Michael Boxall - Brent Kallman Photo credit: Daniel Mick

The Angle

Quantitative Divination – Center Back Edition: Boxall or Kallman

by on 5 March 2018

The 2018 preseason seems to have brought forth a new player controversy. Who should get the start at center back and be paired with Loons captain Francisco Calvo. Michael Boxall or Brent Kallman? FiftyFive.One writer and staff statistics guru, Dave Laidig, looks at the numbers.

One of FiftyFive.One’s regular instigators queried how the stats for Brent Kallman compared to Michael Boxall’s. While this is a bit of a loaded question, knowing that these center backs are likely competing for the chance to start alongside Francisco Calvo meant that I’d try to divine some sort of reality from the quantitative assault of my spreadsheet.

In the defensive third, Kallman won 73% of the aerial challenges and Boxall 56%

First, I turned to the individual stats over the course of the season. Immediately, Kallman stood out for his success winning aerial challenges. In the defensive third, Kallman won 73% of the aerial challenges and Boxall 56%, a figure that’s closer to the league average for challenges in this area. And defending in the air is pretty important considering Minnesota was the league worst in this area, and needs all of the defensive help it can get.

With other individual stats, it’s a bit of a mixed bag and we have to consider the uneven playing time between them. Boxall had more tackles per 90, including successful ones. Kallman has the slight edge with interceptions per 90. Boxall has about three times as many shot blocks per 90, and more crosses blocked.

Next, I looked at the team results when these players were on the field. Center backs work in tandem to organize the defense and reduce an opponent’s scoring opportunities. They are the last line of defense and are positioned centrally. Considering their roles, along with the keeper, these positions may be judged based on the team defense results more than other field players.

With this in mind, I quickly looked at 4 starting CB pairings: Kallman-Calvo (14 games), Kallman-Joseph Greenspan (2 games), Boxall-Calvo (7 games), and Boxall-Kallman (5 games). Bobby Shuttleworth was the starting keeper for all these games. For each of these pairs, I looked at Minnesota’s goal differential, the xG conceded and xGD (the xG differential).

And a brief note about the xG conceded and xGD figures used.All xG info derived from the solid work of American Soccer Analysis; and you can find a variety of MLS info there, including interactive tables. Here, I am not using the actual xG figures, but an adjustment that accounts for the opponent and whether the match was played at home or away. With partial season sample sizes, differences between opponents can easily skew the results. Thus, the xG related comparisons are based on the opponent’s average performance for a home (or away) match. In other words, the CB pairs will be judged on whether the opponents consistently produce more or less than their average xG.

On these measures, the Kallman-Calvo pairing performed the best.  When Kallman-Calvo patrolled the middle, Minnesota conceded an average 0.24 xG above their opponents’ average xG rate. And the xGD (i.e., the difference between Minnesota and the opponents xG) was right where it should be considering the opponents.  And the actual GD was slightly positive, +2 goals over the 14 games.

The two games with Kallman-Greenspan at CB were the worst when one looks at xG; conceding on average 1.09 xG above the opponents’ average. The opponents were getting more quality shots then usual. The expected goal differential was below the team averages (specifically, -1.62 xGD ). Rephrasing, Minnesota opponents were able to add 1.09 xG to their average offensive output and Minnesota fell another 0.53 xG short of its average offensive output. Combined, the xG difference was 1.62 below what we would expect based on the Minnesota and opponent season averages. But this was only two games, and both games were actually draws (GD +0). So while the underlying values are bad, we can’t really say a whole lot about this pair.

All in all, Boxall is more active and shows up more in the standard defensive stats like blocks, tackles and interceptions. But Kallman and Calvo pair were better able to manage the opponents and reduce their opportunities.

The Boxall-Calvo and Boxall-Kallman pairs were similar in xG conceded. When Boxall was at center back, Minnesota allowed opponents to average 0.52 to 0.69 xG above their typical rate (Boxall-Calvo and Boxall-Kallman respectively). And the xGD was below what one would expect given the opponents (specifically -0.76 and -0.71 xGD below the team averages). When paired with Calvo, the actual goal differential was -1, and a full -6 when with Kallman.

All in all, Boxall is more active and shows up more in the standard defensive stats like blocks, tackles and interceptions. But Kallman and Calvo pair were better able to manage the opponents and reduce their opportunities (while not obviously hampering the offense). It’s possible that Kallman’s strength (in the air) is more important for a CB than the other types of actions. Additionally, the Kallman-Calvo pair was primarily a first-half of the season thing – so it is certainly possible that some other factor not mentioned here also correlates with the results (Cronin at full strength perhaps; or some tactical instructions from the manager).

Either way, it’s not an easy call to pick between Kallman and Boxall. Each has some statistical strengths to support a case for playing time. Because I’m more persuaded by team results for center backs, and Kallman adds some offense every now and then, I’d give the edge to Kallman.

FiftyFive.One is now on Patreon. Do you like the independent coverage of soccer news from Minnesota and beyond that FiftyFive.One offers? Please consider becoming a patron.

Tags: , , ,