2018年2月26日纽约时报：算法PK猫咪：预测世界杯桂冠花落谁家
How algorithms and cats predict the winners of the World Cup
算法PK猫咪：预测世界杯桂冠花落谁家
With a half-hearted mew, a white cat called Achilles hovers lazily between two bowls of food adorned with the national flags of Russia and Saudi Arabia. The preferences of this rather ordinary-looking feline, which lives in the basement of the State Hermitage Museum in St Petersburg, Russia, have come under surprising scrutiny in recent weeks.
一只名叫阿喀琉斯（Achilles）的白猫一边小声叫着，一边慵懒地在分别插有俄罗斯和沙特阿拉伯国旗的食盆间踱步。这只看起来并不出众的小猫住在俄罗斯圣彼得堡埃尔米塔日博物馆（State Hermitage Museum）的地下室里，最近几周，他的口味偏好受到了公众的密切关注。
The reason for the attention? He is the official animal oracle for the 2018 Fifa World Cup. The clairvoyant cat signals who will win matches by choosing between two boxes with bowls of food outside them.
为什么受到关注？因为他是2018国际足协世界杯的官方动物"预言家"。盒子外面放着食盆，这位“神算子”选择哪个盒子，就代表哪个队伍会赢得比赛。
According to curators at the Hermitage, Achilles has a good track record for predicting the outcome of football matches after successfully picking the winners of games in last year’s Confederation Cup.
据埃尔米塔日博物馆的馆长说，自打去年成功选中非洲联盟杯的获胜队伍之后，阿喀琉斯预测的足球比赛结果一直很准。
Feline lucky?
幸运猫咪？
Achilles is just the latest in a succession of mystic animals used to predict the outcomes of the World Cup. During the 2010 World Cup in South Africa, an octopus named Paul was shown on national German television predicting the results of matches. The last World Cup in Brazil was blessed by several "psychic" animals, including kangaroos, guinea pigs and camels.
在一连串预测世界杯比赛结果的动物“预言家”中，阿喀琉斯是最新的一位。在2010年南非世界杯中，德国国家电视台播送一只名叫保罗（Paul）的章鱼预测出来的比赛结果。上届巴西世界杯有袋鼠、豚鼠和骆驼"预言家"。
Not everyone, however, is content to leave predictions to prophetic beasts. Bookmakers employ a more methodical approach to set odds using not only historical data on a team’s performance, but also events such as players’ form and injuries to find their favourites. According to bookmakers Paddy Power, the current favourites to win the 2018 World Cup are Brazil at 4/1, followed by Germany at 5/1.
然而，并非每个人都相信动物的预言。博彩公司用更加系统的方法设定赔率，不仅考虑球队的历史战绩，还参考球员分布和伤病情况等因素，最终得出胜率。根据博彩公司帕迪鲍尔（Paddy Power）的预测，这次2018年世界杯的冠军最有可能是巴西队，获胜率是 4/1，其次是德国队，获胜率是5/1。
Some, however, are turning the power of sophisticated statistical models on the beautiful game. Algorithms more commonly used by banks to forecast rising oil prices or changes in the stock market are being tweaked to predict the results of fancy footwork by star players like Cristiano Ronaldo, Lionel Messi and Wayne Rooney.
另外一些人通过建立复杂的统计模型来预测比赛结果。银行常用来预测油价涨幅和股市变化的算法稍作调整可用来预测明星球员的表现，比如罗纳尔多（Cristiano Ronaldo）、梅西（ Lionel Messi）和鲁尼（Wayne Rooney）。
In many ways, according to some analysts, making an accurate call on the results of the World Cup, is far easier than their day job of tracking the ups and downs of the stock markets.
一些分析师称，从许多方面来说，准确预测世界杯的结果比他们追踪记录股价涨跌的日常工作容易多了。
“We know quite a lot about the World Cup because Fifa puts out the game schedule, the rules of the tournament and ranks the teams” ahead of time, says Michael Bolliger, head of emerging markets asset allocation at investment bank UBS, who has turned the financial algorithms he uses day to day onto predicting who might win the World Cup. “So we know a lot more about what we want to model than we do when looking at the investment markets.”
