世界杯德国队小组垫底出局，投行预测模型大失准头

多家投资银行都在尖端的计算机程序上投入巨资，期望可以更好的预测股市，尽管股市在本质上有不可预测性。为了展现算法的厉害，有些金融机构开始预测2018世界杯的冠军。

瑞士的瑞银集团（UBS）和德国商业银行（Commerzbank）的预测却掉链子了。世界杯开赛前，两家银行都用计算机模拟所有比赛并得出结论，最有可能夺冠的是：德国队。

排名世界第一的德国队，却出人意外地在周三以0：2输给了韩国队，这是本届世界杯迄今为止最大的冷门。德国以卫冕冠军的姿态参加本届杯赛，结果不但出局，甚至未能进入16强。

其他投资银行的预测是对是错，还有待时间的检验。高盛（ Goldman Sachs）此前也模拟了100万场比赛，并预测今年巴西会夺冠（同时也预测说德国进决赛），荷兰国际集团（ING）则预测西班牙最终夺冠。

让华尔街挽回一些颜面的，是学术界的模拟计算也认为德国夺冠几率最大，至少德国能进决赛。阿德莱德大学的一项模拟计算认为德国有13%的几率夺冠，因斯布鲁克大学的另一项模拟计算认为巴西有17%的夺冠几率，紧随其后的是德国16%。

这些大学比瑞银聪明的地方或许在于预测的方式：比如说德国夺冠概率是1/6，一旦不准还好说一点，而直白地预测说冠军是谁，一旦预测破产，就挺尴尬了。

不知道有哪台计算机能跟章鱼保罗的未卜先知能力有一拼，保罗曾预测14场世界杯比赛的胜者，对了12场。2010年保罗去世了。其他小动物比如名叫阿喀琉斯的小白猫，也在预测本届世界杯的比赛，不过它们似乎都比不上保罗。

德国队的失败，意味着自1938年以来，德国第一次在世界杯上这么早就出局。过去的20年里，卫冕冠军在小组赛即遭淘汰的有：2002年的法国，2010年的意大利，2014年的西班牙，还有今年的德国。（财富中文网）

译者：宣峰 

Investment banks have spent heavily on sophisticated computer programs, hoping to better predict stock market that is inherently unpredictable. To show off their algorithmic prowess, some financial institutions set out to predict the winner of the 2018 World Cup.

For Switzerland-based UBS and Germany’s Commerzbank, that didn’t work out so well. Before the game started, the banks ran computer simulations of World Cup games and both concluded that the likely winner would be… (drumroll)… Germany.

The top ranked country suffered a stunning 2-0 defeat on Wednesday to South Korea, the World Cup’s biggest upset so far. Germany entered the tournament as its defending champion, but the team was not only eliminated from the World Cup, it didn’t even secure a place in the round of 16.

Predictions from other investment banks have yet to be proven right or wrong. Goldman Sachs ran a simulation of 1 million games and predicted that Brazil would win the World Cup this year (although it also predicted Germany reaching the finals). ING, a Dutch bank, said its models foresaw Spain winning the World Cup this year.

In fairness to Wall Street, simulations that were run in the academic world also saw Germany as a favorite to win or at least reach the finals. One simulation by the University of Adelaide gave Germany a 13% chance of winning, while another at the University of Innsbruck reckoned the Brazil had a 17% chance of winning, with Germany close behind at a 16% chance.

Perhaps the edge that universities have over UBS was in the way they presented their predictions: saying Germany has a one in six chance of winning the World Cup doesn’t look so bad today, while flat-out predicting the winner isn’t a good look.

It’s still not clear whether any computer can match the World Cup divination powers of Paul the octopus, which correctly divined 12 out of 14 World Cup winners. Paul passed away in 2010. Other animals, such as Achilles the cat, are trying to predict World Cup games this year but none so far can lay claim to Paul’s throne.

Germany’s loss marks the country’s earliest exit from the World Cup since 1938. In the past 20 years for defending champions have been eliminated in the group phrase: France in 2002, Italy in 2010, Spain in 2014, and Germany this year.