Showing posts with label FXI. Show all posts
Showing posts with label FXI. Show all posts
- Jason Kuenle

As the US talent pool has developed, the league’s ability to attracted foreign players has increased, and the league generally has matured, the diversity of formations and tactics in MLS has grown. Successful tactics hinge on two primary factors: 1) the individual skills of a team’s players and 2) the tactics and skills of your opponent. In other words, soccer tactics is a chess match; you can make moves with your pieces, but to be effective, those moves need to take advantage of the openings being offered by your opponent. While I hope to analyze how opponent tactics and skills affect MLS formations throughout the summer, the more straightforward analysis is on the chess pieces themselves.

The relative cheapness of US talent is caused by a host of factors including MLS’ single entity structure, European foreign player limits, and the US not having a mother nation that has an expedited passport process for nationals of former colonies. These factor, combined with a limited number of foreign slots, create a dynamic that places a premium on using foreign talent to fill holes in US development. This allows an analysis of MLS players to be a approximate reflection on US player development. The first part of this analysis looks at where managers have most often looked for talent outside of the US to fill out their first team rosters.

MLS First Choice Players and National Origin
PositionNumberDomestic*US DevelopedForeignPercent
CF30812170.0%
CM1460857.1%
LB1872950.0%
AM950444.4%
LM/LW18100844.4%
CB362101541.7%
DM19112631.6%
RM-RW18131422.2%
RB18140422.2%
* Domestic includes Canadians playing in Canada and all US nationals


There are many difficulties in creating the list that underlies this chart (i.e. should a player be classified as a center mid or a defensive mid or who is first choice for San Jose right now), but I don’t believe that there is any systematic bias that brings the results into serious question. Not surprisingly, this chart reflects the strengths and weaknesses of the national team pool; depth on the right and in more defensive midfield positions, left midfield covered well by inverting a right footed midfielder, reasonable depth, though in transition, at centerback. Also, there is a noticeable lack of strikers, attacking midfielders (either in a No. 10 role or as a more attacking box-to-box midfielder), and left backs. It underscores that MLS coaches know the limitations of the domestic player pool and have done a reasonable job augmenting it.

With every team in MLS playing a two centerback system, it is perhaps easiest to explore the use of players as chess pieces there. Tangible examples of combinations possible can be seen in the pairings of Marshall-James, Ream-Marquez, and Borchers-Olave. In functionality, the Marshall-James pairing in function reminds me of the Marshall-O’Rourke pairing in the Crew’s double winning 2008 season. O’Rourke and James fill similar roles, with their strength in covering space balancing well with Marshall’s marking and aerial abilities. The combination of Ream and Marquez are likely the best passing centerback combo in league history. The quality of their deliveries helps create the fluid attacking style that Hans Backe has put together in New York. The aerial prowess of Olave and Borchers protects RSL from crosses played from wide spaces, the tactical area of weakness when playing a diamond midfield. These are all examples of how, on a most basic level, pieces work together.

The above examples intentionally use a domestic-international pairing. It, in part, highlights a lack of diversity in the US central defender player pool. While anecdote does not equal data, looking at Americans in MLS who played centerback throughout college gives us an idea about the development of domestic talent. By grouping these centerbacks in age ranges, we can compare the play style diversity of each age range. The below chart is approximately in order from oldest to youngest.

Age RangePlayers
28+Gregg Berhalter, Jimmy Conrad, Eddie Robinson, Jay DeMerit, Danny Califf, Cory Gibbs, Nat Borchers, Jeff Parke, Bobby Boswell, Scott Palguta
27-24Ryan Cochrane, Jason Hernandez, Ty Harden, Drew Moor, Steve Purdy, Bobby Burling, Chad Marshall, Greg Janicki, Eric Brunner, David Horst, Raushawn McKenzie, Matt Besler, Darrius Barnes
23-19Sean Alvarado, Kwame-Watson-Siriboe, Michael Holody, Chris Sculer, Tim Ream, A.J. DeLaGarza, Ike Opara, A.J. Soares, Omar Gonzalez, Jalil Anibaba, Ethan White, Sacir Hot


The 28+ crowd consists of a fairly uniform type of player; a bigger, stronger centerback who is good in the air and a solid interior defender. In the 25-28 range, you continue to see that model of defenders in Brunner, Marshall, and Janicki, but you all see the emergence of a “utility” defender. These defenders, like Moor, Besler, and Hernandez, while primarily centerbacks, have some positional flexibility allowing them to move more easily into side back or defensive midfielder roles. Within this same age range, the “utility” role also developed in the reverse as players like Marvell Wynne (age 25), Heath Pearce (26), and before he left the league Jonathan Bornstein (26) all moved in from the side, while players like Danny O’Rourke (27), Geoff Cameron (25), Brandon McDonald (25), Patrick Ianni (25), and George John (24) have moved back to central defense either temporarily or permanently. This should not be overly surprising as the earlier chart demonstrated that the US over produces defensive midfielders and right backs.

However, in the 19-23 age range, the central defenders more naturally reflect the diversity required of different tactical systems. While there is still the prototypical back of the 28+ crowd, in a back like Omar Gonzalez, the diversity of the younger backs means getting solid delivery out of the back, no longer a team to move a Geoff Cameron-type from the midfield, they can instead choose Tim Ream and not have to trade defensive training and experience for passing proficiency. A team looking for more speed in its central defensive pairing, no longer needs to pull in a Marvell Wynne-type player from the wing, but can choose Ike Opara or A.J. DeLaGarza.

Again, this is all anecdotal, but if what is occurring in the center back position also occurs throughout the field, it seems likely that MLS coaches will have an increasingly diverse set of players on which to draw on, especially taken in combination with the growing international stature of MLS. Hopefully, this increases the diversity of play, the diversity of systems that coaches are comfortable with implementing, and the development of unique styles of play within MLS. The process of having our domestic pool produce the full array of chess pieces is the gateway to formation flexibility both for MLS and ultimately for the USMNT.


- Jason Kuenle

Like blossoming red and yellow tulips, one of the surest signs of spring is the annual MLS referee discussion. Here was the 2009 version, here was the 2010 version. And again we find ourselves in the same place once again as we enter an MLS season. As part of the discussion from last year’s discussion, I did research on the discipline records in other leagues. However, things came up and I never got to do the analysis. With the same issues arising again, I thought it time to blow the dust of those spreadsheets, update them with some new data, and give the analysis a go.



-Jason Kuenle

In economics, the concept of comparative advantage states that the most effective use of limited resources is achieved by maximizing relative advantage over absolute advantage. In soccer, the same concept can be applied to constructing a formation or lineup. The “limited resources” are the ten field players, “advantage” is measured in objective terms by goals or subjective terms by skills such as positioning, passing, dribbling, vision, covering, tackling, etc. Relative advantage is more important than absolute advantage. The goal is to field a side that maximizes the teams’ advantage. For teams fielding Michael Bradley, this presents a tactical problem in some formations; while he has two areas of absolute advantage, their juxtaposition can present issues in creating relative advantage.


Formula XI: Trickle Up Soccer

Friday, September 03, 2010 | View Comments

Around the web, though not on this site, there has been a lot of consternation over the proposed change to CONCACAF qualifying. Most of this has centered around the probable loss of the home-and-home between the US and Mexico, though some has extended well beyond that. The changes have ramifications for nearly every soccer federation in CONCACAF. Viewing the effects of these changes in their entirety paints a different picture than the quick-draw initial judgments likely created. Different countries will be affected in different ways. Countries are group by similar effect from the bottom of the CONCACAF world to the top.

