【英语科技】为什么我们的预测成功率不及章鱼保罗？

On the face of it, predicting the World Cup looks easy.

Someone from Europe or South America will win. The runner-up will hail from one or the other of those continents. Long before the final game, the original field of 32 teams is gone nearly in its entirety.

In South Africa, the 2010 World Cup has stayed true to that script. Two European teams -- the Netherlands and Spain -- will play for the trophy in Sunday's final at Johannesburg's Soccer City stadium.

But success at the local sports book is in the details, and this is where World Cup predictions hit the rocks. Unlike baseball, which after 100 years has a series of formulas that reliably project success on the diamond, football's number crunchers have yet to master the statistical formula for World Cup success -- us included.

Our methods, which used a weighted mathematical formula that combined four major factors in World Cup success -- a team's current makeup, historical performance, goal differential and location -- delivered mixed results. On the plus side, we correctly predicted 11 of the 16 teams that advanced out of the group stage. Not so good: putting England in the semifinals and giving Brazil the trophy.

Like all those financial experts who in hindsight see the global financial crisis as inevitable, our errors now look crystal clear.

The highly-paid English players, for instance, aren't nearly as good playing with one another as they are in the Premier League, where the stars of Brazil, Argentina, Spain and Africa compliment or even carry them. The French, who have ping-ponged from the final to first-round elimination for four straight World Cups now, are never to be trusted.

So, like those financial experts, we're going to start by blaming the data, or rather, the lack of good data, as well as a few other idiosyncrasies of the world's most popular sport.

1) What makes the World Cup special is that it's the only time when a large group of countries from across the globe play one another with so much at stake. That's also what makes it so unpredictable. No one really knows how Italians will fare against Kiwis until they face each other.

Between now and 2014, nations will mostly compete within their regions, thus providing few hints of how they compare. The Yankees play the Red Sox 19 times each season before the playoffs, but other than a possible exhibition, Brazil won't likely play the Netherlands again for four years, by which time the rosters will likely be completely different. Sure, it's always a smart bet to take Germany over England, but not based on the Germans' semifinal win in 1990.

2) Qualifying results don't tell you anything -- except for when they do.

The problem with paying significant attention to World Cup qualifying games -- as we did -- is that they often involve mediocre, irrelevant competition. This is how England duped us all. The English blazed through qualifying, outscoring the opposition 34-6, but much of that margin was run up against helpless Andorra, Belarus and Kazakhstan. The only semiformidable teams England faced were Croatia and Ukraine. Thus there wasn't much indication of how England would fare against top competition. (Or maybe there was: England didn't even qualify for Euro 2008.)

But qualifying isn't useless with respect to South America, where nearly all of the teams are competent. Eight of the 10 Conmebol teams are in FIFA's top 50, including four of the top 20. So we should've given more credit to Uruguay -- whom we had losing in round one -- just for getting out of there alive.

3) Injuries aren't that big of a deal. Unlike other sports, football teams can figure out how to make up for the loss of a crucial player, especially in those countries with a deep talent pool. We downgraded Ghana because it was without Michael Essien. Same goes for Germany, which was without captain Michael Ballack. Both teams were clearly deep enough to overcome the injuries.

4) Location matters a little, but not that much. Brazil and Argentina are supposed to win World Cups outside Europe, and Europe is supposed to struggle off home soil. To that end, we were feeling pretty good about ourselves until the quarterfinals, when Europe won three games and clearly sent a message that it can do just fine in foreign lands.

5) The World Cup may have the slimmest margins in sports. Particularly this one, with the tournament scoring average challenging the all-time low of 2.21 in 1990.

With goals at such a premium, the margin between a win and a loss is thin enough to defy trends, especially those based on a quadrennial tournament with tons of player turnover. The U.S. advanced on the strength of an unspeakable goalkeeping error by England's Robert Green and a rocket from Algeria that clanked off the post instead of hitting the back of the net. Ghana made it through on the strength of a late penalty kick against Serbia to secure a 1-0 win, and on a tie against an Australian team that barely missed several game-winning chances.

