On B Squads in International Soccer
Author’s Note: This essay has been modified from it’s original published version to correct grammatical and stylistic mistakes, as well as one factual error. Andres Guardado did not participate in the 2021 Gold Cup with Mexico.
The CONCACAF Gold Cup is a biennial tournament to determine the best men’s national soccer team in North America. In the tournament’s history, only three teams have ever won: 11 times by Mexico, 7 times by the United States, and once by Canada. Because the historical record has so consitently favored a Mexican-American duopoly, both nations have routinely competed in the Gold Cup with rosters that were not as strong as they could have been, secure in the belief that either nation could still threaten to win the tournament, even at reduced strength. In some cases, this was done in order to give star players a rest. In others, it was done because the national team was competing in multiple tournaments simultaneously. Sometimes, players have simply been injured.
All three factors played into the most recent Gold Cup, which was won by the United States with a 1–0 victory over Mexico in the final. Having beaten Mexico in the final of the CONCACAF Nations League the previous month, the US elected to allow the majority of the players who participated in that competition, including most of its European-based stars (Christian Pulisic, Giovanni Reyna, Sergiño Dest, etc.), to take a vacation and recover in preparation for World Cup qualifying in September. The US roster therefore consisted primarily of players from MLS, the US and Canadian domestic league, most of whom had little to no prior experience with the national team or were otherwise believed to be on the margins of the player pool. Mexico meanwhile offered vacations to no one, but were forced to bring a reduced strength squad by a combination of injuries to key players (striker Raul Jimenez before the tournament and winger Hirving “Chucky” Lozano in the first game) and the need to send a team to compete in the Olympics in Tokyo (Guillermo Ochoa and Carlos Rodriguez among others).
Because of this situation, when the US eventually defeated Mexico in a final for the second time this year, American fans were inclined to give their team especial credit for securing victory against a strong opponent without relying on any of its usual stars. The Mexicans, for their part, consoled themselves that, had their top guns been available, the result likely (to them) would have been otherwise. On both sides of the border, this phenomenon is usually referred to by saying that either or both sides brought their B squad (or C squad, depending on the opinion of the commentator) as opposed to a full strength A squad. However, in the aftermath of this most recent Gold Cup, it appeared to me that the two fanbases were not consistent with one another in their use of the terms B squad and C squad. In my observation, biased as it may be, American fans tended to categorize their Gold Cup roster as a B squad, whereas at least some Mexican fans tended to categorize their own team as a C squad. In spite of this, in my opinion, the Mexican squad was closer to full strength (although definitely not at full strength) than their American counterparts.
It is not surprising that we Americans should disagree with our southern neighbors on this point; neither side can be expected to be unbiased. More to the point, there is not any objective standard for how many players must be unavailable for a roster to be demoted from A squad to B squad, or from B squad to C squad, nor how important those players must be. Any attempt to define such a standard would be arbitrary. However, I believe that such a standard is nonetheless a worthwhile contribution to the national team discourse because it would reduce the confusion that might result when two observers refer to a roster with different letter grades while implying the same degree of relative quality or the same letter grade while implying different degrees of relative quality. To this end, I have devised such a standard. This essay outlines a procedure for assigning letter grades to national team rosters based on the grader’s personal position-by-position depth chart for the team in question. As such, it does not eliminate the subjectivity involved in comparing the merits of one player with another, but it does eliminate the subjectivity involved in applying those granular judgements to a roster as a whole. Once the procedure has been outlined in full, I will attempt to address various criticisms that may be leveled at this approach.
In order to demonstrate my approach, I will be assigning letter grades to the rosters for the US and Mexico in the 2021 CONCACAF Gold Cup as well as the recently announce US roster for the first three games of World Cup qualifying. In doing so, I am obligated to give my opinions on who is better than whom at various positions for both teams. Readers may disagree with my evaluations of players, particularly with respect to the Mexican team, which I follow less closely. However, the point of this exercise is not for me to advocate my (weakly informed) opinions of individual players. It is to provide a framework for translating those opinions into an opinion on entire rosters. As such, I hope any atomic level disagreements can be set aside in order to evaluate the system on its own merits.