Tier 11
Teams (3): Monserrat, Anguilla, US Virgin Islands
Average FIFA Rank: 200
Matches played under old system: 2
Matches played under new system: 2
Likely FIFA ranking* under new system: 200

Tier 10
Teams (3): Aruba, British Virgin Islands, St. Lucia
Average FIFA Rank: 190
Matches under old system: 2
Matches under new system: 8
Likely FIFA ranking under new system: 170

Tier 9
Teams (5): Dominican Republic, Turks and Caicos Islands, Belize, Dominica, Bahamas
Average FIFA Rank: 180
Matches under old system: 2
Matches under new system: 6
Likely FIFA ranking under new system: 170

Tier 8
Teams (8): Cayman Islands, Puerto Rico, Netherlands Antilles, Nicaragua, St. Vincent and the Grenadines, Bermuda, St. Kitts and Nevis, Grenada
Average FIFA Rank: 160
Matches under old system: 4
Matches under new system: 6
Likely FIFA ranking under new system: 150

Tier 7
Teams (4): Haiti, Antigua and Barbuda, Barbados, Suriname
Average FIFA Rank: 130
Matches under old system: 4
Matches under new system: 12
Likely FIFA ranking under new system: 120

Tier 6
Teams (3): Guyana, Guatemala, Cuba
Average FIFA Rank: 120
Matches under old system: 8
Matches under new system: 12
Likely FIFA ranking under new system: 110

A Canadian soccer fan

Tier 5
Teams (1): Canada
Average FIFA Rank: 100
Matches under old system: 8
Matches under new system: 12
Likely FIFA ranking under new system: 85


Tier 4
Teams (2): Panama, Jamaica
Average FIFA Rank: 85
Matches under old system: 8
Matches under new system: 18
Likely FIFA ranking under new system: 75

The Jamaica starting eleven

Tier 3
Teams (2): El Salvador, Trinidad and Tobago
Average FIFA Rank: 80
Matches under old system: 18
Matches under new system: 18
Likely FIFA ranking under new system: 75

Tier 2
Teams (2): Costa Rica, Honduras
Average FIFA Rank: 50
Matches under old system: 18
Matches under new system: 18
Likely FIFA ranking under new system: 50

Fans of Costa Rica

Tier 1
Teams (2): USA, Mexico
Average FIFA Rank: 20
Matches under old system: 18
Matches under new system: 18
Likely FIFA ranking under new system: 20

The new system funnels its matches and ranking advantages to the middle tiers of teams. Under the new system, tiers three through ten should see an average increase of about 10 spots in the FIFA rankings due to these changes. Those changes are enough to offset the loss of points that the top tiers lose by not playing each other as often.

Canada should get the largest bump with the new system. Because Canada does not play in a qualifier for the Gold Cup, it misses out on points collected by the Central American countries in the Central American Nations Cup and the Caribbean nations in the Caribbean Championship. Beating Aruba in WC qualifying is about equal in points as beating Iceland in a friendly. In fact, it’s worth more points than drawing Spain in a friendly would be. Because Canada’s only realistic chances at the higher multipliers matches come in the Gold Cup and World Cup qualifying, maximizing the number of these games is critical to raising their ranking.

The new system also gives a bit more equality to those nations that will be most important to the growth of CONCACAF. There is a core group of ten countries that have shown some ability to sustain at least decent quality soccer programs; USA, Mexico, Costa Rica, Honduras, El Salvador, T&T, Guatemala, Panama, Jamaica, and Canada. The countries include every country to make the hex round of qualifying, every Gold Cup winner, every confederation Gold Cup runner up, every UNCAF (predecessor to the Central American Cup) winner, and 9 of the last 10 Caribbean Cup winners. By nature of the hex format, four of these countries played less than half the games of the other six. In the new format, likely 8 of these ten teams will play the full 18 matches and the 2 who do not make it will play 4 more matches than they would have under the old system. For these countries, player exposure, gate receipts, and meaningful matches are critical for their development into strong regional competitors.

Another area of group is with tiers seven and eight. Tier seven sees an extreme jump in matches from 4 to 12. Tier eight are teams on the FIFA line of irrelevancy. Under the FIFA ranking formula, a team ranked lower than 150 is given a ranking of 150 for purposes of calculating points. Moving up teams near the 150 line has a positive effect for the entire system. Hopefully, this reduces the number of teams under the 150 mark from around one-half of the confederation to around one-third.

The bulk of the advantages in the new system are targeted at tiers 4 through 8. While there is a short term disadvantage to the preparation of US and Mexico for a World Cup, if the new system can stabilize and grow the middle tiers of CONCACAF the long term advantages outweigh the short term losses. In ten years, a final group of USA, Costa Rica, Jamaica, and Canada has the potential to be a challenging group. If the price of that is the diminishment of the Mexico rivalry, it may be a price worth paying.

* A note on methodology: The “likely FIFA rankings” were calculated by determining the countries current average match score, multiplying that by the number of matches played, subtracting the points from the old system, adding the predicted points from the new system, and dividing by the new number of matches played. The “likely FIFA ranking” is not predictive, but an estimation of what a country’s current FIFA rank would be if the qualification from last cycle had been done in the new format.


The US national soccer team poses for a photograph before their international friendly soccer match in Philadelphia Pennsylvania


Throughout this week, I’ve spent a lot of time analyzing the past and present; beginning with a historical look at the USMNT through rankings and qualification cycles and moving to the current player pool. Today, I’ll focus on what I see as the trends and make a few predictions along with way.

The World Cup

The most immediate future for the USMNT is the World Cup. The table below uses the numbers from Tuesday’s post for Group C participants to project the number of points that should be earned by each team. Using the projected points as a guide, I assembled records that would most closely duplicate that finish.

CountryStatistical ProjectionRecordPoints
England6.7922-1-07
USA5.0491-2-05
Slovenia2.8861-0-23
Algeria2.4000-1-21


To achieve this result, the US would draw England, beat Slovenia, and finally draw Algeria. If things proceed as planned in Group D, this would mean a match in the Round of 16 versus Germany. Statistically, the US is unlikely to move past this round with Germany as the opponent. However, if any of the other countries win Group D, the odds of US advancement to the quarterfinals would increase substantially.

As noted yesterday, the ability of a tier two team to beat a tier one team is highly dependent on the play of the team’s best players. As seen against Spain and Brazil in the final last summer, Donovan and Dempsey can be brilliant. If these two can consistently play like 10s, it gives the US the last piece of the puzzle to make a deeper run into the World Cup. If they can’t, like against Netherlands when Dempsey was out and Donovan ineffective, the US is just short of being competitive against these top quality sides. Exceptional play from these two players and a little luck along the way is all that would be necessary for the US to make it back into the quarterfinals.

...And Beyond

Using the trend line from Monday’s analysis, the US’ trend will reach the top ten during the summer of 2013. Depending on timing and FIFA rules, this could put the US in a position to grab an elusive seed in the World Cup. With Brazil receiving the automatic qualification spot, the top eight teams in the world figure to be seeded. A seed means an easier group and an easier path to the quarterfinals. A look at the current pool of players shows that the US player pool may change enough between now and then to fulfill the trend line’s prophecy.

Multiple players playing deep into European club tournaments is not yet a yearly occurrence for Americans, but it is getting closer. Replicating the gains that have been made and getting the results against top 10 teams expected from a team that has a solid foundation of growing stars is the first step. Edu’s Rangers, Onyewu’s Milan, and Kljestan’s rumored new club Anderlecht are qualified for the Champions League. Adu and possibly Johnson will be with Aris for their Europa League run, joining Parkhurst (Nordsjælland), Guzan and Lichaj (Aston Villa). Teenagers Mikkel Diskerud and Erik Benjaminsen are with Europa entrants Stabaek.

A strong performance in the World Cup would likely set off a number of club moves involving US players. At the older end of the spectrum, Beasley, Donovan, and Dempsey will all be 31 or 32 in 2014 but with moves this summer could all find themselves on clubs capable of regularly making the Champions League or Europa League Knockout Rounds. Outside the World Cup roster, Pearce has been linked to a move to Champions League participant, Bursaspor and rumors persist about Robbie Rogers making a move possibly to Netherlands when his contract is up at the end of this year.

