In Christopher Bierman's book "Football Hacks", he detailed the success model of former Dortmund chief scout Mislin Tate, who had signed Lewandowski and Kagawa for the Hornets. Shinji and Aubameyang. Dortmund had only 10 scouts at the time. This size is a pocket class for a top European team, but their work efficiency is admirable. In 2017, Arsenal paid a price of 1 million euros in order to dig Mislintat. Bierman wrote that the number of potential transfer targets that Mislin Tate’s team submitted to Dortmund each year reached 2,500. They need to use the fastest speed to select the most suitable players for the team. Some top teams have set up their own mathematical analysis models and established their own data analysis offices, but most teams still turn to Analytics FC and SciSports.
在克里斯托弗·比尔曼（Christopher Bierman）的书《足球骇客》中，他详细介绍了多特蒙德前首席球探米斯林·泰特（Mislin Tate）的成功模式，后者已经为勒布朗签下了勒万多夫斯基和香川。 Shinji和Aubameyang。多特蒙德当时只有10名球探。这个规模对于欧洲顶尖的团队来说是小菜一碟，但是他们的工作效率令人钦佩。 2017年，阿森纳为挖矿Mislintat付出了100万欧元的价格。比尔曼写道，米斯林·泰特（Mislin Tate）小组每年向多特蒙德提交的潜在转让目标数量达到2500个。他们需要以最快的速度为球队选择最合适的球员。一些顶级团队已经建立了自己的数学分析模型，并建立了自己的数据分析办公室，但大多数团队仍求助于Analytics FC和SciSports。
In the lobby of a hotel in Madrid, a young man in his 20s is pacing back and forth. He is from Germany, he is funny and humorous, but he has multiple master degrees in economics and management at a young age, but he is not a nerd, but a very eloquent salesperson. His name is Simon Rodel and he is the head of SciSports in Austria, Germany and Switzerland. The Transfer Room is hosting an event at this hotel. This is an online software designed to allow the two teams to directly communicate player transfers. Their purpose is to minimize the intervention of the agents in player transfers. More than 150 club representatives will be present at the event. Rodel needs to contact them one by one in an effort to promote his company to the level of top European teams.
在马德里一家旅馆的大厅里，一个二十多岁的年轻人来回pac步。他来自德国，风趣幽默，但是他年轻时拥有多个经济学和管理学硕士学位，但他不是书呆子，而是一个非常有说服力的销售人员。他叫Simon Rodel，是奥地利，德国和瑞士SciSports的负责人。转移室正在这家酒店举办活动。这是一个在线软件，旨在允许两个团队直接交流球员的转会情况。他们的目的是最小化代理商对球员转会的干预。 150多名俱乐部代表将出席此次活动。 Rodel需要一个一个地与他们联系，以将他的公司提升到欧洲顶级团队的水平。
We sat down, he took out his laptop and showed us his software. Rodel said: "We have business dealings with more than 50 teams, including Lyon, Ajax and the Belgian national team. I myself work for Basel, Frankfurt, Wolfsburg and Paderborn.
"The general manager of Paderborn said that most of their achievements are due to SciSports. They started working with us in the German C. They had almost no transfer budget, but this team is better than their peers. The opponents are smarter, so they can go all the way from C to A. Look at their players. Many of them come from underdeveloped areas of football. Their thin scouting network simply cannot cover those areas."
How did they do it? "We first collect data, such as the first six leagues in England, the first four leagues in Germany, and the first four leagues in Brazil. We will track and update the data in real time for all players in the list. Our calculation model We will keep track of this player and show his influence on the team. We cooperated with WyScout to obtain the video data, then process the data, and finally submit the simplified information to the club. For example, you are a German second team, and We intend to find a striker who uses Lewand as a template. We will show his current ability in the model and predict his future potential. At the same time, we will review the player's performance in the past 6 months.
"In fact, this is like a preliminary screening process. After this process, you can send scouts to conduct field investigations. This is the so-called precision strike, which is efficient and effective."
