World Cup Thoughts – Day Eight
Japan vs Costa Rica, Belgium vs Morocco, Croatia vs Canada, Spain vs Germany
Another big day, lots to talk about, blah blah blah. Let’s get on with it.
Japan 0-1 Costa Rica
Content warning: maths
International football often doesn’t produce a lot of good chances per game.
Japan’s defeat to Costa Rica was a good example of that. Japan “won” the expected goals but, depending on the model you’re looking at, only created 0.6-0.9 xG. When looking at the FBRef model and feeding it into Danny Page’s xG simulator, you can see the problem. They score at least once 60% of the time, but that obviously leaves 40% of the time where they don’t find an equaliser.
Japan created 89% of the xG in this game, but the model estimates that they would only win from those chances 56% of the time. So, despite clearly dominating, the data predicts they’d go home disappointed nearly half the time. What gives? There are so few events in the game. It’s the law of large numbers in action. The fewer chances there are total in a game, the greater the odds that the team which didn’t create as many can win the match. If we double the chances each team took in the match, Japan’s win probability goes up to 78%. If we quadruple them, the Samurai Blue win 93% of the time, despite still only having 89% of the xG.
So should Japan have been more aggressive in their strategy? Yes, but I don’t think it’s a matter of simply turning up the sliders. Japan have spent recent years becoming more solid and hard to beat. They conceded just twice in eight qualification matches. This is a Japan team probably better suited to giving superior sides a tough time, but we saw the flip side here of a low-variance approach. That might be useful in the knockout stages, but they’re now only at a 20% chance (according to FiveThirtyEight) to make it through.
Belgium 0-2 Morocco
Belgium are Eden Hazard’s most famous feature in this tournament.
Roberto Martínez at least seemed to accept this in moving to a 4-2-3-1 shape. Axel Witsel and Amadou Onana made at least some sense as a double pivot and could have offered better protection to the back four. Youri Tielemans was poor against Canada, though he feels like someone who Martínez should be working to get more out of. But the attack didn’t make sense to me. I’m not sure what Eden Hazard does that makes more sense as a ten, and Kevin De Bruyne on the right just limits his involvement. I would’ve gone with a 4-3-3 and used De Bruyne as a central midfielder.
Belgium were fairly positive early on, getting a few decent chances, but then it felt like they just succumbed control of the game to Morocco. This is another good case of xG not quite lying but failing to tell the whole story. Could Michy Batshuayi have scored an early goal? Yes. Did Belgium do a good job of controlling the situation after that? No.
In the first 20 minutes, Belgium led the xG by 0.46 to zilch. In the rest of the game, Morocco led the xG by 0.58 to 0.24. So Belgium created the majority of their good chances in the first 20, then produced very little for the rest of the match while Morocco were gradually turning up the heat. This was something of a smash-and-grab, but Belgium didn’t exactly do enough to definitively put their stamp on the game. Yes, on another day they’d win this, but it’s hard to argue they were remotely good.
Everyone’s talking about how old they are and I agree. But they’re still trying to play like a younger team. Is it not time to drop into a lower block? Martínez doesn’t have to give up his principles in possession, but he needs to protect those ageing centre backs better than he typically does.
This result now puts Morocco as the favourites to qualify from Group F, at 91% according to FiveThirtyEight. A draw against Canada would guarantee it, but there is still a chance of going through even if they lose. Belgium conversely are down to 33%. It’s very likely they’ll need to beat Croatia to do it. That could definitely happen, but one in three feels about right.