Atlanta United 2019 Season Preview / by Tiotal Football

Point-above-replacement values are explained here. Non-penalty expected goals + expected assists are explained here, and you can see all players’ xG+xA in our interactive expected goals tables. Touch percent is the percentage of total team touches by that player while he is on the field, which can be found in our interactive expected passing tables.

By Tiotal Football (@tiotalfootball)

Join me for a moment in a Lovecraftian horror in which time is an illusion and the events of life as we know it, or the meaningful events at least, (soccer matches) progress not sequentially by the steady consumption of the present as measured in minutes and seconds but by the experience of passes being attempted from open play. In this nightmare, as the fates dictate the average MLS team experiences 900 such passes attempted during a match (the total of both teams over 90+ minutes), and using passes experienced as the unit of account, the average 2018 MLS team spent 46% of its matches with a level score line (27% leading and 27% trailing).

Atlanta United spent the joint lowest MLS share of its match experience with the score level (38%) and by some margin the highest MLS share of its matches leading (45%). This is the privileged experience of an excellent team full of competent coaches and players. In truth, Atlanta were the only team in 2018 for which its “leading share” outstripped its “level share” of passes by any material amount.

It may not surprise you then to hear that within the deadlocked experience of a drawn score line, no team was more likely to shatter the walls of its existence and break the equilibrium by scoring a goal than Atlanta United Football Club, throwing up 29 total goals (the most in MLS) against the second lowest open play passes experienced while the game was drawn. Time and time again, Atlanta United found themselves suddenly up a goal and then effectively adjusted their tactics to reduce risk and strike mercilessly against an overextended opponent.

We’ll work our way forward from defense to attack for the 2018 MLS Champions and try to pick out insights about the coming 2019 season as they arise.


After an abbreviated but excellent shot stopping season in 2017 (where in 14 appearances he allowed nine goals against an expected tally of 14, based on the types and locations of shots faced as well as their placement on the goal mouth), Brad Guzan fell back to earth in 2018, conceding 43 goals against a comparable expected value of 41.

To my eye, Guzan had a mostly good year, with only one or two games where his actions harmed the team’s chances of winning (a red card against SKC, for instance). Further, it seemed to me as though he commanded the box fairly well compared to other keepers in MLS. There are some statistics that bear this out, even if its somewhat perilous to put too much weight in them just yet. First, I scanned a plot of all keeper claims and punches in 2018, and decided to draw a perimeter the width of the six yard box up to the penalty spot and christen it “GK command zone” since that’s where the vast majority of claims and punches are occurring. I’m sure many more qualified persons have done this and with greater thoroughness, but this is my #agile approach to it for the purpose getting through this goalkeeper section.

GK Team Successful Crosses allowed into the command zone per game Successful crosses allowed in the command zone per punch/claim Successful Crosses allowed per punch/claim
Columbus 1.50 1.28 2.53
Philadelphia 1.82 1.32 2.70
Seattle 2.74 1.52 2.46
Atlanta United 1.88 1.73 3.19
Los Angeles FC 2.18 1.76 3.21
Portland 2.65 1.84 3.53
Houston 2.03 1.86 3.49
Toronto 1.71 2.15 5.00
New York 2.15 2.28 4.38
DC United 2.53 2.39 4.50
San Jose 2.76 2.41 4.10
Orlando City 2.03 2.46 4.96
Kansas City 1.32 2.50 5.61
Vancouver 2.76 2.76 4.85
Chicago 2.21 2.78 5.22
New York City FC 1.74 2.81 5.52
Montreal 2.74 2.82 4.58
L.A. Galaxy 2.29 3.00 5.23
New England 2.35 3.08 4.96
FC Dallas 2.38 3.24 5.28
Salt Lake 2.50 3.86 5.82
Minnesota United 2.88 4.08 6.88
Colorado 2.35 5.00 8.94

Next, for each team I added up the successful crosses allowed to end in the GK command zone and divided this number by the number of GK claims and punches for an overall measure of how active a team’s keeper is at commanding the area. The results suggest that in fact, Guzan is in fact one of the four most aggressive keepers in the league at commanding the area, again as measured by the number of successful crosses into the command zone per claim or punch. Zack Steffen leads the pack followed by Andre Blake, Stefan Frei, and Guzan, with LAFC’s Tyler Miller rounding out the top five (and Tim Howard at the very bottom). The list looks similar if you were to use headed shots in the command zone per claim or punch also.

