In their time with the US, Landon Donovan and Clint Dempsey displayed impressive attacking versatility.  Donovan played virtually everywhere in attack in his US career, spending most of his time as a withdrawn striker or outside midfielder.  Dempsey started as a wide midfielder and migrated forward later in his US career.  On a 23-man tournament roster, both players gave the US strong options at two or three attacking roles.

It’s unlikely Dempsey has played his last game for the US, but his time is winding down.  The onus is on Jurgen Klinsmann to find the right combination of personnel and formation to energize the future US attack.  Let’s look at some of the options at his disposal.  We’ll assume a base of four defenders, since we always play with a back four (the January 2015 friendly against Chile a recent exception.)  I reviewed the various formations of Klinsmann’s tenure, with a focus on games without Dempsey and Donovan.  This should yield insight into potential approaches without these two players, particularly because Dempsey’s natural inclination to drift between the lines added an additional band to any formation.

Adapting a 4-4-2 with player selection

When Dempsey wasn’t available in 2015, Klinsmann demonstrated a preference for two-striker formations, often in a classic 4-4-2.  He utilized player selection to alter the aggressiveness of the team within the same shape.  Here’s an example of a standard 4-4-2 against Costa Rica: Jozy Altidore and Gyasi Zardes as the two strikers; DeAndre Yedlin and Brek Shea as two true outside midfielders; and Jermaine Jones and Danny Williams as the two central midfielders.


Klinsmann gave this same shape a defensive bent in the recent qualifier at Trinidad & Tobago.  By placing four conservative defenders behind a 4-man midfield that included Yedlin and Fabian Johnson, Klinsmann emphasized a defensive posture (the 0-0 final score was a fair outcome.)


Conversely, the home friendly against Peru showed an offensive interpretation.  Zardes slid out wide with Yedlin on the opposite flank, and Alejandro Bedoya joined Jones in central midfield (Michael Orozco just challenged a ball, explaining his position in front of Jones.)


Deviations from the 4-4-2

Over the years, Klinsmann employed other 4-man midfield variants.  Most notable is the diamond midfield (analyzed here by the incomparable Michael Cox), implemented in the friendlies before the 2014 World Cup.  The benefits of a diamond are a clear responsibility in central midfield roles and less emphasis on wing play from the midfield (as fullbacks help provide width.)  The diamond reappeared in the first half against Germany this past June.  Juan Agudelo and Aron Johannsson started up top, tasked with lateral movement.  Michael Bradley played at the top of the diamond with Williams as the base, and Zardes and Mix Diskerud pinched-in on the right and left.


A drawback of a diamond is the danger of an opponent overloading space around the lone defensive midfielder – this happened in the first half against Turkey in 2014.  Another deficiency of a narrow 4-4-2 (diamond or otherwise) is congestion the middle of the field, making it easier for an opponent to defend.

One option to provide more width in midfield is the 4-1-3-2 used against Colombia.  Rubio Rubin joined Altidore at striker, Johnson played alongside Diskerud and Bedoya midfield, and Beckerman played a defensive midfield role.


We see how much player selection and tactical direction matter across these 4-man midfield options.  In the above photo, the 4-1-3-2 seems like a subtle distinction from the diamond; Bedoya and Johnson could easily squeeze in, push Diskerud slightly higher up, and convert the shape of the midfield.  However, this would change the offensive and defensive responsibilities of the two strikers and the two fullbacks, risking the team’s cohesion.

The 4-3-3, and other adventures with a lone striker

A more dramatic tactical shift is to a single striker system.  Essentially, the bargain is one fewer striker for numerical superiority elsewhere, usually in midfield.  The lone striker system manifests itself most often in a 4-3-3 or 4-2-3-1 (Here, tactical patron saint Jonathan Wilson details the evolution of 4-4-2 to 4-2-3-1.)

In the early days of his tenure, Klinsmann persisted on three central midfielders, even with a Mourinho-inspired trivote.  While he wasn’t the first manager to try this – Bob Bradley played Maurice Edu ahead of Bradley and Jones against Argentina in 2011 – Klinsmann has retained this idea the longest.

The US started with a 4-3-3 in its June 2015 friendly against Holland.  Johannsson played the central role with Zardes on the left and Johnson on the right.  Bradley and Alfredo Morales played central midfield, with Beckerman holding. This shape morphed into a 4-5-1 in defense, as seen below just after Morales pressured the ball.


Late in the second half of this game, the US kept a lone striker, but changed shape dramatically.  Jordan Morris led the line, but Bobby Wood played a hybrid left-forward position similar to how Robinho played with Brazil.  Diskerud played a left-central midfield role, but higher up the field than Bradley.  Yedlin played on the right side of midfield, but in defense, lined-up with Bradley and Williams.  Here is the team in transition just before Wood’s winning goal.


This general structure only lasted 10 minutes, but it was an interesting experiment.  Klinsmann has no qualms about playing a traditional striker in a wide attacking role, and Wood’s use of space created a fluid attack, particularly on the counter.

