Turns Out, Vito Mannone Has Been Good For MNUFC

I’m well aware that nothing good ever comes when you get baited by a bad take on social media. I do whatever I can to avoid it, but sometimes you just can’t help yourself against takes that are egregious. Like this one:

I privately espoused the belief that Bobby Shuttleworth had a case to start against Dallas, but not remotely because of the quality of Vito Mannone’s performances. Mannone played all 23 of the team’s league games to that point and went the full 90 in each U.S. Open Cup game, only dropping to the periphery in friendlies against Hertha Berlin and Aston Villa. He lands in the top 15 in American Soccer Analysis’s tracking of minutes played as of Sunday’s games, and would be two spots higher if not for similar ironmen in goal for Seattle and San Jose (Stefan Frei and Daniel Vega) having two extra minutes of stoppage play. That’s a lot of miles, and if a player with MLS experience is available you might as well give him a run out before he goes on loan.

Adrian Heath didn’t do that. Vito Mannone proceeded to have one of his worst games as a professional, with his five goals allowed only bettered (or, well, the exact opposite) by the eight goals he allowed for Sunderland against Southampton in October of 2014 beautifully immortalized in this YouTube video of a disgusted fan talking over Match of the Day highlights.

In light of this, it’s worth taking up the premise that Vito Mannone should have been benched in favor of Bobby Shuttleworth in Dallas as a null hypothesis. I dove into Mannone’s numbers through the Dallas game (American Soccer Analysis won’t have the data on the Colorado game for another week or so), comparing them to Shuttleworth’s performances both for Minnesota and New England in 2015 and 2016. In short: Bobby Shuttleworth’s overall value came from needing to do a whole lot and doing most of it, while Mannone’s value has been equal compared to the lower amount of work required.

Vito Mannone’s eight clean sheets in 25 games after Wednesday’s game against Colorado are more than the club marked over its first 68 MLS games. When extrapolating the 96 minute numbers out to a 34 game season, Mannone is on pace to allow 19.7 fewer goals than an equivalent 2018 from Bobby Shuttleworth. His save percentage on shots on target outpaces Shuttleworth too: the 70% save ratio for the Italian comes in tenth in the league among goalkeepers with 500+ minutes and outpaces Shuttleworth’s 67% rate in both the 2017 and 2018 seasons, both of which would come in square with the 2019 league average.

The top line isn’t entirely in Vito’s favor, however. Shuttleworth’s two seasons both boasted more saves than Mannone’s 2019 per 96 minutes—Bobby averaging four-plus saves per 96 in 2018 was among just four such seasons for regular starters across the league since 2015, with David Bingham and Jesse Gonzalez on their way in 2019. Each year for Shuttleworth also featured fewer goals allowed minus expected per game, something Mannone’s 0.10 per 96 minutes can’t boast.

Minnesota’s defense being significantly better makes it difficult to gauge the individual impact of a goalkeeper’s performances. The Loons are not only allowing 1.28 fewer shots on goal per 96 minutes in Mannone’s performances than Shuttleworth’s in 2018, but are also forcing shots that are 1.5 yards further from goal with a 0.05 lower expected goal count per shot on goal. In other words, Loon opponents take shots that are 19% less likely to be scored; if the shot quality for 2018 were ported onto 2019’s shots on target per game, the Loons would be liable for 0.26 more expected goals allowed per 96 minutes. Vito’s defensive help trails only D.C. United’s Bill Hamid for having the most assistance from a defense in forcing bad opposition shots. It’s easier to be Vito than it had been to be Bobby.

That also means that Mannone’s counting stats are going to read better and worse than Shuttleworth’s. Mannone makes 0.70 fewer saves per 96 minutes than Shuttleworth did in 2018 and 0.31 fewer than 2017, but seeing a 21% drop in saves per 96 makes sense when the team allows 27% fewer shots on target. The 29% drop in goals allowed per 96 tracks well with the 34% drop in expected goals allowed. It’s also why you need to look at save percentage in context–Mannone’s 70.1% against Shuttleworth’s 66.9% in 2018 and 67.1% in 2017 is certainly better, but there’s a little less value in the jump when there’s less (and worse) shots faced.