Turkey vs United States


In Edu, Bradley, and Torres, the US has a trio of current 8s in the central midfield all of whom should be in their primes in 2014. At Rangers, Edu is already at a club capable of regular European play while rumors involving club interest in Bradley and Torres occurs frequently and could come to a head with solid performances this summer.

I expect Villareal to sell Giuseppe Rossi this summer and Jozy’s maturation in the EPL this year may find him battling for a roster spot for the Spanish club this summer. If he stays, his minutes may be limited this year, but eventually breaking into a starting lineup in Spain would give Altidore fantastic experience before the next World Cup. Davies’ Sochaux has been nearly average in defense this year, but very poor at finding the back of the net. His club situation should give him some stability if he can prove that he can once again play at a high level.

On the domestic front, MLS will need to continue to grow. A sign of this will be their competitiveness in the CONCACAF Champions League. I expect LA, Columbus, Real Salt Lake, and a N’Kufo infused Seattle to all be competitive. Add in an improving Toronto side, and with five teams capable of making the quarterfinals, getting a team through to the semis for the first time seems like a distinct possibility.

Of course, there will be young players who rise to compete for positions, especially in an aging US backline, so it is hard to imagine that the average US player rating will not increase over the next four years. It also seems likely that a US player or two will find a starting spot for a team playing in the Champions League knockout stage and consistently hold that 10 rating. Assuming the US maintains its strong record against weaker opponents, this foreseeable lift would give them all the pieces to sustain a position in the top ten by 2014.

A World Power?

Sports News - June 01, 2010

The US has got a long way to go before it will be recognized as a world power. Yet, a confluence of events sets a reasonable date for this in the future; 2022. The trend line at that point would project a ranking around 5. A top five team and a home World Cup would put the US in position to be a real threat to compete for the championship. A lot would need to happen to get to that point, the first is being the winning bid to host the World Cup. We will know if this happens late this year.

The players who will make an impact in that 2022 World Cup are developing as youth players right now. An entering high school freshman determining whether to play soccer or football this summer will be 26, and in the prime of his career, at the 2022 World Cup. A successful World Cup this year, a successful and growing MLS, and a growing population of kids that choose soccer over other sports will all be good signs that the future of US soccer will be very bright.


Formula XI: Comparing Talent

Thursday, June 03, 2010 | View Comments
Lucio, Clint Dempsey

Yesterday, I presented a framework for assessing individual players. Today, we’ll look at that data to judge the US’ performance relative to their talent. This analysis will focus on two attributes of these national teams: their average talent and the consistency of that talent.

Average Talent

The table below shows the average starter rating as complied according to yesterday’s rules for the teams in the May 26 FIFA Top 25.

TeamAveTeamAveTeamAve
1-Brazil9.4511-Russia8.8221-Nigeria8.73
2-Spain9.5512-Egypt6.4522-Norway7.73
3-Portugal9.3613-Greece8.3623-Ukraine8.64
4-Netherlands9.1814-USA7.7324-Switzerland8.27
5-Italy9.0915-Serbia8.1825-Slovenia6.64
6-Germany9.2716-Uruguay8.45
7-Argentina9.2717-Mexico8.36
8-England9.2718-Chile7.82
9-France9.8219-Cameroon7.82
10-Croatia8.7320-Australia8.00


A look at these numbers reveals that the only teams with an average above 9 are the same teams that finished higher than the US in the hypothetical single table purposed on Tuesday (Czech Republic has an average player rating of 9.36). Also notable is that the current top nine teams all have an average over 9, while Croatia and Russia are close to this level. Below is a graphical view:



The vast majority of countries fall in a consistent pattern. Only Egypt falls drastically outside of the pattern, with France and Slovenia to a lesser extent. The US falls squarely in the midst of this trend.

Drastic deviations from the pattern can occur when a national side under or overachieves its talent base or when the talent bases are incorrectly valued on the international market. In the case of Egypt, whose national team consists mainly of domestic players and thus not having the boost that being involved in European competition gives, I would suggest that the deviation is one of undervaluation of Egyptian talent instead of their overperformance. Slovenia’s plight is similar in that the club market is lagging behind the Slovenian rise in talent. In the case of the France, there is probably a mix between being overrated based on reputation and an element of underachievement. The US' placement on this graph suggests that, on average, players starting for the USMNT are correctly valued on the international market. The results obtained by the US are consistent with that talent and suggest that with the talent available, the US should obtain results that maintain a FIFA ranking that bounces around between 15 and 25.

Consistency

The correlation between average talent and FIFA rank is unsurprising. However, there are multiple ways to a get to an average. Teams can average an 8 by fielding eleven players ranked 8 (Team A) or by fielding three 10s, five 8s, and three 6s (Team B). Because these two teams have the same average, they will likely also have similar FIFA rankings. However, how they are likely to get those rankings in significantly different ways.

On Tuesday, I posed the question of why countries like Nigeria and Cameroon gain over 40% of the possible points against top 10 teams while others like the US, Switzerland, and Egypt are below 20% of points taken against those same foes. The answer, in part, is the consistency of talent on their team.

In the above Team A and Team B examples, Team A should play consistently like an 8. Regardless of which players see the majority of the time on the ball and exert the majority of influence in the game, the level of play should be consistent. This results in regularly beating teams worse than it and regularly struggling against more talented teams. Team B on the other hand has three players that would be equal on the field against a top team. When these players dominate the game for Team B, the team can play well above their average of 8. When these top players do not play well and cannot control the game or the 6s on the team are exposed, this team can also play well below their average of 8.

The below graph looks at the highest ranked three players for each of the teams in tiers two and three from Tuesday’s post versus the percentage of points taken against the top ten results:



Because the number of observations for some of these countries is too low to perform meaningful analysis, the below chart combines the countries with the same combined rating. Honduras and South Korea were combined and given a value of 25.5. Egypt also did not have enough data, but with no logical combining partner, this data point was removed in the combined analysis.

Combined RatingPercent of Points Earned
3022.2%
2931.9%
2831.1%
2722.3%
25.515.2%


This data has a correlation of .58, or what is generally viewed as a strong correlation. This means that 58% of the variation in points earned correlates with the variation in combined ratings. While correlation does not equal causation, the correlation here is stronger than the correlation between points against the top ten and the average talent for these same teams.

Using the trendline for this data, the expected value for the current US team is 22.5%. Applying that ratio to the 19 games that US has played against top 10 teams in the last 8 years would give the US an expected 13 points taken. The US has taken eight in these games, but holding the US to that standard is disingenuous. The US was not putting multiple starters into European tournament knockout rounds throughout the past 8 years. A more accurate average combined rating for the past eight years would be around 25. At that level, the expected value is 10 points taken, or only two more than were taken during this stretch. Moreover, the past 12 months, the US has taken 20% of points against the top ten. Here, the difference between the 20% earned and the expected 22.5% is negligible.

As discussed on Tuesday, the US has been a very consistent team, performing above the average of its tier against weaker competition but under its tier against better competition. In large part this is a function of the consistency of the US talent pool. The US "A" team consists of 7s, 8s, and 9s at nearly every position, but with no superstars the US has had a difficult time breaking through against top teams. Cameroon and Nigeria have had much better success against top teams, but their ranking is below the US because of losses to inferior teams. With the current talent pool at Bradley's disposal, the statistics show that the US’ performance is predictable and comparable to similarly situated teams.

Tomorrow, in the final installment, I’ll look at where all this data gets us and what it might mean for the future of US soccer.


Formula XI: A Framework

Wednesday, June 02, 2010 | View Comments
Oguchi Onyewu


Disclaimer: This post is exceedingly boring and only necessary because I still live by the rules of my high school math teachers; I show my work. If you have no interest in how I created the statistics that I’ll be analyzing tomorrow, feel free to skip this post and rejoin the series tomorrow. For those of you nerdy or bored enough to read this, thanks.

