By Matthias Kullowatz (@mattyanselmo)
When August rolled around, Sporting Kansas City looked poised to make another deep playoff run. Sporting led the Eastern Conference in just about everything that mattered: points, shot differential, goal differential, expected goal differential...you get the picture. But over its last 13 games, Sporting mustered just 11 points and sputtered its way into the fifth seed (see the season progression chart above). It would be easy to chalk it up to bad luck and small sample sizes, but let's first take a closer look at those those 13 games.
|SKC Offense||Games||Shots||Avg. Dist||CrossPct||GoalsF||xGoalsF|
|SKC Defense||Games||Shots||Avg. Dist||CrossPct||GoalsA||xGoalsA|
Interestingly, expected goals suggests that it was Sporting's offense that led to its downfall during the last 13 games. But results--that is, actual goals--shows that SKC's defense was to blame.
Offensively, it seems that Sporting changed style significantly, moving away from getting such a high proportion of shots off crosses. Whether it was tactics or luck, Sporting's finishing improved drastically during that last third of the season, and it made up for a decline in quantity and (apparent) quality of its shots.
Defensively, expected goals didn't detect much of a change, and yet Sporting allowed nearly twice as many goals per game over the last 13 games. Six more goals were scored against Sporting than expected over that time, or perhaps more emphatically, teams playing against SKC scored 140% of the goals they were expected to score. This might sound like a symptom of a backline in flux, but SKC was finally healthy at this point. Matt Besler, Aurelien Collin, and Seth Sinovic all started more than half of these games, and at least two started 11 of the 13 games. Additionally, Sporting was playing well defensively earlier in the season anyway, and that was when it was dealing with injuries and World Cup festivities.
It's tempting to say that Sporting's sharp defensive decline was just bad luck, but there were a few things different about those 13 games worth mentioning. 21-year-old Jon Kempin played more than 300 minutes in goal, a period in which SKC gave up slightly more expected goals than usual, and in which SKC gave up even more actual goals than that. It's a small sample size, but Kempin's tenure resulted in giving up nearly three goals more than we would have otherwise expected.
And during all those other minutes down the stretch, when Andy Gruenebaum was keeping, the average shots he faced were better placed in the goal mouth than the league rate. Three goals better it turns out, making up the six-goal difference between actual goals allowed and expected goals allowed over that rough stretch. Why were they better placed? Maybe something about Sporting's tactics changed during the last third of the season. As mentioned, the offense seemingly attacked through the middle more often, or perhaps via longball more often, generating fewer shots off crosses. This may have led to pushing more numbers ahead, and to more vulnerability on the counter. This is a phenomenon to which expected goals would not be very sensitive.
Or maybe it was nothing. We often get caught up in splitting seasons at arbitrary dates to push a narrative. While I just spent a few hundred words suggesting that something might have changed, I haven't found an obvious cause. After all, New England went through a few streaks of its own, some more explicable than others, on its way to the MLS Cup Final. I simulated 10,000 13-game seasons in which a team allows the distribution of shots that Sporting did, and between five and 10 percent of the simulations had Sporting allowing at least as many goals as it did. The defensive let down could have just been a fluke.
Sporting returns its three most recognizable names and arguably most valuable players in Besler, Graham Zusi, and Dominic Dwyer—captains of the defensive, middle, and attacking thirds, respectively. However, there were some important changes to the roster this off-season, starting at the back.
After giving minutes to Eric Kronberg, Gruenebaum, and Kempin in 2014, it looks as though Sporting will turn to Chilean international Luis Marin. Marin has eight caps for the Chilean national team and is only five years removed from a roster spot in 2010’s World Cup. At 31 years old, he presumably has a lot of goalkeeping juice left. Considering that SKC’s defense was top notch not too long ago with the likes of an aging and overrated Jimmy Nielsen in net, Marin should be seen as a welcome improvement.
Moving up the field, the loss of Aurelien Collin is the biggest blow to the back line. However, Sporting put up another strong year defensively in 2014 largely without Chance Myers, who will be returning this season from a ruptured Achilles injury. Perhaps surprisingly, Myers has matched Collin over the past four seasons in offensive production, as measured by xGoals added to xAssists (xG + xA, adjusted for minutes), a metric measuring total value of all the shots and key passes that a player produces. Though Myers and Collin play distinct positions, in essence this is like a trade of one for the other from Kansas City's perspective. Additionally, Sporting posted its best xGD in Myers’ 609 minutes relative to every other player.** Sporting returns long-tenured Sinovic, the also-injured Ike Opara, and now-broken-in Kevin Ellis. My hunch is that Collin’s absence will not be nearly as costly as many might think.
Though Zusi often plays on th wing, his role is more that of a midfieder. Zusi will be joined by familiar faces Paulo Nagamura and former USMNT capper Benny Feilhaber. Many of Nagamura's minutes are likely to go to Honduran Roger Espinoza, which won't make Sporting's midfield any worse. Though our data isn't refined enough to prove it, I strongly feel that Sporting's play in the midfield is part of what makes its defense appear so effective. Midfielders that can hold the ball, threaten the opposition's defense, and avoid costly turnovers in dangerous areas of the field surely take pressure off the defense. My sense is that this midfield core does exactly that.
Sporting played much of last season with just one striker, Dwyer. Combinations of C.J. Sapong, Sal Zizzo, Toni Dovale, Soony Saad, and Jacob Petersen were employed at the wings. Zizzo is off to the Red Bulls, and Sapong to Philly, so those minutes are likely to be replaced in part by Hungarian international Krisztian Nemeth. Nemeth played in 25 games for Dutch Eredivisie side Roda JC Kerkrade in the 2013-14 season, where he scored eight goals to go along with six assists. More important than his counting stats output is probably the fact that a first-division Dutch club put its faith in him over 25 games. Nemeth is almost sure to be a boost to an underrated attacking unit.
This offense gets some flak for producing low overall goal and expected goal totals, but much of that could be due to pace and style of play. Alex at Tempo Free Soccer shows that Sporting ranks highly in possessions per game and passes per possession, while limiting its opponents' passes per possession to the lowest rate in the league. This patient offense and high pressure defense has worked to maintain a positive possession ratio, shot differential, and goal differential for some years now, and SKC should have the personnel to keep it up in 2015.
Despite research that suggests offensive numbers are more sustainable than defensive numbers over a single season, Sporting’s system has been consistently producing these outputs for at least four seasons. Whether it’s style and pace of play or personnel or some of both, Sporting still manages to produce positive results. Its few roster losses should be mitigated by the new faces and the club's system, and I expect Sporting to finish in the middle of the Western Conference playoff picture with a three or four seed, somewhere behind LA and Seattle.
*Though Sporting's xGD was a little behind that of Seattle and LA, SKC didn't have to face either of those teams until a potential championship game. In other words, its path was paved a little smoother.
**While I would say generally that my first stab at Plus-Minus stats produced a mostly worthless metric, this is a case where perhaps something can be gleaned. Comparing Myers only to his teammates isolates the team variable, and using a player that played enough minutes for expected goals to begin stabilizing but not enough to represent a majority of the team’s minutes, gives us a treatment vs. placebo-like look at Myers’ contributions. Typically full-time players will assume the xGD of the team, which is essentially useless information. In any case, there’s evidence to suggest that a nearly-full season of Myers will adequately fill the hole left by Collin.