That’s partly why passing and possession data can help. While the duo’s 2018 and 2019 seasons are nearly identical in touch rate, the reduced amount of restarts means that Mannone has been used more in open play. With Minnesota incorporating Mannone into the buildup of attack more often, it parses that 11.76% of the team’s shots come on a chain of passes and dribbles that at one point involved Mannone, a 2.4% increase from last season’s numbers with Shuttleworth.

He’s also completing the same number of passes—62.5% with 1.76 more pass attempts per game than Shuttleworth had in 2018—despite an overall more difficult set of passes made harder by the pressure of open play. ASA’s expected passing model suggests that Mannone should complete 63.1% of his passes, due in part to his average pass going around three yards further and 4.7 yards more downfield than Shuttleworth’s did in 2018. As a result, ASA suspects, Mannone has missed around 0.62 passes that he should have completed per 100 passes, good at his passing rate for missing 0.19 makeable passes per game. Shuttleworth, in contrast, missed 1.95 makeable passes per game on a -6.71 passing score per 100 passes.

What’s more, consider the impact of the added vertical aspect to Minnesota’s goalkeeper distribution. Mannone takes plenty of short passes between himself and the center backs, but he also targets through to the far end of the field. Here’s a shot of Mannone’s passing chart from Wednesday’s game against Colorado—green means a completed pass, red means it was inaccurate:

Opta Chalkboard via MLSSoccer.com

The accuracy on the long passes can be a concern, but Mannone’s ability to connect into the midfield—five completed passes into the middle third of the field—stands in contrast to Shuttleworth. Here’s his passing map from June 2018’s away match against… Colorado!

Opta Chalkboard via MLSSoccer.com

Shuttleworth’s misses throughout 2018 were all over the pitch, while his completions were in spots that left little opportunity to move the ball in progressive ways; of his six completed passes against Colorado, just three weren’t directly next to the touch or end line. Mannone, in contrast, completes passes into channels that play runs through, allowing for more passes that let a midfielder or defender play the ball on to someone on a wider overlap.

Being overly conclusive about Mannone’s performances independent of the Loons’s defensive improvements is difficult. The save percentages, however, are up and the distribution improvement paints a more favorable picture. When considering that the defensive-aided metrics are improving in line with how much the back line and midfield has advanced this year, Mannone has definitely carried his weight accordingly. Mannone ultimately exceeds league average in save percentage, goals allowed per shot and per 96 minutes, and saves per 96 minutes, which is why you would be crazy to say he needs to be dropped in favor of a player who hadn’t played and whose contribution to the team was inflated by the abject poverty of his teammates.

That doesn’t necessarily mean that Minnesota fans shouldn’t entertain the thought experiment of what the team would look like with Shuttleworth in net this year. The reduction in expected goals per shot on target yielded by the defense this year would have yielded 0.325 fewer expected goals per 96 at 2018’s shot level; the same save rate that Shuttleworth worked with would yield 0.43 fewer goals per 96 if adjusted to 2019’s shots conceded. The end results for those, however, are still higher than what Mannone’s allowed this year, and it’s a safe bet that the distribution issues would have resulted in more shooting chances regardless of the defenders.

If we’re being as honest as possible, chances are that had Shuttleworth started the Dallas game he would have gotten shelled in ways eerily familiar to Loons fans. The evidence was in fact more voluminous than we want to admit: two years in net without a defense featuring Ike Opara, Ozzie Alonso, and Romain Metanire. Vito Mannone had an objectively bad game against Dallas and could have singlehandedly stopped at least two goals from going from chance to shot, but the track record that Shuttleworth showed in games against Philadelphia in October 2018 or New England in March 2017 suggests that he’s got a five-goal howler in him too.