The procedure begins by defining a position for every player on the field at a given time. This will depend on the formation that the team prefers to play. The US typically plays in a 4–3–3, so for them, the positions will be goalkeeper (GK), left back (LB), left center back (LCB), right center back (RCB), right back (RB), defensive midfielder (DM), central midfielder (CM), attacking midfielder (AM), left wing (LW), right wing (RW), and striker (ST). Attentive fans will note that the US does not generally play with a true AM, instead favoring two CMs who share attacking and defensive responsibilities. However, even in the US system, those responsibilties are not shared equally, and as a fan, I want the US to lean into the true AM role, which I think would suit Giovanni Reyna to a T. In any case, my method breaks down if I cannot distinguish the position of two players on the field at the same time; if you dislike the AM nomenclature, feel free to use a LCM and RCM version when you apply it yourself. Meanwhile, Mexico sometimes plays in a 4–3–3 and sometimes in a 3–4–3. For variety, I will follow the latter formation and name their positions GK, LCB, sweeper (SW), RCB, left wing back (LWB), right wing back (RWB), DM, CM, LW, RW, and ST.
Once positions have been determined, next we pick the best possible 11 players that the team could field. In my opinion, for the US, that means Matt Turner at GK, Antonee Robinson at LB, John Brooks at LCB, Chris Richards at RCB, Sergiño Dest at RB, Tyler Adams at DM, Weston McKennie at CM, Giovanni Reyna at AM, Christian Pulisic at LW, Tim Weah at RW, and Josh Sargent at ST. None of these players were brought to the Gold Cup, but all of them, with the exception of Richards, have been called into World Cup qualifying. I am less confident in my judgement of Mexican players, but my best guess at their best 11 would be Guillermo Ochoa at GK, Nestor Araujo at LCB, Edson Alvarez at SW, Hector Moreno at RCB, Gerardo Arteaga at LWB, Luis Rodriguez at RWB, Hector Herrera at DM, Andres Guardado at CM, Hirving Lozano at LW, Jesus Corona at RW, and Raul Jimenez at ST. Of these, Araujo, Alvarez, Moreno, Rodriguez, Herrera, Lozano, and Corona came to the Gold Cup, but Lozano was injured in the first game against Guatemala and was unavailable in the final against the US. These players occupy the first spots on the depth chart for the position to which they have been assigned. Once assigned, they cannot be listed on the depth chart for another position, even if they can play multiple positions (Edson Alvarez is equally good as a SW and DM for example.) For reasons which will become apparent, this is necessary so that rosters are not penalized for having versatile players in the pool.
Once the first 11 is selected, you pick the second strongest 11 from among all the players who have not yet been named. Then the third, then the fourth, and so on, until all the players that were named to the actual roster are on the depth chart somewhere. Then you trim the depth chart for each position so that the last player on the list for that position is on the roster. Depth charts for different positions will not, in general, have the same number of players, since the worst player on the roster at each position will not necessarily be ranked the same as the worst player in another position. My depth chart for the US based on the Gold Cup roster is below with selected players highlighted in green.
Most players on this roster are the 2nd, 3rd, or 4th best players available, with Jonathan Lewis and Donovan Pines representing notable outliers. I rated Lewis significantly higher before his atrocious performance against Haiti, but Pines inclusion confused me at the time and baffles me afterwards. Although the depth chart above shows 10 players ranked between him and the tournament’s leading RCB Miles Robinson, in reality, I would also have preferred several players from the LCB depth chart over Pines. Contrast this with the US World Cup qualifying roster below.
Most of these players are either the 1st or 2nd choice at their position, with the exceptions mostly attributable to injuries or unstable club situations. A few differences are also probably attributable to differences between my opinion of the players and the coach’s. Namely, it is common knowledge by now that Gregg Berhalter rates Sebastian Lletget significantly higher than most fans do, for better or for worse. Both of these may be compared with my results for the Mexican national team at the Gold Cup.