博利格尔是瑞银集团（UBS）新兴市场资产配置部门的主管，他运用日常使用的金融算法预测世界杯的结果。他说：“对于世界杯，我们已知的信息很多，因为国际足联会提前公布赛程、规则和球队排名。建模时已知的信息相比投资市场多得多。”
Bolliger tested hundreds of possible variables that can influence the outcome of a football match. He found there were just four that made a real difference to his ability to predict games when testing his model against the results in the past ten World Cup tournaments.
博利格尔测试了上百个可能影响球赛结果的变量。通过用模型测试过去十届世界杯的比赛结果，他发现只有四种因素能真正影响比赛结果。
The first was an objective ranking of the teams using their Elo ratings, which orders the national teams using a measure of the relative skill of each team. He also incorporated how well the teams performed in the past 12-18 months during the qualification stages of the World Cup.
首先是根据Elo等级分制度为球队列出客观排名，通过比较球队的技术进行排名。他还纳入了过去12-18个月内各个球队在世界杯预选赛的表现。
While past performance is often used to warn investors against betting on a stock’s future performance, past participation in the World Cup and experience of winning played a significant role in predicting future winners in Bolliger’s modelling.
对投资者来说，股票的既往表现对预测其未来走势非常重要。在博利格尔的模型中，球队参加世界杯和获胜的经验对预测结果十分重要。
“These countries – Brazil, Germany, Italy and Uruguay – have played in most tournaments and have won it,” says Bolliger. “They know how to win a World Cup, but the current players also have more confidence because of the previous guys winning, they believe they can win again.”
博利格尔说：“巴西队、德国队、意大利队和乌拉圭队参加过的比赛最多，且有获胜记录，他们知道怎么赢得比赛，现役球员也更有信心。因为以前的队员获胜过，他们相信自己也能做到。”
Bolliger also incorporated another psychological factor into his model – home advantage. Teams playing on their home turf tend to perform far better than might be expected, because having the crowd behind them gives them a boost.
博利格尔的模型中还纳入了心理因素，即主场优势。球队主场作战表现往往会超出预期，因为观众们加油打气会让他们如虎添翼。
But interestingly he also layered in an algorithm that deliberately introduces an element of chance to the model. This is because in sport, much like in the wider world, upsets do happen. The best team doesn’t always win and footballing minnows can become giant killers.
有趣的是，博利格尔也将偶然因素纳入模型当中。体育运动和现实生活一样，出人意料的事时常发生。夺冠热门可能爆冷出局，不起眼的球队可能成为黑马。
“In the world of investments we do something similar – we bring in this element of unpredictability that helps us make better conclusions,” says Bolliger.
博利格尔说：“投资也是如此，我们会考虑不确定因素，这会帮助我们做出更好的判断。”
Bolliger’s algorithms crunched the data about team ranking, past performance, recent results, and home advantage to calculate the possible outcomes of 10,000 tournaments. Adding a random element to account for upsets that might disrupt the expected results produced a range of outcomes, allowing outliers to do better than expected, for example. He then counted how many times each of the 32 participating teams won, made it to the semi-finals or was eliminated at group stages. It showed that Germany and Brazil were most likely to meet in the final.
博利格尔的算法根据球队的排名、既往表现、最近战绩和主场优势模拟了一万场比赛的可能结果。并且加入了随机变量模拟可能会影响预期的不确定因素，令比赛产生不同的结果，比如，允许偏差较大的离群值。他还分别计算了32只参赛队伍获胜、进半决赛或小组赛出局的次数。结果显示总决赛最有可能是德国队对巴西队。
Forecasting the unpredictable
一切皆有可能，球赛难以预测
The idea that uncertainty can make predictions more accurate might at first seem rather strange, but the world is an unpredictable place. And the business world is one in particular that is subject to the vagaries of human nature.
比赛当中的不确定性让预测的结果更加准确，可能乍一听令人费解，但是这个世界瞬息万变，一切皆有可能。商场尤其受到变幻莫测的人性影响。
One certainty about the World Cup is that the tournament will always feature some things that just cannot be predicted.
