Union Run Out of Luck
I thought the U were leading a charmed life in Orlando as they worked their way through the group stage and into the semi-finals. That luck ended in a 1-2 loss to eventual tournament winner, the Portland Timbers. An unmarked Ebobisse gave the Timbers the early lead. Sergio Santos, after the softest of fouls in the box, skied his PK over the cross bar late in the first half. Both Dennis and Jeff H say they saw that miss coming. Blanco doubled the lead in the 70th minute and the Union looked dead in the water. Andrew Wooten made the finish interesting after he cleaned-up Monteiro's saved free kick in the 85th minute but that was as close as they got. In the end, the result seemed more than fair to me.
A few days later, Portland slipped past upstart Orlando in the final 2-1. Can't say I remember much from the match. I recall thinking that Orlando did a good job here. They had more possession, shots were 14-13, expected goals were 3.3-2.2. Part of the problem was they only managed one shot on target.
 |
| Jeff H will not recoup the cost of his new Subaru through savings on parking at Union matches |
They now proceed with a regular season that is not fully laid out yet. Games are set up through September 16. I haven't checked all the schedules but for the Union it means six more matches. Their season will resume on Thursday at Gillette Stadium (arguably second worst MLS venue behind only Yankee Stadium) against New England. Though the matches will be played in the home parks, few venues will be allowing fans. This is unfortunate since Jeff H recently purchased a Subaru with the expectation that it would mean free parking for Union home matches.
The View from Europe
Just as everyone predicted, the Champions League semis will feature two sides from Germany and two from France. Right. This is first time since 1996 there are no English or Spanish teams in the semis and the first since 1991 with no English, Spanish or Italian teams. How did this happen?
The Cinderella side of Atalanta seemed poised to steal a berth from PSG after Neymar missed at least three wide open opportunities. The Italian side had clearly been outplayed and frankly did not distinguish itself in wracking up 29 fouls. But there they were at 89 minutes with a 1-0 lead. Unfortunately for them, a match is 90 minutes plus stoppage time. Their downfall is chronicled here. Eric Maxim Choupo-Moting, subbed on in the 79th minute, plays a key role in Marquinhos' equalizer in the 90th minute and then scores the game winner at 93 minutes.
On Wednesday, Atletico Madrid might view themselves unlucky losing to RB Leipzig 1-2 on a deflected shot by American Tyler Adams in the 88th minute. On the other hand, the Mattress Makers (no, that really is Atleti's nickname) didn't look to be at their best.
Thursday's match was - wait this must be a typo - Bayern 8 Barcelona 2? When the match began with the teams trading goals in the first 10 minutes, hopes were high for a classic between two high profile sides. Then Bayern pushed through three more by the 31st minute. Suarez got one back and at 4-2, maybe some hope of exciting finish. Nope. Bayern poured in four more as the match took on a practice session quality. The German side was a clear favorite but not like this. Discussions of a Barcelona make over dominated the late stages of commentary.
In the last quarterfinal, Man City entered as heavy favorites over Olympique Lyonnais and mostly outplayed them so of course they lost 1-3. Some serious smash and grab counterattacking from the French meant that City's statistical advantages like 72/28 possession, 18/7 shots, 11/3 corners and 2.2/.7 expected goals didn't mean much. A sad footnote for the match, it was the last appearance of David Silva in a City uniform and we probably won't see him in the EPL next year.
So the semis are set with RB Leipzig - PSG on Tuesday and Lyon - Bayern on Wednesday. PSG are prohibitive favorites and Lyon are heavy underdogs to Bayern. Both matches are on CBS All Access (you can sign up and get a free month). You can also see them on CBSSports (Verizon channel 594 in Philly) but on same-day delay.
Meanwhile, over in the
Warning: Statistical Wonkery Ahead
I cannot remember a season with more improbable results for Newcastle. How about the match with Everton when Florian LeJeune of all people scored two in stoppage time to snatch a 2-2 draw? Or the time Tottenham ran Newcastle ragged and came away with a 0-1 loss? There are some hard luck results too but nowhere near as many, at least based on my recollection. Support for that view comes from 538, which had Newcastle ranked 19th, ahead only of Norwich. So what gives? Were the Magpies lucky? Was there more than meets the eye?