Guzan will return for Atlanta United in 2019. Even if his shot-stopping numbers look more like last year than the outlying 2017, I believe Guzan’s 2019 will be a success if he continues to command the center strip of the penalty area with aggression while only conceding goals at expected rates. And the reason I’m setting the bar for success sort of low for him is the fact that the outfielders in front of him were very, very good at preventing dangerous moments.

Open Play Defending (0.88 goals allowed per game, 3rd in MLS. 0.81 xG allowed per game, 1st in MLS)

As a goalkeeper for Atlanta United, Guzan had to deal with a whole lot less threatening activity than goalkeepers on other MLS teams. The Atlanta outfield players conceded the lowest volume of open play shots inside the penalty area in 2018 (4.3 shots per game vs a league average of 6.0). Unsurprisingly, accompanying this impressive stat was the rate at which they conceded successful passes ending in the penalty area in the first place (eight per match, joint lowest in MLS with Columbus and Kansas City compared to a league average 10).

And when it comes to defending your team’s penalty area, it’s not simply about disrupting the passes that make their way into this zone (although it is, and Atlanta did this well), it’s just as much or more about preventing passes from entering your penalty area in the first place as the graph below shows an obvious correlation.


So if a not insignificant part of preventing your opponents from completing passes into your penalty area is preventing them from attempting passes into your penalty area, how do you do that?

Well, it certainly appears as though most teams who limited their opponents to fewer successful open play pass attempts into the penalty area also possessed more of the ball then their opponents in 2018.

In 2018, Atlanta United found itself on the favorable end of all these defensive trendlines, possessing more of the ball than most teams and conceding even fewer open play passes than their high possession figures would suggest. They also conceded fewer successful passes into the penalty area than the volume of such pass attempts allowed in would suggest, and they conceded even fewer shots in the penalty area than the volume of successful passes allowed into the penalty area would suggest. That is all to say that while Atlanta used possession to successfully limit their opponents’ chances (tired cliché alert), they were also just good defensively when their opponents did have the ball at every node in the defensive “Value Chain” (wasted business degree alert).

And this shows up (at least I hope) in one last defensive scatter plot where I’ve attempted to visualize how effective a team is at limiting its opponents when its opponents have the ball. On the X axis, I plotted the pass attempts conceded in a team’s half divided by the pass attempts conceded which end in a team’s penalty area to try to approximate the frequency with which opponents turn attacking half possessions into dangerous moments against a team (left bad, right good). And on the Y axis, I plotted the pass attempts conceded which end in a team’s penalty area divided by the shots conceded in the penalty area to try to approximate the frequency with which opponents turn dangerous moments into dangerous shots against (top good, bottom bad).


With the defensive effects of possession adjusted for (however poorly), Atlanta shows up as allowing very few dangerous opportunities relative to what they were faced with compared to some other possession-oriented teams like NYCFC, Kansas City, Toronto, and Philadelphia. And the high pressing teams whose overall defensive team metrics look good refract a bit in this lense, with the New York teams and the Revs conceding a high volume of passes into their own box relative to their opponents’ possession, with the Red Bulls happening to successfully snuff out these chances to prevent shots at an above average rate (good center backs?) and the Revs less so. From an Atlanta United perspective, yes they possess the ball. And yes your opponent can’t score when they don’t have it. But also yes, when your opponent does have it, you must find ways to keep it out of your penalty area, and when it does sneak its way in there, you want to prevent a shot, and by these numbers, Atlanta did that better than any other team in MLS in 2018.

And it makes sense to me that this might very well have arisen from the purposeful shift (for about half the season) to fielding three center backs and two wing backs in place of a more traditional two center back and two full back defensive shape. In truth, in games where Atlanta fielded a back four they allowed their opponents to attempt more passes into the penalty area than in games where they played a back three.

To boot, this back three/five also featured a fairly solid midfield three, which in turn ultimately left little room for all of Atlanta’s more celebrated players up front to make it into the starting 11 when it mattered most.

