

[Edgar]

Any updates?

[Dave Clark]

With MLS representing both CAN and USA how is that factored into the national coeffecients? MLS could have 5 teams in the tourney if TFC wasn't so busy losing to Montreal andd Vancouver that is.