By visual inspection, this looks more similar to the US World Cup qualifying roster than the US Gold Cup roster. The last step in confirming this is to count the total number of names on the depth chart, subtract the number of players on the actual roster and divide this quantity by the size of the roster. This measures the average difference in depth chart position between a player on the roster and the player that would take his spot in a theoretically optimum roster with the same positional composition. For the US Gold Cup roster, this value is roughly 1.4, for the US World Cup qualifying 0.4, and for the Mexico Gold Cup 0.6. This does not change much if we adjust for Lozano’s departure and replacement by Rodolfo Pizarro. If we assume he (Pizarro) is third on the LW depth chart behind Lozano and Diego Lainez, then the team’s index is still 0.6 to within rounding error. In my usage, an A squad corresponds to an index of 0, a B squad corresponds to 1, C squad to 2, and so on. By this standard, the US Gold Cup was a B- squad, the US qualifying is an A- squad, and the Mexico Gold Cup was a B+ squad. However, it is easy to define alternative mappings from index scores to grades. If we accept the idea that the Mexican squad which faced the US in the Gold Cup final was a C squad, then this would imply that an increase in index score of 0.3 is enough to cost an entire letter grade, and the most recent US roster is actually a B or B- team, and the US Gold Cup group was either an F+ squad or off the scale depending on whether you want to include the concept of an E squad or not. I prefer my system because it implies that a B squad is one which relies upon the top backups to step up to the task, which feels right to me. It also implies that losing all the top backups is equally detrimental to losing the entire set of starters which is probably false but in a way which corrects against the general tendency to underrate the contributions of squad players.
Get Elijah Schultz’s stories in your inbox
Join Medium for free to get updates from this writer.
Having finally described my system in full, I will attempt to address some natural questions that might arise.
Q. How does this system factor in roster balance?
A. It doesn’t. My method assumes that the coach has chosen a sane proportion of players for each position on the field. This is not always the case. Note that based on my depth chart that the US Gold Cup team had no LWs. Jonathan Lewis played there in the first game, played like hot garbage, and thereafter the US either played STs there out of position or completely changed formations to a 3–5–2. If I continue to work on this idea, addressing this problem is the thing I’d like to focus on. Applying something like a Gini coefficient to the rank-weighted number of players who can play each position might be able to generate a ‘fragility’ index that could supplement my prior work.
Q. Why can’t players be counted in more than one position’s depth chart? If a player can play multiple positions, then your model will often underestimate the cost of bringing certain players
A. This is a correct observation. I even noted the issue with respect to Donovan Pines earlier. However, if players could go on multiple depth charts, then versatile players who can play multiple positions would hurt their squads index ranking even if they get called in. For example, Sergiño Dest is widely acknowledged as the US’s best RB and LB. If he were on both depth charts, because there’s only one of him in reality, calling him up doesn’t eliminate all the extra names from your depth chart. You can argue that this indicates a fundamental flaw in my approach, but I think the simplicity of the method makes up for the theoretical weakness. One solution might be to say that players who get called up can only appear on one depth chart, but players who do not may appear on arbitrarily many. This would unfortunately increase the amount of time spent formulating the depth chart, since the number of players that need to be considered for each position increases, but I wouldn’t object to this modification if people prefer it.
Q. Some players are more valuable than others, even if all are starters, but your system counts all absences equally. What gives?
A. If I knew how to systematically measure or estimate the value of individual players to their teams, I wouldn’t be posting it for free online. I’d be selling the technique to professional teams. I have vague judgements about which players I’d most prefer to see at a given position. The quality of my subjective data don’t allow me to be more detailed. Also, effects from differential player quality will be most significant when the number of absences is small. If you only have one player missing, even if he is your best player by a wide margin, I don’t think the language of A squads and B squads is the right way of describing what’s going on. I think it is more informative to simply observe that Player X is missing and that’s bad.
Q. You say Player A is better than Player B, but I say Player B is better than Player A. How can I trust anything you say when you hold such terrible opinions?
A. Rest assured I hold far more terrible opinions than you could possibly imagine, but I hope it’s pretty clear that the procedure I’ve described doesn’t assume anything about which players are better or worse. You, the fan executing the algorithm, get to decide that. If someone else uses this method to generate a grade based on rankings you disagree with, then you’re free to argue about those rankings and make your own grade based on what you think. Half of being a soccer fan is arguing over which players are better than each other, and it would be the worst kind of technocratic insensitivity for me to take that away from you. There’s a good chance my Mexico depth chart is wildly off-base. Nothing would thrill me more than for a Mexico fan to take my system and their own depth chart to compute a more accurate grade.
That about covers everything I have to say about this matter. Odds are, the audience for this essay is pretty small, but feel free to reach out if you have a question I didn’t cover or just want to talk USMNT.