可以确定的是，世界杯比赛总会有些出人意料的事发生。
Who, for example, could have guessed that Uruguay striker Luis Suárez would bite an opposing Italian player on the shoulder during a group stage match at the 2014 tournament in Brazil? Suárez was given a nine-match ban for the incident, something many pundits say contributed to Uruguay crashing out of the next round of the competition.
比如，谁能想到2014年巴西世界杯小组赛乌拉圭前锋苏亚雷斯（Luis Suárez）会咬对阵的意大利选手？苏亚雷斯为此遭禁赛九场，许多专家称乌拉圭随后的比赛惨遭淘汰与此有关。
Similarly the surprise decision by the Spanish Football Association to sack their national coach Julen Lopetegui just one day before the start of the World Cup in Russia could easily introduce similar levels of unpredictability to this year’s tournament.
西班牙足协在俄罗斯世界杯揭幕战的前一天临阵解雇国家队主教练洛佩特吉（Julen Lopetegui），给本届世界杯增添了同样的不确定性。
Unpredictable events and temperamental player behaviour can swing the result of a match, just as they can in business negotiations or the stock market. Pre-game nerves, match-day pressure and trouble at home can turn even a team of footballing superstars into a shambles.
同商业谈判和股市一样，不可预测的事件和球员的意气用事能扭转比赛结果。赛前感到紧张、比赛中遭受压力，以及团队的内部问题都可能将一队名将变成一盘散沙。
“Players' mindset will play a massive role in successes and failures at the World Cup this summer,” says Jennifer Cumming, a sports psychologist at the University of Birmingham. This can introduce a random element to matches that is difficult to predict.
卡明（Jennifer Cumming）是伯明翰大学的运动心理学家，她说：“运动员的心态对今夏世界杯的成败尤为重要。”这又为比赛增添了难以预测的随机因素。
“On average, about 86% of a result in a match between two teams is random,” says Andreas Heuer, professor of complex systems theory at the University of Münster, who has studied how to predict football scores. “If the players were robots of equal skill, the chances of winning would be like throwing a dice. With the right data, you can predict a match.
霍耶尔（Andreas Heuer）是明斯特大学复杂系统理论的教授，研究如何预测足球比分，他说：“一般来说，86%对比赛结果的预测都是随机的。如果球员是水平相当的机器人，获胜的几率就和掷骰子一样。只要数据正确，就能预测比赛结果。”
“But when the psychology of the players comes in, prediction gets a lot harder.”
“但如果还要考虑球员的心理状态，预测就变得复杂得多。”
Heuer points to a well-studied phenomenon known as loss aversion, which was first identified by Nobel Prize-winning psychologist Daniel Kahneman. It says that humans objectively feel losses far more than they do gains. And it seems to play a major role in matches where the score is 0-0 in the 80th minute, according to Heuer’s analysis.
霍耶尔指出，最早由诺贝尔奖获得者卡尼曼（Daniel Kahneman）提出的损失规避现象认为，人们对失败的感知往往比成功更强烈。这一现象后来被广泛研究。根据霍耶尔的分析，这种现象在比赛进行到80分钟比分仍是0-0时表现得最为明显。
“There are far fewer goals in the final 10 minutes of a football match than one might expect statistically,” says Heuer. “The activity of the teams goes down and they take less risks. They seem to see a one-point draw as being a better option than the chance of getting no points.”
霍耶尔说：“足球比赛最后十分钟的进球比预期少得多。球队活跃性下降，不愿意冒险。他们认为平局得一分比输球得零分要好。”
On the other hand, a draw is not a good result given that teams might get three points for a win. Yet football players and managers seem to prefer to play it safe, especially in the early stages of a tournament like the World Cup where table points matter.
另一方面，相比获胜得三分来说，平局并不算是好结果。然而球员和教练还是愿意打安全牌，尤其是在积分比较重要的世界杯这样的比赛初期。
“This is just irrational behaviour,” adds Heuer. “But people tend to avoid losses.”
