Welcome to The Graphical, wherein I mine some of the available data on Nashville SC’s recent results to provide some additional insight. A result that was closer than the underlying numbers expect requires a dive. Also we’re a couple days behind on content here, so let’s get right to the thing.
A statistically unlikely result!
The game was a nail-biter, and at one point looked doomed to a draw (briefly, before VAR swooped in to save the day). It… shouldn’t have been. Based on the shot quality that NSC was able to find, scoring only one goal was pretty unlikely:
| Nashville goals | Probability |
| 0 | 3.32% |
| 1 | 20.98% |
| 2 | 33.63% |
| 3 | 26.22% |
| 4 | 11.79% |
| 5+ | 4.06% |
It’s obvious that, with an xG value of 2.3592, outcomes of either two or three goals were most likely. However, seeing that either of those individually was more likely than the combined likelihood of scoring fewer than two goals is still pretty stark!
With a major assist from ASA‘s Eliot McKinley, here it is represented graphically:
I’m not exactly breaking new ground here that Nashville’s finishing was poor on the night. But if you assume finishing is largely down to luck (or at the very least, not consistent from game-to-game – about which more in the next section), it’s pretty stark to see exactly how poor it was.
Is Nashville’s finishing just bad?
So far this year, Nashville has been struck by more bad luck than any team in the league. There’s not quiiiiiiite enough data to make the opponent-adjusted power rankings public, but NSC is the second-best team in the league on xG terms (NYCFC is in first by a clear distance), and below-average in terms of actual goals scored.
While slightly more of that bad luck has come defensively (an impressive tidbit, given that the team hasn’t given up an Actual Goal in over four games – that’s just how bad the Cincinnati and Montreal goals were), there’s no doubt that the offense has been a major culprit: the team has 11.07 xG so far this season, and just seven goals to show for it.
In the micro, the results have been skewed by bad finishing. But is this a bad finishing team? Wes and I touched on it in the podcast Tuesday, but it’s worth exploring. I’ll go through the whole roster here. All data via American Soccer Analysis:
| Player | 2021 minutes | 2021 goals | 2021 xG | 2021 G/xG | 2013-2020 MLS minutes | 2013-2020 MLS goals | 2013-2020 MLS xG | 2013-2020 MLS G/xG |
|---|---|---|---|---|---|---|---|---|
| C.J. Sapong | 299 | 1 | 2.87 | 35.8% | 16851 | 58 | 65.31 | 88.8% |
| Jhonder Cádiz | 332 | 2 | 2.65 | 75.5% | 483 | 2 | 1.63 | 122.7% |
| Randall Leal | 562 | 2 | 1.00 | 200.0% | 1981 | 4 | 2.33 | 171.7% |
| Daniel Lovitz | 578 | 0 | 0.18 | 0.0% | 11495 | 2 | 3.16 | 63.3% |
| Hany Mukhtar | 339 | 1 | 1.07 | 93.5% | 1468 | 5 | 4.51 | 110.9% |
| Walker Zimmerman | 590 | 0 | 0.32 | 0.0% | 15176 | 16 | 11.88 | 134.7% |
| Alex Muyl | 442 | 1 | 0.41 | 243.9% | 8988 | 12 | 18.58 | 64.6% |
| Aníbal Godoy | 573 | 0 | 0.44 | 0.0% | 11122 | 6 | 6.69 | 89.7% |
| Handwalla Bwana | 105 | 0 | 0.52 | 0.0% | 1580 | 4 | 3.67 | 109.0% |
| Alistair Johnston | 590 | 0 | 0.03 | 0.0% | 1798 | 0 | 0.41 | 0.0% |
| Dominique Badji | 184 | 0 | 0.56 | 0.0% | 10809 | 33 | 35.53 | 92.9% |
| Dax McCarty | 565 | 0 | 0.13 | 0.0% | 22775 | 15 | 10.75 | 139.5% |
| Dave Romney | 590 | 0 | 0.34 | 0.0% | 9873 | 4 | 4.40 | 90.9% |
| Tah Brian Anunga | 17 | 0 | 0.03 | 0.0% | 1044 | 0 | 0.09 | 0.0% |
Randall Leal and Alex Muyl are better than their career averages so far in 2021. Every other player is below his career average. Is there something endemic to Nashville SC’s poor finishing this year? Or is it just poor luck on low trials?
I think you’d have a hard time making a statistical case for the former, given the limited sample size to date. That’s not to say there’s a guarantee that it gets better, but regression (or progression?) to the mean is far more likely than continuing at the current rate.
That doesn’t mean you get to go back and say “actually the Cincinnati and Montreal games were wins, thanks.” It does mean that, as you project forward, it’s fair to expect players’ performances will be closer to the massive sample sizes they’ve already put together historically (including a couple thousand minutes in Gold for many of them) than the generally-poor output we’ve seen to date.
Across the way
Austin has built too much of its offense this season via the cross. Nashville’s defense is elite, in part, because it has relegated opponents to relying on crosses to generate their chances.
So what did we expect Sunday evening? And lo, it came to pass. Or cross. Whatever.
Nashville was able to force a team that’s already going to cross a bunch into generating basically all of its offense through that method. That’s not gonna be a real good time when you’re trying to get balls over and around Dave Romney and Walker Zimmerman.
Joe Willis is a very good goalkeeper, but it’s easy to put up clean sheets when the vast majority of opposing offense is generated via the cross: shots on the end of crosses are inherently low-probability when it comes to converting. I harp on it when Nashville is over-reliant on the cross, and so it’s only fair to credit the team when an opponent is forced into doing the same.
It probably comes as no surprise that Austin, despite attempting a boatload of crosses, managed to connect on very few of them for shooting attempts, and the lone attempt on-target was from distance, rather than any sort of generated chance:
That’s good defense, to me.
What did you see in the Opta data? Poke around and feel free to share your thoughts on social.
Club and Country 