/cdn.vox-cdn.com/uploads/chorus_image/image/59999639/usa_today_10872854.0.jpg)
Following this weekend’s games (in which Minnesota United FC get added rest), Major League Soccer reaches a natural break that comes from both the World Cup and clearing through the first third of the season. At this point, certain teams have established themselves in a hierarchy in each conference; out in the East, Atlanta and both New York sides have started to cruise, and in the West, Sporting Kansas City have jumped out well, with FC Dallas, Portland, and LAFC not far behind. On the flip side, others have already become mired by poor performances; Colorado and San Jose out West, and Montreal in the East all are starting rebuilds poorly, with DC’s lengthy dry spell continuing.
Minnesota United, so far, have fallen a little bit further down. They currently sit ninth in the Western Conference, four points under the playoff line with three games past the one-third pole. They’re around a poor median mark in away wins—like the Loons, six teams in the West have one or fewer—and their home form is actually above average, sitting fifth on points. Of their five wins, two (away vs. Orlando, home vs. Vancouver) are against current playoff teams, with their lone draw against their top conference rival in SKC; of their eight losses, all but three (home and away vs. San Jose, away vs. Seattle) have been against current playoff teams. In other words, Minnesota are competing well against the not-that-great teams and are coming up short against the class of the league.
As we approach this break, it’s as good a time as ever to assess what the top teams are doing well and rate where the Loons are coming in. To do this, I took the team-by-team averages per game of all of the detail stats available at WhoScored.com, then ran a correlation function in Google Sheets between each stat and the team’s average points per game. From there, I ran an R-squared function to help determine whether or not the differences between PPG rank and each given stat showed significant enough fit to suggest that it has descriptive value.
I’m going to throw a huge caveat out before I get too far: no data scientist would feel comfortable divining tons of knowledge from this early in the season. The only R-squared numbers that would be deemed remotely significant (read: R-squared<0.9) in the 107 pairs are between PPG and total points and wins; after losses and WhoScored’s average player rating, none are greater than 0.7. Part of this means to take all of the information with a grain of salt, though for my money the grain of salt is about as illustrative as we get at this moment.
Without further ado, here are the 12 correlations to points per game whose R-squared values exceed 0.4:
Stats Correlating With Higher Points Per Game
Statistic | Corr | Rsq |
---|---|---|
Statistic | Corr | Rsq |
Points | 0.9752715807 | 0.9511546561 |
Wins | 0.9597430503 | 0.9211067225 |
Losses | -0.8721511967 | 0.76064771 |
PlayerRating | 0.8507009621 | 0.7236921269 |
OGoals | 0.8169894644 | 0.6674717849 |
OpenPlayPG | 0.7509792824 | 0.5639698826 |
OtherAssistPG | 0.7426959169 | 0.5515972249 |
ShotsInBox | 0.7362781189 | 0.5421054683 |
GoalsPG | 0.7310454101 | 0.5344273916 |
AccCorner | 0.6912655282 | 0.4778480304 |
CornerPct | 0.6816708081 | 0.4646750906 |
ShotsOTPg | 0.6417631944 | 0.4118599976 |
From here, we can lump out some axiomatic ones. If a team has more points and wins, it will almost certainly get more points per game; likewise, if a team has fewer losses when all teams have played at least 10 games, it’s no shock that said team will have more points per game. We can also assume that WhoScored has optimized their Player Rating statistic to credit players for their positive contributions and demerit their negatives, given that they have data on a ton of games.
As a result, we can look at the remaining eight and find five general themes that top teams in the league do each game.