We’ve covered shot volumes and the extent to which Atlanta suppressed its opponents’ shots in 2018, but what about shot quality? Did they by chance allow fewer shots at the expense of allowing much scarier ones (a thing that happens for instance when a team dominates the ball but leaves itself open to counters)? No. Atlanta’s opponents took shots from open play with an average xG value of 0.10 per shot, the second lowest in the league and just fractionally worse than the opponents of Columbus Crew, leaders in this metric. For context, the league average xG value per shot from open play is 0.115 and the worst team by this measure is Orlando City, allowing its opponents an average xG value of 0.13 per shot.

You can see why this might be the case when looking at shot locations around the box. Atlanta United conceded both fewer shots (and impressively, a smaller share of their total shots) inside the six yard box, the dangerous zone extending beyond the six yard box, and the penalty area as a whole than the average MLS team.

In Search of a Defensive Weakness: The Left

That all sounds fairly impenetrable as MLS defenses go. If pressed to point towards a weakness in Atlanta’s defense, I might bring up the defensive left channel (attacks down the opposition attacking team’s right wing). The distribution of successful passes from open play into a team’s own penalty area for an average MLS side in 2018 was 49% from the center, 24% from the defensive right, and 27% from the defensive left. For Atlanta United, who again conceded some of the lowest totals of passes into this area in the league, 49% came from the center, a league low 19% from the defensive right, and a league high 32% from the defensive left. The left side of the team is the one that faced the most injuries in 2018, notably Greg Garza and his immediate backup Mikey Ambrose for half the season, and Chris McCann (who will not return in 2019) filled in more than anyone else here. And to be fair, perhaps there was something going on in front of the left back that leaned ball progressions this way. All in, it’s not great consolation for opposition attacks.

The Defense in 2019: Expect Tweaks

As to how this defensive overview maps to the 2019 season in front of us, Atlanta’s defense was mostly retained, only really losing left back Garza in the offseason – a player who when healthy was one of the league’s best at the position. Many hope he will be replaced by local phenom George Bello (!) but perhaps more realistically to start the season by Brek Shea (!?), who signed in the offseason as a free agent and is a larger human than almost anyone in MLS. Also bolstering the left side, either as depth or at starting left center back if there were to be a reshuffling on the back line is the new arrival of out-of-contract-in-Turkish-second-division Florentin Pogba (?!).

Now, new manager Frank de Boer, despite having set his teams up exclusively in the 4-3-3 for most of his career, appears ready to implement a three center back approach following Atlanta’s success with this down the stretch in 2018 (de Boer experimented with a back three during his brief stints at Inter and Palace). So you’d think that most of what worked well in 2018 should remain intact, and perhaps the 2018 analysis above is quite relevant when we think about 2019 although there’s much to learn about how de Boer’s midfield will function.

However, illuminating quotes from an interview with The Athletic’s Felipe Cardenas and Atlanta United captain Michael Parkhurst revealed that Frank de Boer plans to take a more conservative approach in defense, with fullbacks tucking in more and the team making a more concerted effort to avoid defensive 1v1 duels, having a spare man at the back whenever possible. And this idea is propped up by de Boer’s own statements to Roger Bennett about not wanting to change last year’s attack but feeling like there was plenty of room for improvement in the team’s defensive style.

Having just walked you through an epic lullaby of defensive analysis which gently suggested Atlanta’s 2018 defense was excellent, this sentiment from the newly arriving coach gives me tremendous pause – especially since he appears to be switching the central midfield from a three to a two man setup, which… I don’t know. It just feels like a lot more will be flowing down hill at Atlanta’s (now more conservative) back line.

Is it simply that de Boer is not comfortable managing a team that defends with the Bravado and risk-taking that Tata Martino forced upon his defenders -- that he’s making some tweaks so that he can coach what he knows -- or has he misdiagnosed a problem that didn’t exist?