The first two posts in this series looked at the US’ place in the world, a growing second tier nation that struggles against top tier teams. To delve further than this requires looking at the players that comprise this team and similar teams around the world. As I said in the first post of this series, rankings are an impossible thing. This is doubly true when we move to the thought of rating individual players in a team sport. However, to really understand the current state of US soccer, it is a necessary endeavor.

Because most rating systems of individual players would require either a completely subjective rating or a PhD in mathematics and a QB-rating-like statistical formula, I am opting for a disaggregation-reaggregation concept. The theory behind this rating is that club success is dependent on aggregate individual success. By rating a club on its current successes and assigning the club’s success level to the players responsible for it, a rating for each player can be obtained.

The rating used starts at 10 and works its way down. In this rating, individual clubs are assigned a rating based on their best performance either in international club tournaments or league play. This rating is then assigned to each of the starters and regular subs of the club. Bench players are assigned a rating of two less than the club rating except for those bench players playing on a UEFA Champions League knockout round club. These players are assigned a rating of one less than the club rating.

While there is a level of error introduced by assigning the same rating to every starter on a club, the competitive transfer market in global football keeps this level of error relatively low. Because clubs, unlike national teams, can address positional deficiencies in the open market, the quality difference among the starters should be relatively small. A standout player on a “7” club, will be bought by a “9” or “10” club.

For space and usefulness, I’ve included only the ways for a club to be at least a “6”. The assignment of ratings are the result of way too much of my free time studying international results and player transfers. These ratings are not perfect (an underdog making a deep run in a tournament can overrate its players) and by virtue of the data type the ratings cannot be tested for significance. However, it will allow an analysis of what national teams look like from a “talent” perspective.

Tournaments

Sports News - May 31, 2010


The international tournaments offer the best opportunity to judge teams from different countries against each other. The interplay between the UEFA Champions League and the Europa League makes is relatively easy to determine where to break one rating level from another. The other competitions are a little harder to set break points because they are dominated more by individual nations. The best example of this is the Mexican domination of the CONCACAF Champions League. It would be difficult to claim that the quarterfinal losers belong in the same category as the quarterfinal winners. To a lesser extent, CONMEBOL, CAF, and AFC all suffer similar issues. The one thing that the Club World Cup results have shown in the last few years is that UEFA rules the club world, CONMEBOL is comfortably in second, and the top team from the other three confederations are all about equal. Below is a table showing the rankings for clubs in international tournaments.

RatingUEFACONMEBOLCONCACAFCAFAFC
10CL KO



9EL KOCLi QF


8CL GP4
EL GP
CLi KO
CLi GP3
CS SF
CL SFCL SFCL SF
7EL POCLi GP4
CLi R1
CS R2
CL QF
CL GP
Cup SF
CL KO
6CL R2
EL R3
CS R1CL GPCL R2
Cup GP
CL GP
Cup SF

Competition AbbreviationsRound Abbreviations
CL = Champion's League
EL = Europa League
CLi = Copa Libertadores
CS = Copa Sudamericana
SF = Semifinalist
QF = Quarterfinalist
KO = Knockout Stage Participant
GP = Group Stage Participant
GPx = x Place in Group
(i.e. GP3 = 3rd place in group)
PO = Playoff Stage Participant
Rx = x Round Participant
(i.e. R2 = 2nd Round Participant)


Leagues

Sports News - May 16, 2010


Aggregating individual club success in international tournaments for each countries gives an approximate view of the relative strength of the domestic league. Except for MLS, each country’s top division is divided into thirds, with the top third rated at the highest number, the middle third rated at the middle number, and the bottom third rated at the lowest number. For league sizes not divisible by three, the following patterns were followed, 20 teams: 6-8-6, 16 teams: 5-6-5, 14 teams: 4-6-4. MLS is split into two because the league structure shrinks the difference between the top and bottom teams. For a country’s second division a 25%-50%-25% ratio is used with the top quarter of the second division of equal rating as the bottom third of the top division. For the big four, this means second divisions of 7-6-5.

RatingsUEFACONMEBOLCONCACAFCAFAFC
9-8-7England
Spain
Italy
Germany




8-7-6France
Russia
Ukraine
Romania
Portugal
Brazil


7-6-5Netherlands
Turkey
Greece
Belgium
Switzerland
Denmark
Argentina
Uruguay
Mexico

6-5

MLS

6-5-4Scotland
Bulgaria
Czech Republic
Norway
Austria
Serbia
Israel
Cyprus
Sweden
Slovakia
Poland
Croatia
Belarus
Colombia
Venezuela
Ecuador
Chile
Paraguay
Peru
Bolivia
Costa Rica
Honduras
Egypt
Tunisia
Japan
South Korea


For clubs in international play, their final rating is the higher of their international tournament rating and their league rating. Thus, for Rangers, they would be an “8” even though their position in the Scottish Premier League would make them a “6”.

For a player who has been with multiple clubs during a season, the higher rating for the player will be used. In Landon Donovan’s case, this would make him a “9” as he played in the Europa League Knockout Stage with Everton. Interestingly, he was a “9” the year before as well sitting on the bench for a Champions League Knockout Stage club.

By assigning the club rating to individual players and averaging those ratings for starting lineups for each country’s national team, a reasonable representation can be made of the nation’s underlying on-field talent. Tomorrow, we’ll look at what that data says about the US compared to the world.


United StatesVS.Spain


As discussed yesterday, the US is growing as a soccer nation. The next question is how do we compare with the world? As the world perception goes, the US is not a current threat to any of the established soccer powers of the world. That perception is held by a number of fans here as well. Project 2010 was mocked when it was released and many today believe that the US is far from being a competitor. But is that perception reality?

The chart below shows a variety of teams in the world. Most were chosen because they are among the best in the world, a few were chosen for regional or World Cup reasons. For each team, the percentage of points taken against teams with ELO rankings in the top 10, ranked 11 through 25, 26 through 50, and over 50 was calculated along with the overall percentage of points won from matches after the 2002 World Cup through February 1, 2010. The final column is the points earned by the country's team, if the team was to play a hypothetical home and away series with every country (including themselves) in the ELO top 75 using the calculated percentages of points earned. This results in 10 games at the 1-10 percentage, 15 games at the 11-25 percentage, and 25 games at both the 26-50 and the 51+ percentage. The order of the table is ranked by these points. Think of it as if the world championship was decided by single table rather than playoff format.


CountryTotal1-1011-2526-5051+Pts
1Spain80.8%79.2%79.6%78.2%83.8%362
2Netherlands71.2%48.3%57.1%80.8%89.3%336
3Brazil69.7%59.6%61.1%76.5%81.8%328
4England68.2%38.1%61.4%75.0%90.0%326
5France70.7%53.7%63.9%61.5%92.7%321
6Argentina65.0%34.9%66.6%73.5%84.4%318
7Germany67.0%33.3%75.4%71.4%78.9%313
8Czech Republic65.1%40.5%47.6%79.2%78.7%304
8Italy66.3%48.3%70.8%58.3%82.3%304
10Portugal66.7%38.3%42.4%62.7%90.0%290
11USA65.6%14.8%55.6%66.7%86.8%289
12Croatia64.7%41.7%51.1%56.7%85.7%285
13Mexico62.5%38.9%53.6%56.2%78.6%274
14Egypt68.4%14.3%63.0%63.6%75.1%273
15Denmark59.1%38.5%44.4%69.7%68.8%271
16Russia59.1%22.8%59.3%46.7%86.5%267
17Cote d'Ivoire65.9%11.1%71.4%48.1%81.6%265
18Switzerland56.3%17.6%59.0%50.0%83.3%264
19Ukraine54.7%18.5%55.6%60.0%70.4%257
20Greece59.1%39.2%48.3%38.1%88.2%256
21Chile52.4%22.2%42.6%51.0%83.3%253
22Japan62.6%15.2%43.1%52.1%79.3%245
23Uruguay50.6%17.6%72.9%54.5%54.0%239
24Australia61.7%20.8%44.4%41.7%76.7%230
24Turkey55.9%17.6%62.1%40.5%68.2%230
26Cameroon59.7%41.7%41.7%35.2%75.2%228
26Colombia48.0%33.3%34.9%54.1%63.6%228
28Sweden50.3%23.3%21.1%53.0%74.4%224
29Paraguay47.7%45.1%31.5%39.6%70.5%221
30Nigeria62.7%44.4%41.7%28.6%74.4%219
31Ghana52.3%25.0%33.3%51.5%59.4%211
32Honduras54.6%0.0%50.0%43.9%62.4%204
33South Korea55.0%27.8%40.7%27.5%70.0%200
34Costa Rica50.0%0.0%17.5%45.3%67.2%185
35Algeria49.8%0.0%46.7%33.3%54.6%174
36Slovenia40.6%20.0%12.8%31.1%63.4%165