Total Offensive Goals and Goals Per Game (correlations: 0.817 and 0.731, R-squared: 0.667 and 0.534)
Gasp. Teams that score more goals win more points. While this might be obvious, it’s definitely instructive that MLS continues to be a league where truly outscoring teams matters as a way to gain points. It’s also not a shock that Minnesota United’s spot in the total goals scored list mirrors that of their ranking in the Supporter’s Shield. The Loons are tied with Toronto FC for the seventh worst goals scored number this year; they’re seventh worst on points in MLS this year. It’s fair to suggest that the Loons’ best spots on the field are in attack, but given the low resulting number of goals it’s hard to suggest that Minnesota can be fully comfortable yet.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/11501725/usa_today_10855530.jpg)
Open Play Goals Per Game (correlation: .751, R-squared: .564)
At the outset, it’s important to define what an open play goal means. Per WhoSocred, open play goals mean any goal that happens aside from a set piece (i.e. directly from a free kick or corner kick). Counters are included in open play goals, but their comparative correlation and R-squared show them to be insignificant to PPG; additional checks as a control variable suggest they show no significant relation to open play goals. More succinctly, teams that score open play goals in the standard run of play do better. Again, it’s a bit axiomatic: if you score more goals, chances are you’ll win more games, but the fact that open play goals come out as the biggest indicator of success as opposed to, say, corner kicks or penalties, show that success comes from the offensive Plan A paying dividends. Minnesota in contrast ranks fourth last in this mark in front of only Montreal, Vancouver, and Colorado.
Other Assists Per Game (correlation: .743, R-squared .552)
This one also needs some extra explanation. WhoScored groups their assist types by through balls, crosses, throw-ins, corners, and “other,” which encompasses short and medium balls into the scorer. Overall assist counts per game show some correlation, but whereas the other types show weak and/or negative correlation (crossed assists are a weak negative, oddly enough), these short play assists are showing to be an indicator of a team that is connecting play well and getting results. Given that a reliance on crossing or long balls seems to go in tandem with a team trying to find extra space versus controlling the ball deep, this isn’t a huge shock. Moreover, it’s also not a shock that a counter and cross team like Minnesota ranks just ahead of Colorado in the basement of MLS here.
Shots Per Game Taken In 18-Yard Box and Shots Per Game on Target (correlation: .736 and .642, R-squared: .542 and .412)
MLS teams aren’t yet turning into the Houston Rockets in eschewing low likelihood shots. That said, it’s notable that while overall shots per game appear to be weakly correlated to success, quality shots show a much stronger correlation, with shot counts outside of the box having almost no noticeable correlation. It also follows that teams with a greater amount of shots in the 18 are getting a higher amount of shots on target per game—the correlation between those two is, for our lower standards, quite strong (.761 correlation). All of this adds up to teams having more success when they’re taking higher volumes of high-percentage shots. Here’s a surprise: Minnesota United also score poorly on these rankings, coming in fourth worst in shots per game on target and third worst on shots in the 18.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/11430515/MTL-3.0.jpg)
Accurate Corners Per Game and Corner Completion Percentage (correlation: .691 and .682, R-squared: .478 and .465)
These two go close to hand in hand, with all teams getting between 3.8 and 6.6 corners per game and most teams in a cluster around 4.8-5.8. If you’re able to get any regular corner completion, it’ll put you further ahead of the pack. More importantly, any completed corners mean some sort of close chances were created. The three top teams in points per game all have at least three accurate corners per game, with Orlando the other strong set piece side. As Loons fans might be aware, corner kicks are not quite the team’s strong suit. They’re bottom of the league in accurate corners per game with 1.3, and are third-worst in completion percentage with a shade under 30%, trailed only by Colorado and Seattle.
So, in general, the shared traits of good MLS teams are as follows. They score a lot of goals per game. They get those goals primarily through open play, generating the goals through short- to medium-length assists. They pump a lot of shots in on target and from the penalty area, and they get solid to good accuracy on their corner kicks.
By all of these tokens, Minnesota United don’t do the things that good teams regularly do. They don’t score a ton of goals, generate fewer shots in the box and far fewer on target than other teams, they rely on a higher proportion of crosses (and, might I add, complete them at the worst clip in the league) when successful teams generate chances on short plays. Where they lead in counting stats is often in defensive categories, specifically in blocked crosses and shots; the problem is that the weakly negative correlations suggest that the increased activity at the back is less a sign of proactive work and more a symptom of suffering pressure.