Midfield Pressure & Possession (55% possession, measured in share of OP passes, 4th in MLS)

Whereas the 2017 Atlanta United was defined firstly by its furious high press, the 2018 iteration wavered between an elite high pressing team when playing a 4-3-3 and a more measured situational pressing team when fielded with a back three. While there are several ways to measure these high press elements I’ll stick with a more traditional definition of High Press Passes per Defensive Action (opponent passes started in opponent half divided by tackles and interceptions in opponent’s half). The chart below summarizes how Atlanta’s pressure morphed over time with the team ultimately settling for a level of high pressure that’s more intense than the average MLS team but much less intense than the 2017 Atlanta United or any year’s NYCFC or Red Bulls.

The impacts of the formation on pressure can be seen in the xPass scores as well. While on average, Atlanta United allowed their opponents just a half a percent higher in pass completion percentage than the xPass model would suggest, in games where they played three center backs, their opponents enjoyed pass completion rates that exceeded the expected modelled rates in their own buildup third (less high pressure from Atlanta) but fell short of the expected pass completion rates in Atlanta’s defensive third (tighter defense from Atlanta).

If one pillar in Tata Martino’s style guide was pressure, the other was possession. And interestingly, when you filter out passes in which one team had a man advantage due to a red card, you can see that while Atlanta’s share of the total passes being completed in its games (one way to measure possession) was the 2nd highest in the league when the team was trailing or level on the scoreboard, it dipped to a near MLS-average 9th when leading. I believe this captures the team’s concerted efforts to reduce risk and strike with pace on the counter attack when leading late in a game.

Note: Despite what some suggested, Atlanta’s shift towards counter play when leading is not an indicator that they were a “bunker and counter” team, despite the fact that they often found themselves leading after mostly dominating possession and scoring first.

If I had to guess, this is one aspect that Frank de Boer aims to change in 2019. I could see him wanting to be on the ball more when leading to “dominate” games (as he says), but “guess” is an appropriate word here.

If you’re a team that wants to have the ball and dominate play, turnovers can ruin your day – specifically because a turnover in buildup can create a decent game-changing opportunity for your opponent, and if they capitalize on such an opportunity, then the state of the game changes such that your otherwise sustainable and dominant plan A might now be inadequate. If we shift our focus for a moment towards on-the-ball turnovers (turnovers coming from poor first touches and being caught in possession), and if we measure a team’s tidiness in terms of pass attempts per turnover, then Atlanta United look like kings in the middle and final thirds, but league average in its own defensive third. And they are slightly worse on the left side of their own defensive third than the right (although this is true for the average MLS team as well). This might be one area where de Boer clamps down. He might want his defensive line to remain courageous, but also to show more caution on the ball than they did in 2018 (i.e. have Leandro Gonzalez Pirez get rid of it from time to time). For completeness, below is the average touch position and touches per turnover for the critical returning players as well as incoming/outgoing and fringe players. We’ll get to this in a bit, but if the manager favors possession over direct attacking, there’s reason to lean towards Ezquiel Barco and away from Tito Villalba. For all of the rumblings in Atlanta about Barco giving the ball away too easily, the numbers suggest otherwise with both Miguel Almiron and Barco similarly above average in terms of ball retention compared to other attacking midfielders.

Open Play Attacking (1.35 goals scored per game, 7th in MLS, 1.5 xG created per game, 1st in MLS)

Atlanta attempted the second most open play shots inside the penalty area per game in MLS last year (7.6/game behind NYC’s 7.9/game vs a league average of 6.0/game).

Atlanta United generated more open play shots than the average MLS team in every area inside the penalty area and found a higher percentage of these penalty area shots in the more dangerous areas (inside the six, the area around the penalty spot) and poetically, despite creating more shots than the average team, they took a slightly below average number of shots from outside the box.

Curiously, while the excessive shot volumes and superior quality of shots from open play landed Atlanta United with the highest open play expected goals created in MLS (51ish expected goals over the 34 game season), this output did not generate league leading goal tallies. In fact, Atlanta ended up converting only 90% (3rd lowest in MLS) of this xG total into 46 open play goals on the year, 7th best in the league. It’s an interesting development after a 2017 in which the Alchemists United of Atlanta converted 40 xGs worth of open play shots into 59(!) goals (149% G/xG). Miguel Almiron is mostly to blame in 2018 for putting so many high quality chances off target in the first half of the year, whereas in 2017 almost everyone lit it up, crushing their expected goals. Let’s just move on.

