FIFA 2014 World Cup - Official Discussion Thread

[7,052,770]

The U.S. now makes the knockout stage with a pair of draws or a win in either.

[7,052,770]

Excited to see the bro tears when USA finishes last in the group with two losses.

538 gives the US around a 2/3 chance of moving on.

[7,052,770]

Excited to see the bro tears when USA finishes last in the group with two losses.

538 gives the US around a 2/3 chance of moving on.

I saw that. Not only is it silly, but if there's one thing we know we cannot predict results from soccer rankings or whatnot.

His model takes a lot more into account than "soccer rankings." It looks at how the team has been playing together lately, how good the individual players are, homefield advantage, etc. You can read more about it here. It also adjusts some for play during the tournament, so Ghana playing well is going to raise their chances of advancing, and Portugal's huge loss hurt its chances.

I trust Nate Silver not to put something out there that's not complete garbage, and he even admits that it's very hard to predict these things.

[7,052,770]

Well all your predictors predicted Ghana would be losing horrendously, and look what happened...

538 said there was a 19% chance of a draw today. Just as in his election forecasts, he never makes predictions of who wins and who doesn't, he just gives probabilities. We can judge his accuracy over the course of the tournament by seeing if things he said had a 20% probability ended up happening around 20% of the time.

[7,052,770]

Well I'm rooting Ghana... because I'm Ghanaian. And I'm rooting against the US because Ghana cannot advance if the US does well. So that's that.

Wait, I thought you were an Ethiopian nobleman or something?

[7,052,770]

He's half Ethiopian half Ghanaian

Ah, ok. Well that all makes perfect sense then.

[7,052,770]

ahahahahaha

You really shouldn't gloat about a draw when you all you could talk about beforehand was how the US would definitely lose...

[7,052,770]

ahahahahaha

You really shouldn't gloat about a draw when you all you could talk about beforehand was how the US would definitely lose...

At least the Belgians can finish you off if the worst comes to fruition.

Maybe so. You still gloated after being wrong.

[7,052,770]

538 now has the U.S. probability of advancing to the knockout stage at 76%, up from 64% this morning.

[7,052,770]

538 now has the U.S. probability of advancing to the knockout stage at 76%, up from 64% this morning.

Wtf? This is plain idiotic. Any serious metric is going to show 40ish, maybe 50ish. I mean, whatever their possibility that the US wins yesterday and thus goes through automatically, is out of the equation now.

Here's the article: http://fivethirtyeight.com/datalab/how-the-portugal-draw-boosts-the-u-s-s-world-cup-advancement-odds/

[7,052,770]

http://fivethirtyeight.com/features/was-the-u-s-robbed-against-portugal-it-depends-on-what-time-means/

538 with a fascinating analysis of stoppage time.

[7,052,770]

(Source: 538)

[7,052,770]

English soccer fans taunting Germany about the World Wars was/is(?) a very common cheer when the countries played each other.

[7,052,770]

most of our players are young and promising (Yedlin, especially, has been sensational this World Cup) and will continue to improve.

What age do soccer players tend to hit their prime?

[7,052,770]

most of our players are young and promising (Yedlin, especially, has been sensational this World Cup) and will continue to improve.

What age do soccer players tend to hit their prime?

Usually their late twenties.

Ah, so two of our goal-scorers should still be improving 2 cups from now. Nice.


Nate Silver had an interesting piece on whether the USA has plateaued or will keep improving.

[quote author=538]From this perspective, the upward trend has been extremely steady. In fact, other than the 2002 team arriving in the quarterfinals a little ahead of schedule, it’s been almost perfectly linear. Since 1986, the U.S.’s Elo rating has improved by almost exactly 50 points every four years.