I dissected the season using a couple of tools - expected goals data available at 538 (on downloadable Excel spreadsheet no less!) and a formula I developed to estimate how many points a team should earn based on their goals scored and allowed (call it ETP for now). With the expected goals data, we can see whether Newcastle scored or gave up more or less than they should have based on the types of shots taken and non-shooting actions (methodology is explained here). Of course we can debate whether differences between actual and expected reflect luck (good or bad) or some inherent trait in the team. With the point formula, we can get a sense of whether a team the benefited or was hurt by the distribution of the goals across the games. To see how that matters, say a team scored two and allowed two over three games. They could come away with three, four or six points from those three games depending on how the goals were distributed. Again, we can debate how much of that is luck or "clutch" performance.
On the expected goals front, Newcastle scored four more than expected and allowed 10 less than expected. If I do a game-by-game retally based on expected goals (calling matches with less than a .3 difference a draw), I get Newcastle with just 16 points. That's the biggest difference in the league. All teams results are presented below.
| Expected | Actual | Difference | |
| Manchester City | 99 | 81 | -18 |
| Liverpool | 97 | 99 | 2 |
| Chelsea | 92 | 66 | -26 |
| Leicester City | 75 | 62 | -13 |
| Manchester United | 72 | 66 | -6 |
| Wolverhampton | 63 | 59 | -4 |
| Southampton | 59 | 52 | -7 |
| Everton | 56 | 49 | -7 |
| Watford | 56 | 34 | -22 |
| Arsenal | 50 | 56 | 6 |
| Tottenham Hotspur | 45 | 59 | 14 |
| Brighton and Hove Albion | 44 | 41 | -3 |
| Burnley | 44 | 54 | 10 |
| Sheffield United | 44 | 54 | 10 |
| Crystal Palace | 33 | 43 | 10 |
| West Ham United | 30 | 39 | 9 |
| Aston Villa | 28 | 35 | 7 |
| AFC Bournemouth | 27 | 34 | 7 |
| Norwich City | 21 | 21 | 0 |
| Newcastle | 16 | 44 | 28 |
But it looks like the distribution of expected goals was unlucky. A team that scores 34 and allows 68 should get about 27 points. Even allowing for that though, Newcastle would still have finished dead last.
So what about the differences between expected and actual? There has been talk about how good Dubravka is - maybe that's part of it. Turns out that is entirely possible. In the 18/19 season Newcastle allowed 12 fewer goals than expected. In the 17/18 season they yielded five fewer than expected in the 12 matches he was between the sticks. In the other 26 matches that season, they were seven better. So this looks to be a consistent thing with Newcastle that Dubravka has accentuated.
On the offensive side, there is no consistency. In 17/18 they scored five less than expected, in 18/19 they were one goal better. So I started to play around with using expected scored but actual allowed. That would mean 34 scored and 58 allowed. With ETP, that says they should end up with about 31 points. But if I go through game by game and retally using expected scored and actual allowed, I get 46 points. In short, it wasn't only that they were yielding fewer goals than it expected, it was also when they didn't allow those goals.
I'm concluding that about half of the difference in the point total looks to be related to a defense that consistently outperforms expectations but the other half was a lucky distribution of goals allowed. That would say that 30 points is about right. For a smell test I went through match-by-match and based on my sense of the game, what was a realistic outcome. I got 31 points. There are just too many matches like the 1-0 wins over Man United and Chelsea in which the opposition played well enough to have earned at least a draw. My conclusion is that this is a not a mid-table squad and absent some serious upgrades, next year will be another slog to avoid relegation.
Ironically, the only signing to date is Mark Gillespie, a goalkeeper. But we have heard about interest in Bournemouth's Callum Wilson and Joshua King. That might help. I wouldn't have minded getting Adam Lallana but Brighton already snatched him up.
Two notes on the table above. Check out Chelsea's "underperformance" by 26 points. That is largely driven by allowing 15 more goals than expected. It is possible that Chelsea's save percentage of 54.8% (the lowest in the league by almost 10%) had something to do with that. The correlation between the difference of expected to actual goals allowed and save percentage is 86%, as in save percentage explains 86% of the variation in the difference. Certainly Kepa has his critics, though some wonder about the defense in front of him.
And then there is demoted Watford, a massive 22 points worse than projected. They scored 15 fewer than expected, the largest by far but also yielded almost eight more than expected. Haven't gone through their season in detail so I haven't sorted out how much might be simply bad luck. But this is a side that had done well in their first two seasons back up in the EPL.
Okay, enough of the wonkery...though I do hope to continue going over this data for insights into how the game works. I blame 538 for making the data so accessible.
Go enjoy some of the August NHL action.
No comments:
Post a Comment