The question coming into the break is whether or not the Loons can address these via tactical shifts or if personnel moves are needed. Scoring goals can come from placing higher tactical responsibility on getting the ball to Christian Ramirez in the box through more patient buildup and short passing plays in the final third. Ibson can be pushed into playing the ball tighter with Darwin Quintero centrally versus out to the wings to force more of this buildup. Goals can also come from adding a Designated Player on the left wing or in central midfield to take shots as the team presses deep.
As far as getting better shots, MNUFC can rely on Miguel Ibarra, whose shooting accuracy has been solid to great this year, to take more shots in the box as opposed to far shots from the likes of Alexi Gomez. The technical staff can (again) push Christian Ramirez higher, giving him reign to camp in the box and target him for close shots. The Loons can also find ways to put their holding midfielders higher into the box, sacrificing a compression of the opposition defense for greater numbers able to shoot close. Likewise, assuming the goal is to increase the forward efficiency of shots, Minnesota can push for an improvement given that Ramirez, Abu Danladi, and Mason Toye are all within 0.3 shots on target per 90 of the league median, which includes all 314 players that have a shot on target this year.
The team probably can’t expect better accuracy on corners due to a number of height disadvantages in the first team. There are only three players in the standard starting XI that stand over six feet: Christian Ramirez and Michael Boxall at 6’2”, and Eric Miller at 6’0”. However, it can bring players like Brent Kallman, Bertrand Owundi Eko’o, or Wyatt Omsberg (6’2”, 6’2”, and 6’4” compared to the 5’11” of Francisco Calvo) in to be a tall aerial threat. If Calvo is moved following the World Cup, Minnesota can look to replace him with a center back that provides adequate size to push him forward on corners. More realistically, Minnesota can adopt a short corner play here and there. On that play, Loons who would be expected to crash the box for headed shots could instead break to the penalty spot for more accurate shots, or passes through the chaos to forwards hovering in the six-yard box.
I can find myself admitting that the tactics the Loons have taken are ones that play to the team’s strengths (or lack thereof). The teams that play through the middle and get assists through short passes show an ability to roll with elite pairs in the attack and attacking midfield. Minnesota’s major focal point shift has been the addition of Darwin Quintero, a player situated centrally in part because his skill level as a ball player is seen as more influential in the center than in a wing space where his comfort level and pace are greater outliers. Their most skillful players—Quintero and Miguel Ibarra—are most comfortable wide, with their top creative central player in Ibson far more apt to look for longer plays to move the attack quickly than players looking to patiently draw the attack into the final third. A coach works to derive the best from what he or she has, and I see why there’s not enough confidence in the talent to execute more complicated tactics.
With that being said, there are enough moves that Minnesota United can take with the personnel they have to improve. Wide players can inch inward to get closer passes for better chances in the box. Feeding Christian Ramirez with closer shots can pay dividends. Miguel Ibarra’s value to the team comes in large part because he’s been a positive at three main positions in the lineup. Brent Kallman’s ability going forward (bolstered by his best Vince McMahon impression against FC Cincinnati) can assist on set pieces more than what Calvo has shown this year.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/11500221/usa_today_10877892.jpg)
Ultimately, it seems that what would help the Loons is a willingness to use their talent down the middle. Both Miguel Ibarra and Darwin Quintero can work their ways inside from the wings, with Quintero using pace and ball control and Ibarra using a constant eye to support play. Swapping from a double-8 and #10 to a straight trio in central midfield like they rolled out against FC Cincinnati and to start the second half against Sporting Kansas City helped to anchor play in the middle of the park, and it gives license to Rasmus Schuller to shuttle from box to box (along with allowing Ibson to do... whatever it is he does). Keeping those lanes clogged also helps to alleviate the pressure in the defensive third and establishes a base to apply greater pressure as the opposition builds play in their back line and midfield.