Imagine that the trend continues. Right now, the U.S. team’s Elo rating is nearly 1,850, which places it 15th in the world. Add another 50 points, and by 2018 it would be at 1,900 — somewhere around eighth or 10th in the world and near where Belgium and Uruguay and France are today. That’s a team that could be a dark-horse World Cup contender. By 2026, its Elo rating would be 2,000 — not far from where Argentina, Germany and the Netherlands began this year’s tournament.[quote]

[7,052,770]

This is my problem with Nate Silver. He presents his predictions as merely "probabilities." So he does not predict that Brazil will advance, but instead that Brazil has a 78% chance to advance or whatever. But if Brazil wins, he and the media applaud him for correctly predicting that Brazil would advance. But if Brazil had lost, he would have just said, "Ah, but I said there was a 22% chance that they would lose, so I'm not wrong!" To check whether or not his probabilities are right, you'd have to look at if, over time, a team with a 78% chance to advance does actually advance in 78 out of 100 games. You can't just say, "Oh, he got the winner in all 8 games right, he's a soccer prediction genius!"

Yes, absolutely that's how he should be judged. At the end of the World Cup, someone should analyze his probabilities for each game and see if, for example, he was "wrong" around 1/4 of the time he said there was a 75% chance. I suspect we'll find out his odds were very good.

[7,052,770]

For everyone's reference:



It's too small of a sample size to really judge how good his probabilities are, but on first glance they seem pretty good. He really blew Group D, but that's what happens when England is surprisingly bad AND Costa Rica is surprisingly good in the same group. You can see how his formula gave Costa Rica more respect as the tournament went on.

Here is a very rough breakdown at the levels of odds. It's still too early for any valid mathematical analysis (I guess I could run the Chi-Square test if anyone wants), but again his numbers appear to be doing pretty well:
CATEGORY   Y   N   PERC%
0-14         1   15   6.3%
15-24       4   26   13.3%
25-34      15   37   28.8%
35-44       8   17   32.0%
45-54       6    6   50.0%
55-64       5    1   83.3%
65-74      10    1   90.9%
75+          7    1   87.5%

[7,052,770]

An update for those interested:



You can argue the sample size is still too small, but 538's probabilities have been very good so far.

[7,052,770]

fivethirtyeight still has Argentina as a more likely champ than Germany even having to play an extra match. I assume they're not factoring in Neymar and Silva not playing in the semis though.

Probably not yet. I think they only recalculate stuff like that once per day.

[7,052,770]

Here are a couple more charts I calculated:

The average age of each team, weighted by minutes played:
AVG   COUNTRY
24.9   South Korea
25.3   Ghana
25.5   Switzerland
25.6   Nigeria
25.7   Belgium
25.8   Algeria
25.9   Australia
26.1   France
26.3   Bosnia-Herzegovina
26.3   Germany
26.4   England
26.5   Netherlands
26.5   Costa Rica
26.6   Chile
26.8   Cameroon
27.0   Colombia
27.1   Argentina
27.2   Brazil
27.5   Ecuador
27.7   Japan
27.8   Uruguay
27.9   Mexico
27.9   Spain
28.0   Russia
28.1   Croatia
28.4   Italy
28.5   Greece
28.7   Portugal
28.8   Iran
28.9   United States
29.0   Ivory Coast
29.2   Honduras

This is updated through today's matches.

[7,052,770]

And here is only weighted by only goals scored (not counting own goals or PKs after extra time):

AVG   COUNTRY
22.0   Cameroon
22.9   Switzerland
23.7   Belgium
24.0   Ecuador
24.1   Colombia
24.7   Nigeria
25.0   South Korea
25.2   Brazil
25.2   Costa Rica
25.2   Germany
25.5   Italy
25.7   Argentina
25.8   France
25.9   Algeria
26.0   England
26.0   Ghana
26.0   Iran
26.0   Ivory Coast
26.5   Chile
26.7   Greece
26.8   United States
27.0   Russia
27.2   Netherlands
27.3   Bosnia-Herzegovina
27.3   Uruguay
28.0   Croatia
28.0   Japan
28.3   Portugal
28.6   Mexico
30.0   Spain
31.0   Honduras
32.3   Australia

[7,052,770]

When taking margin of victory into account, yesterday's match was the least likely result in World Cup history