As you can see in this analysis, the United States does fairly well, placing 11th behind ten countries recognized as traditional footballing powerhouses. A closer look at the table begins to reveal tiers of countries.

Above 300 points is the top tier, the soccer elite. These nine teams include the only 8 teams to have made a World Cup Championship match in the last 50 years. The only team from this tier not in that list is Spain, who finally broke through their string of underperformances with their Euro 2008 Championship and have been the most dominate team in the world over the last 8 years.

Around 250, there is another break, with only one team between 253 and 239. This second tier contains a number of the teams traditionally viewed as just below the powerhouses. Sides like Portugal, Mexico, Russia, Denmark, and Cote d’Ivoire; having the talent to make a run deep into the World Cup, but not considered threats to win the tournament.

Using 300 and 250 as guideposts, I’ll set a third break more arbitrary break at 200. A break here creates a third and fourth tier. The third tier primarily contains teams that are either growing as a soccer nation; Japan, Australia, Cameroon, Honduras, or have fallen back from being in the second tier; Uruguay, Colombia, Sweden. An analysis done four years previous to this one would likely have put the US in the third tier.

Finally, because of the teams chosen, tier 4 here consists only of a few teams. But these teams represent what tier 4 is, teams traditionally viewed as also rans; these teams have taken 50% or less of the possible points against all competitors during the last eight years.

Aggregating the teams in each tier allows for a further analysis of how the US compares. To do this, I’ve removed the US from tier 2 numbers and averaged the percentages of every team in each tier:


1-1011-2526-5051+
Tier 148.4%64.8%72.7%84.7%
Tier 227.6%53.7%54.8%81.0%
Tier 326.0%43.1%43.5%69.0%
Tier 46.7%25.7%36.6%61.7%
USA14.8%55.6%66.7%86.8%


Comparing the US to the aggregated tiers, a trend becomes obvious. Compared to even the top tier, the US is great at beating lesser teams. This is not just a case of being in CONCACAF, as Mexico’s percentage against the 51+ category is below both the tier 1 and tier 2 aggregates and Honduras is below the tier 3 average. The US is also very good against teams 26 through 50, falling between the aggregates for tier 1 and tier 2, but being much closer to the tier 1 percentage. Moving up to matches against teams 11 through 25, the US is again between tiers 1 and 2, but closer to the tier 2 number.

However, it is clear that the US is not in the range of tiers 1 or 2 when is comes to getting points off of the top ten teams. While the tier 3 number is high due to “giant killers” Paraguay, Nigeria, and Cameroon that all have 1-10 percentages of over 40, removing these outliers only brings the tier 3 average down to 20.1%. This still places the US somewhere between tiers 3 and 4.

This is why the US is not yet taken seriously by the rest of the world and why they are not seen as a threat to win international tournaments. As discussed yesterday, the US has become better against poor competition. Their bottom end numbers have evolved from their tier 3 days, but their top end numbers are well below their current tier 2 peers. The performance against Spain and in the finals of the Confederations Cup may be changing that perception, but more results are necessary for the US to turn the corner against top 10 teams. If they can, there would be little that distinguishes the US from second tier powers like Portugal, Mexico, and Croatia.

But the distribution of results leads to a question: what separates teams like the US, Egypt, and Switzerland who do relatively well against all teams except the top 10 from teams like Nigeria and Cameroon who have a better points percentage against the top 10 than they do against teams 26 through 50? Tomorrow, I’ll lay a groundwork for that analysis.


Formula XI: Where Are We?

Monday, May 31, 2010 | View Comments
Obama, Biden, And Bill Clinton Pose For Photo With US National Soccer Team

Rankings are an impossible thing. There is no art or science that can take two teams of similar talent and say one is definitively better than the other. But by drawing comparisons to others national teams, by looking at what results are obtained, and by analyzing player pools, a better understanding of the United States’ place in the world of soccer is hopefully obtainable.

The goal of this weeklong series is to look at the strengths and weaknesses of the US team as objectively as possible through comparison with other teams throughout the world. This process will result in a lot of data being thrown your way. I’ve attempted to focus each post on the highlights of the data being presented, though I welcome comments delving deeper into what’s presented.

The Growth of the US Team

Rankings

A logical place to start answering the question of where are we, is to answer the question where have we come from. For some perspective on the situation, rankings are a good place to start. Because FIFA changed their ranking formula in 2006, it’s impossible to use those rankings to obtain a consistent measure for more than this current world cup cycle. That amount of data would be pretty meaningless. However, rankings using a modified version of the ELO rankings used to rank chess players has been consistently applied to every official international soccer match ever played. Charting the US ELO rankings from their reappearance in the World Cup in 1990 to the end of 2009 looks like this:


I’ve added an exponential best fit trend line to show the approximate path that the US has taken in the last two decades. As you can see, there is a definite up trend though there are periods of success relative to the trend line (mid 91 - mid 93, mid 95 - mid 97, and mid 99 - mid 06) and periods of underperforming the trend line.

Another way to look at this data is by the amount of time each year that the team was above plateau markers. The table below starts in 1995 (the first year that the US cracked the ELO top 30).

Percentage of Days at or above ELO Plateau Numbers (1995-2009)
Top 30Top 25Top 20Top 15Top 10
199547%0%0%0%0%
199672%0%0%0%0%
19970%0%0%0%0%
199821%1%0%0%0%
199949%42%41%0%0%
2000100%100%69%0%0%
2001100%68%34%7%0%
2002100%92%18%2%0%
2003100%100%100%0%0%
2004100%100%92%29%0%
2005100%100%100%57%0%
2006100%47%47%40%0%
200785%46%40%4%0%
2008100%82%74%10%0%
2009100%100%99%63%5%

As you can see, besides the slip endured by sending a “B” team to Copa America, the US has been consistently a top 30 side since 2000. 2003 through 2005 can make a strong claim as the best run of form that the US has ever had with a three year run almost entirely in the top 20. However, by the rankings, the most successful single year belongs to 2009, when the team first cracked the top 10 and had the highest percentage of days in the top 15.

Qualification

USA team 2005

Another logical place to look at the progress of the national team is the cyclic World Cup qualifying process. The CONCACAF qualifying process of three group semifinals and the “hex” finals began for the qualifying for the 1998 World Cup. In the four qualifying cycles, the US has dropped the following number of points compared to the 48 possible points in each cycle; 1998 – 17, 2002 – 20, 2006 – 14, 2010 – 13. The total of these points dropped is 64 points. 57 of those 64 points have been in at least one of the following categories: 1) away to Mexico or Costa Rica or 2) to teams with an ELO ranking worse than 50.* Significant trends exist for both of these categories.