'Small details in the World cup can send you home,' Spain's Andres Iniesta wisely remarked earlier this week.

That said, predicting 11 of the final 16 teams isn't a bad result, considering the thin margins that separated those that advanced from those that went home early this year. We ignored widespread optimism about the rise of African football and predicted just a single African nation would advance. Though we got the country wrong -- we said Nigeria, not Ghana, would go through -- our numbers correctly predicted Africa's struggles on home soil.

In the quarterfinals, we had Spain dusting Portugal, Brazil thumping Chile and Argentina sending Mexico to its usual second-round exit. All of those prophecies came true. We correctly had the U.S. departing in the round of 16 (albeit to Germany) and Spain lasting until the end.

But we flubbed on England and France, misread Italy's age as valuable experience and were wrong about Serbia, which looked strong during qualifying play. We figured Japan and South Korea couldn't succeed outside their home nations, and we still can't figure out how Slovakia slipped through.

If you can, please let us know.

预测世界杯足球赛的结果表面上看起来很容易。

来自欧洲或南美的某个国家将会捧杯。而获得亚军的也将是一个欧洲或南美国家。

Associated Press预测极其准确的章鱼保罗2010年的南非世界杯也是这种情况。荷兰队和西班牙队这两支欧洲球队周日将在约翰内斯堡的足球城(Soccer City)体育场争夺冠军。

但预测体育比赛结果的成功存在于细节中，而细节正是世界杯预测的短板。棒球就不同了，经过100年的发展，它已形成了一套可以有效预测场上比赛结果的公式。而足球在技术统计方面还拿不出一个可以成功预测世界杯比赛结果的统计公式，我们也做不到这一点。

我们的方法是，使用一个加权数学公式来预测世界杯比赛结果，这一公式体现了在世界杯比赛中取胜的四大因素──球队目前的组成、以往成绩、净胜球数和比赛地点，预测结果令人喜忧参半。从好的方面说，从小组赛中晋级的16支球队有11支都被我们预测对了。不太好的一面是，我们错误地预测英格兰将杀入半决赛，巴西将会夺冠。

神奇的章鱼保罗成功地预测了德国队在南非世界杯上六场比赛的结果。以下视频展示了它为德国对阵西班牙的半决赛做预测时的情形。就像所有事后诸葛亮般宣称此次全球金融危机不可避免的金融专家一样，我们的错误现在看来十分清楚。

比如说，那些高收入的英格兰球员彼此间配合的默契程度并不像他们在英超联赛中那样好，在英超联赛中会有一大帮巴西、阿根廷、西班牙和非洲球星围着他们转。而在最近四次世界杯比赛中，法国队既有杀入决赛的时候，也有小组赛就被淘汰的时候，它永远无法让人信赖。

所以，像那些金融专家们一样，我们也要从批评手头的数据开始，更确切地说，抱怨手头缺乏良好的数据，并抱怨足球这项世界最流行运动的其他一些特征。

会预测世界杯比赛结果的神奇动物不只章鱼保罗一个，新加坡的长尾鹦鹉马尼(Mani)也为世界杯决赛算了一卦。它“说”，荷兰队将成为今年的冠军。1) 足球世界杯赛之所以特别，是因为它是唯一一项世界大批国家汇聚在一起押下如此重注的比赛。而这也大大增加了世界杯比赛的不可预测性。在意大利队和新西兰队开始比赛前，没人真的知道两队会有何种表现。

从目前到2014年这段日子，报名参加下一届世界杯赛的各国将主要在各大洲内部进行比赛，因此提供不了什么可借以将他们相互比较的线索。在美国棒球职业大联盟的联赛中，纽约扬基队(Yankees)和红袜队(Red Sox)在决赛中相遇前，相互间已进行过19场赛季常规赛，但巴西队与荷兰队除非打表演赛，它们未来四年内不大可能再同场竞技，而届时两队的球员可能已与本届世界杯赛完全不同。当然，打赌德国队将战胜英格兰队总会是聪明的做法，但德国在1990年世界杯半决赛中战胜英格兰的事实却不能成为下赌注的基础。