霍耶尔说：“这并不是理性的行为，但是人们趋向于规避损失。”
Teams can also become stuck in a particular strategy during matches because of the time and effort they have already invested in this. In economics, this is known as the sunk cost fallacy, where people have a tendency to persist in a failing endeavor because they see what they have already invested as being too good to lose.
球队在比赛中也可能困在某种策略中，因为他们已经投入了太多时间和努力。在经济学中，这叫沉没成本谬误，即人们往往会坚持一些毫无回报的努力，因为他们已经付出了太多。
These are problems that regularly crop up in business decision making, too. Those who understand when it is worth ignoring their human instincts to take a risk are likely to do better, says Heuer.
同样的问题也经常出现在商业决策中。霍耶尔说，那些能够战胜本能勇于冒险的人往往能够做得更好。
Steve Begg, professor of decision-making and risk analysis at the University of Adelaide in Australia, normally spends his time modelling how uncertainty can impact the oil and gas industry. He too has applied this financial markets-modelling work to predicting the outcome of the World Cup.
贝格（Steve Begg）是澳大利亚阿德莱德大学决策风险评估的教授，他主要用模型分析不确定性对石油和天然气产业的影响。他也把这个金融市场模型运用到了预测世界杯比赛结果中。
Using a technique known as a ‘Monte Carlo simulation’, which was initially developed by scientists working on the atomic bomb in the Manhattan Project, to calculate how the form of each team might change during the tournament. He ran 100,000 simulations of possible outcomes in the forthcoming 63 matches of the tournament.
贝格运用负责曼哈顿原子弹计划的科学家发明出的“蒙特卡罗模拟技术”（Monte Carlo simulation）计算比赛中的球队构成，对接下来的63场比赛进行了10万次模拟。
“Uncertainty is crucial in predicting the chance of an oil or gas field being economic,” says Begg. “In the World Cup, it determines the many ways the whole tournament might play out. Probability is subjective, it depends on what you know. It doesn’t need data – you use what information you have to assign a degree of belief in what might happen, and thus make decisions or, in this case, a judgement on who wins. The crucial thing is that your information and reasoning is not biased.”
贝格说：“不确定性对预测石油和天然气涨跌很重要。同样，在世界杯中，它也在很大程度上决定整个比赛的结果。可能性是客观的，取决于你的已知信息。用已知的信息猜测可能会发生什么，然后做决定或是判断谁会赢，不需要运用数据。重要的是，你不带偏见去获取信息和进行推理。”
This is something Bolliger insists is important too – emotion, no matter how much you want your national side to win, has to be taken out of the equation.
博利格尔认为还有一点很重要——情绪。无论你多想要你们的国家队赢，都得把情绪考虑进去。
“Investors can learn a lot from the qualities of successful soccer players,” says Bolliger, who ran his simulations during late night shifts at work. “Agility, balance and calm are all things I will try to use in my interactions with clients.”
博利格尔在上夜班，手头上运算着模拟，他说：“投资者能从优秀的球员身上学到很多。敏捷、平衡和镇定，也是我和客户交往时努力做到的。”
And the winners of the 2018 World Cup, according to some of the most powerful forecasting tools in the business world? According to UBS, Germany will be the winners, while Beggs predicts that Brazil have the best chance of being champions.
根据商界最有力的预测工具，到底谁将是2018世界杯的最后赢家？根据瑞银集团的预测，德国将最终折桂，但是贝格预测巴西最有可能成为冠军。
But for those wanting to take a gamble on the winners of the World Cup, there might be another option alternative to popping to the bookies. One study published in 2007 found that a country’s stock market can respond both positively and negatively to World Cup results, perhaps driven by changes in investor sentiment or mood. The researchers from the Norwegian School of Management found stock prices could slide by up to 49 points the day after a national team lost at the World Cup.
当然对于那些想趁着世界杯赌一把的人，他们可能选择别的赌注方式。2007年发表的一项研究发现，世界杯比赛结果对一个国家的股市既有积极也有消极的影响，可能是由投资者情绪变化引起的。挪威管理学院的研究者发现，在国家队输了世界杯比赛之后的一天，股价最多可能下跌49个点。
Of course, you could just take a day off and enjoy the football.
当然，你也可以不用下注，请一天假好好地享受球赛。