Any serious US fan knows about the constant struggles of the US at Azteca and Saprissa. For the US to turn the corner and really be a world powerhouse, those are games that the US will eventually need to get some points from, but a consistent inability to win in those places tells us little about how the US has improved over time. The other category however, shows a more usable and positive trend:

US Games Versus ELO 51+ teams in World Cup Qualifying Cycles
CycleWinsDrawsLoses% of Points Taken
199864261%
200254163%
200684078%
201081183%

As was shown with the US’ ELO rankings, compared to previous US teams, there is a trend showing continual improvement by the national team. While this data shows that the US is getting better compared to itself, how the US compares with other nations around the world will be the focus of the next installment.

* The other 7 dropped points cannot be categorized in a way creates enough observation to undertake statistically significant analysis.


Netherlands vs USA


Yesterday, for the fifth time in the last 12 months, the US faced a top 10 team. The US will face another one on June 12th in facing England. A look at some statistics from these games shows that the Netherlands game was both quite different and very much the same as the Confederation Cup games last summer.

OpponentPossessionShots - On Goal - Goals
Italy58%22-13-3
Brazil59%23-11-3
Spain56%29-8-0
Brazil59%31-13-3
Netherlands65%10-6-2

All stats from FIFA except Netherlands from ESPN




While the Netherlands possession numbers are higher than those from the Confed, there is a drastic difference in shots taken. However, removing the penalty kick from the Dutch totals a strong trend appears for the US defense. The table below looks at the percentage of shots on goal and the percentage of shots on goal scored by these top 10 teams.


Opponent% of Shots on Goal% of On Goal Scored
Italy55%25%
Brazil48%27%
Spain28%0%
Brazil42%23%
Netherlands56%20%



These numbers confirm that the game against Spain was a fantastic defensive performance by the US. However, the rest of these numbers do not stray that far from the general output of these top teams. Below are the statistics in yesterday’s games for the other top 10 teams that ESPN has statistics for:


TeamShots-On Goal-Goals% of Shots on Goal% of On Goal Scored
Germany9-1-011%0%
Argentina5-2-140%50%
Spain6-3-250%67%
France11-4-036%0%
England14-7-350%43%
Italy12-5-042%0%
Average10-4-140%25%



These numbers show some similarity to the US numbers above. The conversion rate of 25% shots on goal to goals is similar to the percentage allowed by the US against top ten teams. The 40% shots on target number is made slightly lower by Germany’s terrible performance yesterday, but the 35%-50% range for the other teams is only slightly lower than the 40-55% that the US has run against similar teams. To expect a much better overall defensive performance percentage-wise than the US has produced against top 10 teams, is to expect the US to defend better than top 10 teams.


The US does struggle in another category. Another look at the shots attempted shows that against good competition, the top 10 teams averaged around 10 shots attempted. That is a far cry from the average of 25 that the US allowed in the Confederations Cup, but a very similar number to that allowed yesterday.


The real challenge is combining the two efforts. If the US can begin to limit top teams to a number of shots in the low to mid teens while keeping percentage like those found in the later half of the Confederations Cup, the US would have a chance against these teams. Getting to the point of allowing 15 shots, with 40% on goal and a 25% scored percentage means giving up 1.5 goal per game to these top sides. While this will not always be enough to beat these teams, it would be enough to keep them competitive and put them in position to draw or win these games a position that the US has rarely found itself in.


While Bornstein has correctly been made the goat for a terrible decision in the box, the US defense overall did well. There were few balls from a very talented Dutch midfield that sent players through on goal. As has been noted elsewhere, the play that resulted in the penalty was well covered by the US and on the second goal a better developed partnership between Edu and Bradley likely would have resulted in Bradley staying at home on van der Vaart who provided the assist instead of pinching in on Sneijder whom Edu was covering..


Whether yesterday's drop in shots conceded is an aberration, a result of the game being a friendly, or a result of the Dutch style of play is impossible to tell. However, if it is a trend that continues it could be the beginning of the US being more competitive in games like yesterday’s.


USA Starting Eleven


The Best XI, the “A” team, rightly gets the majority of analysis by pundits and bloggers. The arguments surrounding who should be in the “A” lineup are as complicated as they are simple; given a formation, is the player the best in that position. What “best” means is, of course, complicated, but the goal of putting the best lineup on the field is a simple concept.

The “B” team is a whole different animal.

The “B” team serves multiple purposes; to analyze this animal, I employed a slightly simplistic model looking at the last 11 years of data. It’s not a perfect model, but the best I could do with accessible data, and I think it does give a reasonable approximation of the player pool in each year. The top 10 field players and top goalkeeper in minutes were marked as first team; the second ten field players and second goalkeeper in minutes were marked as second team. No attempt was made to divine the depth chart that may have been altered because of injury nor were field positions strictly accounted for. On average, each team from each year ended up with one goalkeeper, four defenders, three midfielders, two midfield/forward combos, and a forward. On average, these “B” team players play a substantial amount, each logging an average of 400 minutes each year.

By defining the relationship between the players on the “B” team and their past and future presence on the “A” team, four types of players emerge: 1) Transitory – players who were on the “A” team before and return to the “A” team after a year on the "B" team; 2) Developmental – players who are later on the “A” team, 3) Veteran – players who had played on the “A” team in the past, but never return to first team minutes, and 4) Nevers – players who never do make the “A” team.

Each of these types play a role on a balanced “B” team. Transitory players are often working their way back from injury, a poor run of play, or being out of favor with management. These are often players in their prime like Cherundolo in 2007, Bocanegra in 2006, and probably Feilhaber in 2009. Developmental players are often not quite ready for the spotlight, but seasoning and exposure will give them a better chance to develop into top players. Veterans, used correctly, bring an experienced presence to the “B” team when they need it, like Cobi Jones did at the 2002 World Cup and DaMarcus Beasley could in 2010. And though the Nevers may seem on the surface to have no useful place, names like John O’Brien and Ben Olsen show up on this list, with Jay DeMerit likely joining them as great contributors to the USMNT success who never logged first team minutes.

“B” Team Breakdowns by Cycle


1999-20022003-20062007-2010
Transitional664*
Developmental7611*
Veteran1098
Never5117
* For the current cycle, when necessary to categorize the players, I have made projections as to who will eventually become or return as an "A" team member. These players include: Adu, Davies, Edu, Feilhaber, Guzan, Holden, Marshall, Pearce, and Rogers.


By looking at the individuals who fell into these categories during the past three world cup cycles, patterns begin to appear. Each of the three cycles shows a different primary type. The 1999-2002 “B” teams were dominated by experienced players either winding down their careers as veterans or fighting through injuries keeping them out of their place on the “A” team. Notable players on these “B” teams were Pope, Reyna, Stewart, McBride, Keller, Moore, Balboa, Harkes, Friedel, Jones, and Regis. With a lack of club seasoned youth in the talent pool at that time, these players filled minutes with their experience.

The 2003-2006 cycle, though it can lay claim to one of the most successful “A” team runs of form ever for the US, was a lost cycle from the “B” team prospective. It was an era dominated by Nevers and Veterans. None of these players carry the weight of Reyna or Harkes and few compare favorably to Moore, Jones, or Regis. What may be most notable about this class is the failure of the developmental players, with the exception of Howard, to become anything more than a role player for the national team. The other five developmental players during that cycle were, Cory Gibbs, Steve Ralston, Josh Wolff, Eddie Johnson, and Brian Ching. In four years, those were the only six players to first put in “B” team minutes and later play for the “A” team.

The current cycle has been dominated by youth and the average age of the “B” team shows it:

“B” Team Average Age

CycleFirst YearSecond YearThird YearWorld Cup
1999-200227.827.026.428.4
2003-200627.527.528.529.5
2007-201027.625.925.7


The average in the first year of the cycle has been remarkably consistent. The second and third years however were drastically different, with this current cycle being over a year young than the average of the previous two cycles. In the World Cup year of the cycle, the average has increased substantially. This increase is again likely this year as few new young players will be brought into the pool for the World Cup squad. Natural aging and the seemingly ubiquitous inclusion of a veteran or two for the World Cup nearly guarantee this. However, the youth have been well served in this current cycle. A look at the “B” team players under the age of 25 shows the increase in activity compared to the previous two cycles.