Josef Martinez was one of only 11 MLS players to start every match in 2018. And that was extremely good for Atlanta United. He scored 21 goals from open play: seven assisted by Julian Gressel, five by Villalba, four by someone named Miguel Almiron. Three goals were unassisted and Andrew Carleton and Darlington Nagbe each set up one. All 21 goals were from inside the penalty area, and seven were from crosses. He scored eight with his head, seven with his weaker left foot, and six with his dominant right. He also introduced the world to a new penalty kick run-up in route to converting eight of nine from the spot. I’m just saying Martinez can do all the striker things, and his 21 open play goals are backed up by 20.4 expected goals worth of shots. For how unbelievable his scoring record has been since joining Atlanta United, at least in 2018 it seemed unnervingly sustainable based on the underlying numbers. Martinez returns in 2019 as do his top aids in Gressel (0.34 assists per 96) and Villalba (0.31 assists per96).

Oh and I almost forgot. Josef Martinez assisted five goals on 4.5 expected assists. In fact, for all the talk of the dynamic duo with Miguel Almiron being the provider for Martinez in his record goal-scoring year, Martinez actually assisted just as many of Almiron’s open play goals as was the reverse.

That reminds me: this player Miguel Almiron. He was immense for Atlanta in 2018 and he’s not returning this year (studying abroad in the UK) and that should be a big deal. The front office was never going to find a perfect like for like that can replace his engine without the ball, the tracking back to break up counters, the timely pick-pockets of the opposition defensive midfielder, and his ability to run right by a defender without even allowing the guy to get close enough to challenge. But from a strictly attacking output/final third perspective, Almiron’s threat to defenders is clear. He completed the most passes into the box per game on the team, and off the ball, he made tireless and sacrificing runs into the box which led to him receiving more passes in the box per game than anyone not named Martinez or Villalba. This makes him not just a traditional “between-the-lines” #10, if those still exist.

And this leads us to perhaps the biggest question for the 2019 Atlanta United attack. If Frank de Boer intends to play a 3-4-3 then he’ll soon face the fact that he’s got to fit four elite or elite potential attackers into three spots. To start, It’s Josef’s team so he’s in. And you can also go ahead and pencil Pity Martinez into one prong of the trident as the expensive Almiron replacement with the immense pedigree. In truth, the playmaker from River Plate does appear to be the player most capable of matching both the incisive passing and the shot volumes of the departed #10. That leaves Barco and Villalba remaining to battle it out for the last spot. Based on who’s been playing the most in preseason it appears as though de Boer is leaning towards Barco. The young DP’s first touch is immaculate and if you scroll back up to the turnovers plot, you might imagine why a possession-oriented manager might also favor the youngster.

I would offer up a caution here. A return glance to the Penalty Area Danger chart will highlight that Barco loves to get on the ball and pass, and despite the objections of many that he slows the game down and passes sideways too much, he did in fact thread needles into the penalty area       more than anyone not named Almiron in 2018. However; you also need someone making the runs to be on the end of those passes.

He may have been most noticeably the creative hub, but Almiron also put in the work, made frequent hopeful runs and found himself receiving passes in the penalty are at rates similar to the best attackers in the league. Similarly, Hector Villalba, despite assisting Josef more often than any player not named Julian, made himself available in the box time and time again, attempting to get in behind and stretch the back line. His attacking output per game reflects this. The prospect of fielding all four players at once is tantalizing, but if you are dead set on a 3-4-3 that’s not going to happen, and you need to make sure you have players who can create, but also enough run support to draw defenders off Josef in the penalty area. A suboptimal scenario for Atlanta United is Josef Martinez looking for space in the box and both Pity and Barco wanting to be on the ball orchestrating in similar areas just outside the area. Conversely, if Pity Martinez can replace all and more of Almiron’s attacking output (especially the runs into the penalty area and the shots), then the Barco/Villalba controversy is less important. In the end, I have a certain faith in the fast player, but neither is going to replace Almiron’s defensive contribution.