2）从资格赛的比赛结果中得不出任何结论──只有事后才能得出结论。

如果像我们那样把大量注意力集中在世界杯的资格赛上，问题就在于其中涉及的竞争对手常常平庸到不重要。英格兰队就是这样让我们所有人上当的。资格赛期间英格兰队势如破竹，进了34个球，只输了六个，但净胜球当中有很多是从安道尔、白俄罗斯和哈萨克斯坦这些弱队身上拿到的。英格兰队遇到的还算比较强大的队伍，也就只有克罗地亚和乌克兰。所以从中很难看出英格兰队跟顶级对手比起来表现会怎样。（或许有一点可以说明问题，2008年英格兰队连欧洲杯的参赛资格都没有拿到。）

但在南美球队身上，资格赛比赛结果并非没有用处，因为这个地区几乎所有球队都很强大。南美洲足球协会(Conmebol)10支球队中有八支都名前国际足联排名前50位，其中四支球队位列前20名。所以，我们当初起码应该相信乌拉圭队晋级到16强，仅凭他们能入围这一点。

3）伤病情况并不那么重要。和其他体育比赛不同，足球球队在损失一位核心球员之后，可以想出补救办法，特别是在那些后备人才充足的国家。我们调低了对加纳队的预测，是因为它损失了埃辛(Michael Essien)，同样调低了对德国队的预测，因为它的队长巴拉克(Michael Ballack)没有上场。两支球队的后备力量明显都足以弥补这些伤病情况。

4）地点因素有一定作用，但作用不是那么大。本来，巴西和阿根廷应该是要赢得欧洲以外的世界杯的，而欧洲球队远征海外会出师不利。在这方面，我们一直对自己的预测相当满意，但到四分之一决赛的时候就不是这么回事了。在这一阶段，欧洲球队赢得了三场比赛，清楚地告诉我们，他们在异域照样可以做得很好。

5）在各种体育比赛中，世界杯的净胜球数量或许是最低的。特别是这一届世界杯期间，平均净胜球数量有可能比1990年创下的历届最低2.21粒还要低。

净胜球数量这么低，足以打破固定趋势，何况世界杯四年才举办一次，期间还有大量球员进进出出。美国队晋级，是因为英格兰队守门员格林(Robert Green)的惊天“黄油手”，以及阿尔及利亚的一脚远射击中门柱而未能入网。加纳队晋级，是因为它在最后时刻靠点球以1比0的比分打败塞尔维亚，并且和连失多个进球机会的澳大利亚打成了平局。

西班牙队的伊涅斯塔(Andres Iniesta)本周早些时候很有道理地说，世界杯当中的小细节就可以打发你回家。

不过呢，16强当中有11支球队都被我们预测到了，这个成绩还不算坏，何况净胜球数量很低，就是靠这点微薄的优势，一些球队晋级，另一些球队在今年提前打道回府。我们忽略了人们对非洲足球崛起的普遍乐观，并预测只有一支非洲球队晋级。虽然预测的是尼日利亚而不是加纳，具体球队搞错了，但我们还是准确地看到了非洲球队在本土的艰难。

我们预测，在四分之一决赛中，西班牙将打败葡萄牙、巴西打败智利、阿根廷打败墨西哥，让墨西哥像以往那样在小组出线后就出局。这些预测都对了。我们还成功地预测到美国将在16强比赛中出局（只不过以为是输给德国），而西班牙将坚持到最后。

但我们对英格兰队和法国队的预测栽了跟头，并错误地把意大利队的年龄看成一种宝贵的经验优势，也过于看好资格赛期间看起来很强的塞尔维亚队。我们以为日本和韩国都不会在本国以外的地方赢球，而且到现在我们也搞不清楚斯洛伐克是怎么打进16强的。

如果你搞清楚了，不妨告诉我们。