“B” Team Under 25 Players

1999-20022003-20062007-2010
Ben Olsen
Jovan Kirovski
John O’Brien
DaMarcus Beasley
Josh Wolff
Carlos Bocanegra
Steve Cherundolo
Cory Gibbs
Tim Howard
Conor Casey
Eddie Johnson
Bobby Convey
Sacha Kljestan
Justin Mapp
Heath Pearce
Jonathon Spector
Freddy Adu
Jozy Altidore
Maurice Edu
Brad Guzan
Drew Moor
Charlie Davies
Benny Feilhaber
Stuart Holden
Robbie Rogers


While not every player under 25 on the “B” team goes on the make the “A” team or land a big transfer, there is a strong history for these players making big jumps soon after seeing major minutes for the national team:

PlayerFirst “B”
Team Year
MovedFromTo
Carlos Bocanegra2002Jan 2004Chicago FireFulham
Cory Gibbs2003Jan 2005Dallas BurnFeyenoord
Tim Howard2003Sum 2003NY MetrostarsManchester United
Bobby Convey2003*Jul 2004DC UnitedReading
Benny Feilhaber2007*Aug 2007Hamburger SVDerby County
Jozy Altidore2008Jun 2008NY Red BullsVillareal
Maurice Edu2008Aug 2008Toronto FCRangers
Brad Guzan2008Jul 2008Chivas USAAston Villa
Charlie Davies2009Jul 2009HammarbySochaux
Stuart Holden20092010Houston Dynamo???
* Played with the "A" team before playing with the "B" team. First year represents the year with the "A" team.


It is apparent that Bob Bradley has made an effort to get younger players substantial experience with the national team. Whether the increase in transfer frequency is a product of Bradley use of the “B” team to showcase younger members of the US player pool is impossible to know. Has Bradley pushed too far with the youth, depriving the team of veteran leadership from the bench? If you had to choose between Torres and Beasley for the last place on the flight to South Africa, which way would you lean? With the influx of so many young, talented players, will next cycle be dominated by transitional players who bounce from "A" team to "B" team from year to year? Should developing future "A" team players always be the priority of the "B" team?


Formula XI

Thursday, January 07, 2010 | View Comments
Maths Class


I, Jason Kuenle (not to be confused with the Jason that started this whole blog), like numbers, statistics, and logic. Often my posts here reflect that. But because these posts have elicited such comments as “this is a discussion that is most likely taking place by men with pipes and smoking jackets sitting in antique leather chairs”, “I'm not convinced that you can determine too much from stats in soccer”, and “my head hurts now”, in the future I will designate my posts of this type by starting the title “Formula XI”. They will also be tagged as “FXI”. I have gone back and tagged a couple of my previous posts that fit into this category as FXI in case any newer readers are interested in the backfile. If you have numerophobia, if you are subject to migraines, if you don’t like the style of these posts, you will now be able to avoid them.

The goal of Formula XI posts is to provide in-depth logical analysis of US Soccer, particularly for the national team. I am not anti-passion; when the US scored its first goal against Spain this summer, I literally fell out of the booth I was sitting in at the Sweetwater in St. Paul, Minnesota due to excitement (and that literally does mean literally). But, so many of the arguments on soccer websites and blogs has become so based on blind passion and opinions that there seems little room for the analysis and reasoned debate that facilitates learning. While a passionate fan base is essential for the progress of the sport here in the US, I enjoy my debates cloaked in Hugh Hefner jackets, wafting a glass of port, and engaging in the more genteel tone that commenters on Match Fit USA have generally maintained.

Perhaps there is a bit of sarcasm about the jackets and the port, but there is none surrounding my observation of the readers of this blog. I have learned a substantial amount about US soccer from the contributors and commenters on this blog. I hope that Formula XI posts can be tool for analyzing the game we love and a place to further all of our knowledge about the game in the US. I hope to learn as much from reading comments and engaging in subsequent discussion as I do from researching and writing these posts.

The first official post under this title will be up in a few hours. If you have ideas for research-based questions or analysis that you would like explored, drop a note in the comments below or shoot me an e-mail at secondhalfsub@gmail.com. Due to constraints on my time and my own research interests, I can’t promise you that I’ll be able to write on every subject, but if you’ve got a great idea, I’ll be happy to do some research and credit you with the idea. The time that it takes me to write these posts from start to finish means that there will be no schedule for when they are done, but as long as there are few if any other writers taking on these types of projects, I’ll be trying my best to fill the void.


The Central Midfield: A Hypothesis

Tuesday, September 29, 2009 | View Comments
Ricardo Clark


A few days ago, I posted some numbers about the US central midfield and the effects of individuals and midfield partnerships on the number of goals scored and conceded. Two general trends emerged from the analysis; that replacing more defensive players with more offensive players increases the number of goals conceded and that replacing defensive players with more offensive players does not increase the number of goals scored. While there are sample size and control issues in the analysis, and I fully concede that the results are not significant to a level of confidence that I would make the decision of who should play and who should sit on them, they raise interesting questions. Why is the US so reliant on a defensive midfielder? Why does playing an offensive midfielder not result in more goals for the US?

The Defensive Mid

Fair warning: the hypothesis laid out here is supported with statistics that are not statistically significant within scientific confidence levels (the sample sizes are too small and I’m too lazy to figure out the confidence intervals), nor are they properly controlled to specifically isolate the variable, but they are interesting and pass the smell test.

That the addition of more defensive-minded midfielders lowers the goals conceded rate should surprise no one. If that rate went up, why would you ever play with one? However, more than just a general effect on the game, who the US has conceded goals to when a DM is on versus not on the field is telling of what they bring to the table. During its 13 “A” games, the US has conceded 18 goals; 8 to strikers, 7 to midfielders, 3 to defenders. Below is a table that looks at the positions being played by those who scored against the US broken out by the presence of at least one defensive mid.

MinutesGCFWCMLMCBRB
DM739742010
No DM4701143202


For the US, the presence of a defensive mid seems to have a twofold effect. First, it has a direct effect of reducing strikes directly from central midfielders for .6 goals/GE to .25 goals/GE. Second, but more important number wise, it seems to have an indirect effect at reducing strikes from the rest of the pitch. LMs and RBs should be the primary defensive responsibilities of our RBs and LMs respectively. The two side back positions and Donovan while playing left mid are the most likely candidates for pinching in to fill central midfield gaps that occur when a defensive midfielder is not in the lineup. This means that their primary defensive responsibilities are left unattended resulting in a probable increase in opportunities for opposing side midfielders and backs.

Based on his playing time and style it should be no surprise that Michael Bradley was the most defensive-minded midfielder on the pitch for all 470 minutes when there has been no true DM. The results are as much an indictment of his inability to shut down the midfield as it is a testament to the ability of Clark and Mastroeni to do so. Until the unlikely event of playing a meaningful game without Bradley or a DM in the lineup, it is impossible to know if the US needs to play with a DM or just that whenever Bradley plays, he must be paired with a defensive midfielder.

The Lost Talent

That offensive minded midfielders do not contribute more to the goal scoring is more of a conundrum. Logically, more possession be it through dribbling or passing should lead to more goal scoring opportunities. In the interest of entertainment and mental gymnastics, I am going to take a shot at a theory that is definitely still a work in progress. In fact, I am not sure that I believe what is written below, but it is the best I could come up with and I hope it stimulates a good conversation in the comments...

What if the current US strikers are not good at playing in a possession based offense? The list of true strikers that the US has used in “A” team matches is Altidore, Casey, Ching, and Davies. Here’s what I would judge as each respective player’s greatest strength: strength to out physical a defender, heading, holding up the ball, and speed.

