Avoiding the Drop

After the World Cup draw in December, I did a prediction post on the World Cup. This post wasn’t based on how I saw each group ending up or who I wanted to win; rather, it was based solely on the multiple regression metric found in “Soccernomics”.

Love it or hate it, this metric does at least lay forth an idea: a national team’s success or failure can be directly linked to how well their respective country compares to their opposition in the following three areas: team experience, gross domestic product per capita, and population. So I took their goal values for rank, went through each World Cup group, and figured out who SHOULD win each game of the group stage.

While I’m not so naive as to think that this is how the World Cup will play out, it did bring up some interesting scenarios (for example, if I’m doing this right then Japan should win group five over the Netherlands). But what’s to say that I’m doing it right? And just how accurate is this measurement anyways?

Well, we’ve got one way of checking: let’s use the African Cup of Nations and see how well the metric predicted winners.

I mentioned the areas of comparison above, but these are the specific weights we’re using for each piece of information:

• Home field advanatage is worth 2/3 of a goal.
• Having twice as much international experience as your opponent is worth about 1/2 a goal.
• Having twice the population of your opponent is worth about 1/10 of a goal.
• Having twice the GDP per capita of your opponent is also worth about 1/10 of a goal.

It should also probably be mentioned that we’re obviously distorting the intended use of this information a bit; while this is helpful for determining long term results (and is fairly accurate), no metric can accurately predict what actually happens on the field in a given game. You might be able to tell at the beginning of the season that you’ll score fifty goals over the course of a year; what you can’t tell is which games you’ll score them in. I’m essentially abusing a statistical model for my own amusement.

That being said, it’s possible that it still works in one-off situations. Let’s check it out.

Group A
Algeria, Angola, Malawi, Mali

Ultimately, this group was won by Angola, with Algeria as the runners-up. Mali was kind of the dark horse based on the quality of their midfield, and while they were actually ahead of Algeria on goal difference they lost on head-to-head results (the first tiebreaker in the Cup of Nations).

That’s reality, though. In our imaginary world, this group ends like this (the fake results are in italics):

Match #1: Angola 2-0 Mali
This match (the opening match of the tournament) actually ended 4-4, with Mali storming from behind in the last ten minutes to score all four of their goals. “Soccernomics” agrees with the first seventy-five minutes; Angola should’ve won this.

Match #2: Malawi 0-2 Algeria
Algeria outmuscles Malawi on every single statistical model and should have won this. However, they actually lost the match 3-0.

Match #3: Mali 0-1 Algeria
For the first time, the metric and the real life result agree; Algeria really did win this.

Match#4: Angola 2-0 Malawi
The metric agrees with real life here, too; the issue for Malawi is their insanely low GDP per capita of $835 per year. That probably effects them in a ton on non-soccer related ways, of course, but it’s also killing them in my imaginary soccer results thread.

Match #5: Angola 1-1 Algeria
“Soccernomics” correctly predicted a draw here, although in reality it was scoreless; Mali, incidentally, hinted that this match was fixed (a draw ensured that both teams would be guaranteed to advance out of the group).

Match #6: Mali 0-0 Malawi
Neither one of these teams really holds any edge over the other, and a scoreless draw made the most sense. Mali, however, ended up winning this 3-0 in real life. That means that twice in this group (this match and the opener) Mali performed significantly better than one would think they’d be able to.

In the end, however, “Soccernomics” agrees: Angola should’ve won this group, while Algeria should’ve come in second.

Group B
Burkina Faso, Côte d’Ivoire, Ghana

This is a rough group for Burkina Faso; everything’s pretty even between these three countries in terms of GDP per capita and population, but Burkina Faso’s got about half of the international experience of the other two.

Match #1: Côte d’Ivoire 1-0 Burkina Faso
This game was actually a scoreless draw, and the metric kind of supports that result as well. I’ve been rounding up (you can’t score half a goal), and Côte d’Ivoire’s got twice the experience, more money, and more people; all told, I figure rounding up is worthwhile. However, outside of experience Côte d’Ivoire doesn’t dominate Burkina Faso in any way, shape or form, so a scoreless draw could also be a decent result. That said…name one player on Burkina Faso. Côte d’Ivoire should’ve won here.

Match #2: Côte d’Ivoire 0-0 Ghana
Ghana’s got the edge in experience and population, while Côte d’Ivoire has the edge in GDP…but it’s not enough of an edge to grant either of them a goal. In reality, this match was a 3-1 win to Côte d’Ivoire, with Ghana’s loan goal coming as a stoppage time penalty at the end of the match.

Match #3: Burkina Faso 0-1 Ghana
See the notes for Match #1, as this is pretty much the same. The only difference is that Ghana really did win this game 1-0 in real life.

So, ultimately, Côte d’Ivoire and Ghana would advance (they’d also be in a statistical dead-heat as far as tiebreakers are concerned, meaning that they’d decide the winner of the group based on the fewest number of yellow and red cards; I’m giving the group to Côte d’Ivoire, since when I played as Ghana’s manager on FM they were always getting stupid red cards.