As an experiment, I asked him to use a computational model to find several players similar to Lewand. The search process is very fast, and we can add more filtering parameters. "Millik, Dolberg, Giroud, and Wilson are the final result," he said, pointing to the screen. Among them, Wilson is the most eye-catching. The Bournemouth forward is often associated with Chelsea and Manchester United. Many people feel incomprehensible about this, but data analysis shows that Wilson is a very talented player.
Speaking of searching for players, some people will ask why the team does not look for players in the game "Football Manager"? Even if you buy genuine games, you don’t have much money. SciSports founder Blowell said that some teams do do this. He said: "Many teams will refer to "Football Manager", but the data in "Football Manager" and "FIFA" come from amateur observers, and they will inevitably have some subjective tendencies. Most people will prefer their own players, but Our data is objective."
说到寻找球员，有人会问为什么球队在“足球经理”游戏中不寻找球员？即使您购买正版游戏，也没有多少钱。 SciSports创始人布洛威尔（Blowell）说，一些团队这样做。他说：“许多球队会引用“足球经理”，但“足球经理”和“ FIFA”中的数据来自业余观察者，他们不可避免地会有一些主观倾向。大多数人会更喜欢自己的球员，但是我们的数据是客观的。”
Rodel added: "And as far as the game itself is concerned, the game data is not real-time data. They set up a working group for each league, and the working group assesses the players' abilities. We are different, we have an artificial intelligence system. , The artificial intelligence system will analyze competition videos from all over the world up to 1,000 times. This means that the data obtained by the artificial intelligence system is both accurate and extremely time-sensitive."
This system is not only used by teams like Paderborn, but also some top teams. For example, Arsenal signed Torreira from Sampdoria using Kanter as a template.
Arsenal bought a set of its own data system from US data company StatDNA in 2012 for £2.1 million. Bierman wrote in "Football Hacker": "Arsenal has done a very good job of confidentiality. Even the full name of the American company is not found in Arsenal's annual financial report, only consisting of acronyms. : AOH-USA LLC."
Arsenal's operations director Almstadt quickly moved from behind the scenes to the front desk. Almstadt studied at the London School of Economics and Economics and worked at Goldman Sachs for three years, and has an MBA from Harvard Business School. He told Arsenal CEO Gazidis: "Looking at a team lineup is like looking at a set of securities portfolio, which contains 30 asset items of varying value."
Almstadt’s task is to improve efficiency in both transfers and wage expenditures. Almstadt said that if StatDNA had been used previously, Arsenal would not waste a lot of transfer funds on Shammach and Park Ju-young. Bierman wrote: "The data show that as early as playing for Bordeaux, Shamacher's expected goal value was relatively low, because he always shot from an inappropriate position. The system also showed that serious technical defects made Sha Mach’s contribution to the campaign is also quite limited." Finally Wenger was persuaded and agreed to introduce this system from StatDNA.
Almstadt的任务是提高转移支付和工资支出的效率。阿尔姆施塔特表示，如果以前使用过StatDNA，阿森纳将不会在Shammach和Park Ju-young上浪费大量转移资金。 Bierman写道：“数据显示，早在为波尔多效力时，Shamacher的预期目标值就相对较低，因为他总是从不合适的位置出手。系统还显示，严重的技术缺陷使Sha Mach对竞选的贡献也相当大有限。”最后，Wenger被说服并同意从StatDNA引入此系统。
It is foreseeable that in the near future, more and more top teams will build their own data analysis models. SciSport’s Rodel said: “Some teams are building their own data systems. Their own system must be more in line with the actual needs of the club. For example, we can define a typical left back, but RB Leipzig knows what they need. What's the left back of the United States? Some big clubs sometimes refuse our service, because there are very few players who can see their eyes. It is enough to use their eyes. Right here, I just met with a Juventus representative After meeting, although he invited me to visit Turin, I can see that they are skeptical about whether it is necessary to cooperate with us. They are Serie A bosses and it is their usual practice to buy famous players."