Scoring in a counter attacking and set piece offense relies on these traits. The ability to shield off a defender when both are racing for a pass up the wing loses its effectiveness when playing a half-field offense. The ability to head in a cross is less necessary when the attack is built through the middle as opposed to the wings. Pulling down a long ball and holding it for an on-rushing midfielder loses its effectiveness when there are no long balls. The ability to blow past a defender on a release loses its effectiveness when the defense is already settled in the box. To be an effective striker in a possession offense the most important traits are not strength, speed, or height, but are positioning, timing, and vision. While I’m confident that Altidore and Davies will develop this, their national team and club playing experience until now (primarily New York Red Bulls and Hammarby) has not made this an area of expertise.

Someone who does have those abilities is Clint Dempsey. With his questionable looking performances on the right side and the emergence of Feilhaber and Holden as viable replacements, a move up top for Dempsey could be a step in the right direction for building a possession side. Kenny Cooper is another who jumps to mind as having better than average positioning. What has kept him off the team or on the bench is that he is not as quick, is not great with his head, and does not use his strength to take on defenders like the other strikers above who are more suited to a counter attacking style. Either or both of these players would likely raise the effectiveness of Torres and Feilhaber. The question remains is the benefit to Torres’ and Feilhaber’s style outweighed by the loss of effectiveness in the counterattacking game?

Putting together a best XI is not about picking the best eleven players at their respective positions. It is about choosing players with complementary skills. The strikers that play now are chosen because their skills mesh with a back line that has a propensity to panic against high line pressure and serve long balls down the field. They mesh with the apparent need (see above) to play a defensive midfielder whose passing ability does not fit into a possession scheme. They mesh with Donovan’s ability to place perfect outlet passes on the counter. When Feilhaber or Torres take the field, none of these things change. Counterattack and set piece opportunities come at the same rate. I would argue that this is why the goal scoring rate with them in does not change. Perhaps, it is a case where their talents are wasted on the team.

For you conspiracy theorists out there, if this analysis is correct, then using a counter attacking style makes it a lot easier to justify playing a central midfielder who has a knack for finding the back of the net on the counterattack and set pieces himself instead of creating for others, i.e. Michael Bradley.

Finally, if you want to see a more possession style played by the US and like one of the guys currently sitting on the bench or not making the team, I propose this as the most viable (Bob might actually consider playing it the next time Michael gets suspended) option:

GK: Howard
Def: Spector, Onyewu, Bocanegra, Pearce
Mid: Feilhaber/Holden, Feilhaber/Torres, Clark, Donovan
FW: Dempsey, Altidore

Or you can just wait a couple years until Jozy’s and Charlie’s position, timing, and vision are much better and the US can use either of these attacking styles effectively.

There. I’ve given you all some talking points. Argue below.


Evidence of a Central Midfield

Thursday, September 24, 2009 | View Comments
USA v Mexico, FIFA World Cup Qualifier


Essien, Pirlo, Mascherano, Xavi, Iniesta, Kaka. These guys can deliver devastating tackles, brilliant balls over the top, immense pitch coverage, fantastic one-touch passing, penetrating runs, and jaw dropping dribbling. As diverse as their talents are, they all line up in the center midfield and create the backbones of their teams. They are charged either with being the engine for their team or shutting down the other team’s engine. Only in the central midfield can you find this diversity of talents and roles. While the quality of those above midfielders may not be available in the US talent pool, that level of diversity certainly is.

Eight players have staffed the US central midfield in the 13 “A” team games that have been played in 2009. While Mastroeni and Beckerman have played in games with the US “A” team, their time with the MNT is likely coming to a close; Mastroeni because of age and Beckerman because of better options. The rest of the US talent pool in the center of the pitch is young but experienced for their age. Bradley, Clark, Edu, Feilhaber, Kljestan, and Torres average only 23.3 years old but 20.7 caps. They are still developing, but have shown the brilliance of their potential and the mistakes of their youth.

I wanted to see how the US plays with these players on the field in a more concrete way than just by my own feelings about them. Similar to the hockey plus-minus stat, the table below shows each players minutes, the goals scored while they have been in the central midfield, the goals conceded while they have been in the central midfield, each of those numbers per game equivalent (where game equivalent equals 93 minutes with the extra three added for average second half stoppage time), and the difference between the two.

MinGSGCGS/GEGC/GE+/-
Bradley101618141.61.30.3
Clark5631252.00.81.2
Feilhaber310671.82.1-0.3
Kljestan288260.61.9-1.3
Mastroeni201431.91.40.5
Torres66232.84.2-1.4
Beckerman10000.00.00.0
Edu1000.00.00.0


But that’s only part of the story, because these are midfield pairings, who you are paired with matters too. Here’s where the subjectivity of this comes in. I am going to classify the central midfielders into three groups: more useful when the opponent has the ball (defensive), more useful when the US has the ball (offensive), equally useful (balanced). I classify Clark, Mastroeni, Beckerman, and Edu as defensive; Feilhaber and Torres as offensive; and Bradley and Kljestan as balanced. Below is a chart with all the different combinations that the US has played this year.

MinGSGCGS/GEGC/GE+/-
Def-Bal4941122.10.41.7
Bal-Bal252250.71.8-1.1
Bal-Off210562.22.7-0.5
Def-Off121322.31.50.8
Def-Bal-Off45020.04.0-4.0
Def-Def45112.02.00.0
Def-Bal-Bal18000.00.00.0
Bal14000.00.00.0
Def7000.00.00.0


The two tables show one thing very strongly; Sacha Kljestan has been terrible for the US this year. While Torres’ defensive numbers are not great, the statistical analysis there suffers from an overall lack of minutes and being skewed by the three man center midfield disaster that was the Costa Rica first half. Kljestan has no such excuses. His removal from a lineup has generally been followed by a US goal (Mexico) or two (at El Salvador), while his insertion is generally followed by an opposition goal (Italy and Brazil). Because of terrible play when Kljestan is in the lineup, the Bal-Bal combination has suffered greatly. The other three combinations that the US has frequently used shows a stable trend:

MinGSGCGS/GEGC/GE+/-
Def-Bal4941122.10.41.7
Def-Off121322.31.50.8
Bal-Off210562.22.7-0.5


As the more defensive players are replaced by more offensive players, the goals conceded rises; however, the number of goals scored does not significantly change. An analysis of the numbers separated from player bias, stylistic leanings, and form arguments shows that the US should start its matches with a pairing of a defensive midfielder and a balanced center midfielder with an attacking option off the bench to replace the balanced player when a late goal is needed. In that case, there is a slightly higher likelihood of scoring without selling out the back against the counter. In games where an Off has replaced a Def by red card (Italy when Feilhaber moved to the middle after Clark was tossed) or substitution (Mexico, again Feilhaber for Clark) the US has allowed 3 goals and scored 1 in 74 minutes; whereas a Off has never been brought in for a Bal.

Until Edu returns, Clark is the only top defensive midfielder the US has. Likewise until Jones is eligible and healed, Bradley is the only good option as a balanced midfielder. While their play might be ugly, the objective results of their pairing is hard to argue with. In 370 minutes with them on the pitch together, the US has scored eight goals and conceded one or 2.0 goals scored per game and 0.3 goals conceded per game.

Central midfield pairings can be defensive, possessive, creative, attacking, or some combination of these things. The goal should be to find the pairing that gives a team the best chance to win. For whatever reason, the US bleeds goals when anything nearing an offensively minded midfield is played, and those more offensively minded players have not contributed to the attack enough to offset that loss. When healthy and not suspended, objectively the pairing of Clark and Bradley puts the US in the best position to win. That being said, the statistics make a strong case for looking at the pairing of Torres and Clark; as the strike rate with Torres on the field is the highest of any midfielder and Clark’s low concede rate may balance that midfield nicely.



    Sponsored

    Archive

    Legal


    Privacy Policy