A rethink of the 4-3-3 appeared against the Czech Republic in 2014.  Here, Julian Green and Joe Gyau flanked Altidore in a classic front three.  Bedoya and Joe Corona played ahead of Diskerud in the center, while Diskerud operated as deep-lying playmaker (though more mobile than his future teammate).  In the picture below, Altidore and Bedoya just swapped roles in the build-up to Gyau’s shot.


Playing with a 3-man frontline enabled the US to press high and maintain its desired position.  The midfielders exchanged well in a strong 1st half performance, dictating play and enabling the US to grab a goal (Bedoya, off the Diskerud rebound) and out-shoot the Czech Republic.  As you can see from the above image, the formation also enabled Timmy Chandler (bottom left corner) and Johnson (top right) to join the attack.

What’s interesting is that despite the popularity of the 4-2-3-1 , Klinsmann largely abandoned the formation without Dempsey, though it briefly reappeared against Brazil to offer a reminder of why player selection matters.

Who does the US build around?

While it’s near-impossible to predict a complete World Cup roster years in advance, a national team still needs core players to form an effective tactical system.  Once in place, a manager can approach friendlies with the goal to see which supporting players complement the core, and the various tactics that maximize the team’s performance.

To aid in this evaluation, Jörg Seidel of Goalimpact generously provided his quantitative analysis of key players in the US attack.  What I love about the Goalimpact methodology is its top-down approach: the only on-field metric that matters is a player’s goal differential.  From there, the algorithm accounts for a player’s age (with an assumed peak at age 26), the league, teammates, opponents, and the prior history of the player to calculate a Goalimpact value (a more detailed explanation is here.)

Goalimpact scores are not evenly distributed.  To provide context, players with the following scores are essentially good enough for:

  • 100: A Bundesliga team that will avoid relegation.
  • 110: An average Bundesliga team
  • 130: A Bundesliga Champions League qualifier
  • 150: One of the top 200 or so players in the world

Let’s start with Donovan and Dempsey to see their long shadows quantified.  The solid red line is the player’s Goalimpact score, and the light red line is the expectation line (click here for a thorough overview of how to read the chart.)


Donovan ascended rapidly during his time in San Jose.  With LA, Donovan’s score hovered around 120 from mid-2005 to the end of 2009 – his actual score and expected score converged during the prime age of 26.  However, Donovan defied aging expectations, and peaked just before turning 30 during his 2012 loan to Everton.


Dempsey also had a late peak.  His move to relegation-battling Fulham reflected his MLS potential.  Dempsey’s improvement from 2009-2012 mirrored that of Fulham, and he registered a high score of 130 with Tottenham at the of 2012.

Transitioning to players around whom Klinsmann could build a future attack, here is Jozy Altidore’s chart:


Altidore had an incredibly high initial projection – among the top-rated talents in the world.  As covered in part 1, Altidore had very sporadic club performances from 2008 to 2011; that time period lowered his projection, but a move to AZ (thanks Earnie!) changed his career trajectory.  Altidore is just now entering his prime, and is the top-rated striker in the US national team pool with a Goalimpact score of 131.

The next tier features strikers bunched together closely: Wood and Zardes both have a Goalimpact score of 110, and Johannsson is at 107.  To cover other younger strikers discussed in part 1,  Agudelo is at 93 and Wooten is at 87.  Given that none of the strikers offer themselves as a superior option to Altidore, the ability to play wide – either Zardes dropping to midfield or Wood/Johannsson running the channels – could be a key distinguishing factor.

Moving deeper in the attack, one player who has set himself apart from others is Fabian Johnson.


Johnson provided quality for the US in the past, but his past two years at Hoffenheim and Gladbach featured revelatory performances, and raised his Goalimpact score to 123.  Johnson also provides critical versatility; he will be on the field somewhere for the US, the only question is where.  Playing as a wing forward in a 4-3-3 is likely too high up the field, though he should be the preferred left midfield option in a 4-man midfield configuration.  Similar to Altidore and the rest of the strikers, Goalimpact sees Johnson as a clear step above other outside options like Shea (107), Bedoya (105), or Zusi (104).  A wildcard is Julian Green, a player who, despite the consternation, has a Goalimpact score of 107 with a projection around 130 at age 26.


For the US, it appears that all tactical decisions cascade from striker back down the field.  Simply put, what initial system maximizes the forward line?  Which players make that system effective?  Also, which players have enough uniqueness or flexibility to offer an alternate plan if the initial system doesn’t work?

That system starts with Altidore up top, and my personal preference is to include Wood.  Wood complements Altidore well: the combination provides the US with strength, pace, mobility, and finishing.  Additionally, both strikers will be near their prime at the 2018 World Cup, minimizing the risk of a sudden decline.

The rest of the attack follows accordingly based on the strength and style of the opponent, and the strengths of the rest of the players.  Against a CONCACAF opponent at home, breaking a bunker matters, so more technical players like Diskerud, Johnson, Bradley, and Darlington Nagbe could feature in midfield.  If we’re looking to counter-attack at Copa America Centenario,  Yedlin might man the right flank while Williams offers more protection at holding mid.

A premium must be placed on connectivity in the attack, and unless a surprise attacking winger or two emerge in the next few years, a two-striker formation best ensures that connection.

Click here for Part 1 –  a focus on the club careers of US attackers