Importance of 2nd game

[SNUSA]

Recently, SoccerAmerica put out (or quoted) a stat about how important the first game of group play is. 87 percent of winners go through to 2nd round, only 4 percent of losers do.

A friend of mine who is good with numbers said he wasn't surprised, and furthermore thought that the results would be very similar for the 2nd game of group play.

Since I had some time, I decided to take a look at that...

The percentages are portion of teams with the listed result that went to 2nd round. 1st game results is first column, then 2nd game results.

So, you suggested that Soccer America's stats on the importance of the first game of group play weren't that shocking, and that the game 2 stats would look similar.

I had a chance to check it out.

First number is how many go through with that result in the first game, second is that result in second game.

W 87% 70%
D 61% 62.5%
L 4% 15%

So... I'm really not sure what it means. First game seems to be more predictive. But I'm not sure how much of a statistical difference that is.

Anyone with thoughts on what this all means (besides "It's better to win")

[Vivo Sohando]

I'm not surprised by the wins, but by the draws. Your numbers say that nearly 2/3 of the teams that draw the 1st or 2nd game go through.

[USA2010?]

Keep in mind that, statistically speaking, these are not independent variables. If you lose the first game, you are probably one of the two weaker teams in the group, which would mean you would probably lose, or at least not win, the second game. Same logic for winners. One of the keys I see from the data is the importance of getting at least a point from each game. One win and two draws gets you through. Drop a point, and you're sitting on 4 - flip a coin at that point.

Still nice research.

[Iplayedforchelsea]

I hope we really play well in the second game.

[SNUSA]

USA2010 -- Yup.

I hope to have time tonight to look at 3rd games, which could be even more wonky since there's often a very different psychology involved -- that is, teams either already know they're in or out. Which doesn't stop them from playing hard, necessarily (for some reason, the Poles come to mind...)

If I get that done, though, we should be able to see what percentage of teams who win ANY game go through.

[SetPeace]

I'd like to see us bag two wins, then take off for the Ghana match and rest our starters (easier said than done). We got incredibly lucky to get out of our group in Korea on the last day of group play (a Portugal draw or win against South Korea could have sent us home early, and the Koreans would have advanced over us had they played for a tie). A win and 2 draws would be OK, but going 1-1-1 in this group could be risky in terms of advancing.

[troutseth]

The one thing I think the stats need to take into account; what happened in game one. The stats for game two are great, but what are the percentages of teams that advance with a win/draw/loss in game two AFTER, a win/draw/loss in game one. So in essence, you would have three different sets of percentages for the game two result. That would give you a better idea of "what are our chances if we XXXXX game one?"

My non-mathematical gut says we need to have at least 3 points heading into Ghana and then pull off the win. I think 4 points is a ticket home in this group for us unless we really start scoring more than we have (and can take home a tie breaker).

[SNUSA]

So, let's look at the third game, where things could be odd, because you have teams that only need a draw, teams that don't need any points, teams that are eliminated, etc.

We'll now have the result and percent that go through, 1st game, 2nd game, 3rd game:


W 87% 70% 78%
D 61% 62.5% 47%
L 4% 15% 23%

It seems this all does support the idea that the results matter a bit less (in a direct sense) as each game goes on (though as Bruce reminds us, the same results in opposite order would have gotten us through in 2002).

And a draw in the last game is much less helpful than elsewhere.

Another thing to look at:

Over the last two WCs, 14 teams have won a game and not gone through. Only 9 have lost any game and still gone through. I think that means that any team that has a loss has only a 28 percent chance (1 in 4, roughly) of advancing, regardless of the other results.

[And while I wasn't specifically looking for it, I think there was only one group winner with a loss -- Nigeria in 1998 (2W, 1L)]

Anyhow. Guess there's nothing too surprising there. But perhaps interesting. And it gave me something to do while waiting for the next Studio90...