Shape and Roster Considerations

Gonzalo “Pity” Martinez
Brek Shea
Florentin Pogba
Anderson Asiedu 

Chris McCann
Greg Garza
Miguel Almiron (!)
Sal Zizzo
Andrew Wheeler-Omiunu
Mitch Hildebrandt
Oliver Shannon 

I’ve hinted at it above, but the team looks to keep its “back three with wingbacks” foundation, but instead of Tata’s more playoff-hardened “three in the middle to win balls plus Martinez-Almiron FUSION up top,” de Boer appears headed towards the single striker flanked by two playmakers who I would expect to interchange and/or tuck-in often given the now two-man central midfield losing a body.

In preseason, Franco Escobar has been injured and it’s clear that while he is starter material, it’s unclear whether he would slide right into the right sided center back spot or reprise his play-off right wing back hero role, a role that most would suggest Gressel fits perfectly into – though the fact Tata played him centrally throughout the playoffs makes this something to watch in 2019.

Due to this injury, Miles Robinson has stepped into the sweeper center back role so far with Michael Parkhurst on the right and Gonzalez Pirez on the left. Parkhurst is as slow as Geoff Cameron is Republican Robinson is fast, so this is an intriguing choice, and suggests that what de Boer values in his flanking center backs is the ability to split lines with forward passes (something Parkhurst excels at) and not the wheels to keep up with a tricky winger should a tricky winger slip through midfield (something Robinson most definitely possesses.

In the center of midfield, one would think 2018 signings Eric Remedi and Darlington Nagbe would be shoe-ins for the starting double pivot; however, something’s up with Nagbe who has sat out of training amidst rumors of personal issues with club personnel. At time of writing Nagbe has said to the press that he loves being in Atlanta and he’s ready to go, but this is exactly what Carlos Carmona said before he was transferred out of MLS late in preseason 2018. Jeff Larentowicz has filled in beside Remedi in preseason, and this reminds me that every year you think Larentowicz, now 35, has finally become a bench player, is a year he starts 35 games for a top team in MLS.

We’ve talked about the left wingback role, with Bello the heir-apparent as soon as he can beat out man-mountain Brek Shea for the role. The smart money suggests the veteran demi-god Shea starts the year on the left, but it’s very much up for grabs, and one imagines Greg Garza would not have been sold if it wasn’t partly to clear the path for the 17 year old George Bello, the club’s top homegrown prospect.

The right side is also up for grabs in the sense that there are two starter quality players who both fit the role perfectly in Julian Gressel, who provided more service for Josef Martinez in 2018 than anyone and Franco Escobar, physically dominant in almost all matchups against MLS wide players. It seems almost certain that both players would start somewhere in the eleven, but in what is a running theme, it could very possibly not be the case.

This leads us with the front three which we’ve discussed above ad nauseum: Josef Martinez likely flanked by Ezequiel Barco and Pity Martinez with Villalba the first sub off the bench and likely a rotational starter. Andrew Carleton remains in the supporters hopes and dreams, and while the very limited sample size of his work to date shows signs of promise, so far has failed to force a manager’s hand for more than a few appearances.

2019 Expectations

Everyone is not going to stay healthy. I mean, you just know this about sports. There are always one to two key players missing. My guess for Atlanta United is that they can win either the Shield or the Cup if one to two key players are missing at a time, either with a little luck in the early part of the season or a swift loss in CCL. They nearly won the Shield dealing with mostly independent stretches of missing Garza, Villalba, Barco, Nagbe, and even Almiron in 2018. They won the cup with everyone mostly healthy in the playoffs.

With CCL and MLS given equal priority, I think Atlanta is built about as strong as an MLS team can be built given the roster rules. The squad has starter quality depth backing up the striker and playmaker positions in Villalba (or Barco) and starter quality depth in the center of midfield in Gressel (if Escobar plays wingback) and Larentowicz. Gressel can play wide right if Escobar is out or playing CB and one of Shea or Bello will ride the bench on the left. As with most teams, critical injuries to the center backs would be a problem, but there’s enough versatility in Larentowicz and Escobar to make it work. But the above depth and potential for rotation only works across multiple competitions if everyone is generally healthy.

A reasonable effort would be to prioritize CCL and the making the playoffs to compete again for the cup.