Group C
Benin, Egypt, Mozambique, Nigeria

Let’s take this moment to talk about how much population can figure into the metric. Nigeria’s got 154.7 million people living there; Benin has 8.7 million. Guess which one of those teams has a better international record? Better international records translate into more matches being played (if you qualify for more tournaments you’re automatically going to play more games); that’s true of Nigeria, who’ve played three and a half times more matches than Benin. So guess which team isn’t going to score in this group?

Match #1: Egypt 0-0 Nigeria
…except for when they face Nigeria. Nigeria’s managed to cancel out Egypt’s GDP advantage by straight-up doubling their population. Their caps are about the same, too, so Nigeria is actually the one team in Africa that had a chance to not get pummelled. In reality, however, the Pharoahs won this 3-1.

Match #2: Mozambique 0-0 Benin
This match actually ended up 2-2; neither team is really set up to compete in this tournament based on the “Soccernomics” rules for success.

Match #3: Nigeria 4-0 Benin
Nigeria has 17.6x more people than Benin, and three times the number of international matches under their belt. Benin makes a little more money, but it’s not enough to get them even a tenth of a goal. In reality, this match ended 1-0 to Nigeria; still a win, but not as convincing (especially since the goal came from a penalty kick).

Match #4: Egypt 3-0 Mozambique
Egypt’s got a lot more money, experience, and people than Mozambique, so they win this pretty easily. This game actually ended 2-0 to Egypt.

Match #5: Egypt 3-0 Benin
Egypt’s smaller population prevents them from making this worse against Benin than Nigeria did. In reality, Egypt won this 2-0.

Match #6: Nigeria 3-0 Mozambique
Mozambique’s bigger, which cuts into Nigeria’s main advantage against other teams in the tournament. This match really did end 3-0 to Nigeria.

So based on goal difference (given to them by population), Nigeria wins the group with Egypt in second place; in reality, Egypt won the group without losing a match, while Nigeria came in second.

Group D
Cameroon, Gabon, Tunisia, Zambia

I can’t name an international Zambian player playing abroad, but they’ve played a ton of matches; in fact, only Egypt has more experience than Zambia. That’s not to say, however, that Zambia’s a successful country; they’ve never been to the World Cup and have only been runners-up for the Cup of Nations twice. They do, however, play a lot of games; in reality, that actually ended up paying off for them, as they won this group on tiebreakers over Cameroon and Tunisia after not losing a match.

Match #1: Cameroon 2-1 Gabon
Gabon makes a LOT of money for an African country; their GDP per capita is $14,545. That’s outweighed slightly by the fact that there 1,475,000 people living in Gabon, which means that any advantage they’d have on GDP is outweighed by the disadvantage they suffer on population. Gabon did win this match 1-0 in reality, however, which must’ve really pissed Samuel Eto’o off.

Match #2: Zambia 0-1 Tunisia
Tunisia’s got the second highest GDP per capita in the tournament; at $8,002, they’re making seven times more money than Zambia. In reality, however, this match ended as a 1-1 draw.

Match #3: Gabon 0-2 Tunisia
Gabon’s money is cancelled out by Tunisia’s money; once experience and population come into play, Tunisia are the clear victors. This match ended up as a scoreless draw when the teams actually met.

Match #4: Cameroon 0-0 Zambia
Neither team really has an edge on paper; in reality, the presence of much better players on the Cameroon doesn’t show up on paper and Cameroon won the match 3-2, but it took an 86th-minute goal for Cameroon to break a 2-2 deadlock.

Match #5: Gabon 1-1 Zambia
Gabon’s monetary advantage really shines here; they make fourteen times more money than Zambia. Unfortunately for them, Zambia’s experience and population cancel out Gabon’s solitary goal; on paper it’s a 1-1 draw, but in reality Zambia won 2-1.

Match #6: Cameroon 1-1 Tunisia
Experience is equal; Tunisia makes more money, Cameroon has more people. This game ended as a 2-2 draw in real life, too.

On paper, Tunisia should’ve won and Cameroon should’ve come in second. In reality, Zambia won (on goals scored; Zambia, Cameroon, and Tunisia were all even on four points), with Cameroon in second place.

So how did the metric fare when compared to the actual results? Let’s look.

Correct Score: Four Matches (19%)
Correct Result: Ten Matches (48%)

All told, that means that the “Soccernomics” method was right a little less than half the time when compared to actual results from the tournament. That’s better than 33%, which (since each match has three possible outcomes) means it’s better than rolling a die; I’m not sure I’d use it much past that, however.

What is interesting is that it was very accurate when predicting results on the long-term. Seven out of the eight to advance from the group stage were correctly selected using the “Soccernomics” method; in fact, only Group D – which was decided between three teams on the seventh tiebreaker – showed a variance. So on the longer timeline, the metric works pretty well.