Current clubs, big or small, are gradually accepting the two-pronged plan of data and intuition. Brouville said: "Data has proven that it can help people save costs. In the past, computer science practitioners may also think that there is no way to communicate with old-school footballers, but nothing can be done without cooperation. Now it is different. Now, we need everyone to work together to build some wonderful projects, not behind closed doors. Intuition and data. Both are indispensable."
Steele is a little more conservative. He said: "Data analysis still has not become a mainstream signing tool. At the seminar, some people always jump up and say that data analysis must improve the efficiency of communication. He will say that data analysis needs to be understood by football players, whether it is Sports director, general manager or CEO. But strangely, no one has ever said it the other way around. No one has ever said,'Football players should learn about data analysis right away.'"
In reality, data analysts need to make adjustments based on the preferences of coaches and players. The task of the video analyst is to translate the research conclusions of the data scientists into football language, and then pass them to the first-line coaches and players.
Stevens explained: “Performance Analysis can adapt to different environments like a chameleon. We are wage earners in the football world. How to do it depends entirely on what the boss wants. Take Crystal Palace as an example. From Dyce to Frank De Boer and then to Hodgson, the coach changed one after another, but we didn't move because we could adapt to them.
史蒂文斯解释说：“性能分析可适应变色龙等不同环境。我们是足球界的打工仔。如何做完全取决于老板的要求。以水晶宫为例。从戴斯（Dyce）到弗兰克·德布尔（Frank De Boer），再到霍奇森（Hodgson），教练接连改变了，但我们没有动弹，因为我们可以适应他们。
"We use video data when we work. Sometimes the newly appointed head coach will ask us to issue an 80-page report, but I will ask,'Why? Do you really read it?' We do do some written reports. , But that is required for due diligence. We usually prepare a 13-page report for the coach before the game. These 13 pages also include the cover, but in fact there are only 12 pages. These 12 pages include: predictions of the opponent’s players, The opposing team’s lineup information, one page about how the opponent will attack when the ball is in control, one page about what the opponent will do when they don’t have the ball, one page for positioning ball analysis, one page for tactical adjustment tendencies (for example, whether to push the ball or stay patient) , There is also a table showing the situation of each opponent’s player, basically that’s it."
During Hodgson's coaching, we focused on the targeted placement of players. Stevens said: "The commendable thing about Hodgson is that if we say that the other party will do this when in possession, then he will say,'Well, yes, since they did it, then we will do it.' We need the last Develop targeted methods, otherwise, what is the point of our analysis for a long time? Similarly, we can easily find the weakness of the other party through data, but only finding out is not enough. We have to find a way to take advantage of the weakness of the other party. For example. How do we defend when the opposing goalkeeper holds the ball? If we play 4-3-3, does the forward trident keep pressing or defending back? They have a delayed midfielder, okay, let No. 9 retreat and oppress Is this late midfielder or No. 8 pressing forward? We will all make our own suggestions for these issues."
Assuming the coach is willing to listen to our opinions, what will the players think? Will they accept the use of data analysis to deploy tactics? At Arsenal, few players are interested in Emery's nuanced video analysis course. At Manchester United, Van Gaal's lengthy and rigorously worded post-match analysis session greatly undermined the players' confidence. Later, Van Gaal changed his strategy and sent each player's mistakes in the game to their mailbox by email. After receiving the emails, some players deleted them without looking at them. Later, Van Gaal installed a tracking plugin to check whether the players opened the emails. The players simply opened the emails with their mobile phones and then proceeded to do other things on their own. Obviously, the use of video analysis to explain tactics is no problem, but it also requires excellent personnel management skills. Some Premier League teams are using an app called Pushfor. This app is very popular in the legal world. Its function includes the ability to send information about whether the customer (here refers to the player) has read each page to the sender (here refers to the coach). ). No matter which team he coaches, Allardys encourages the data analysis department to send analysis materials to the players. These analysis materials may raise the interest of video analysts, but the questions raised by the players are sometimes even difficult to answer. .
假设教练愿意听取我们的意见，球员会怎么想？他们会接受使用数据分析来部署策略吗？在阿森纳，很少有人对Emery细致入微的视频分析课程感兴趣。在曼联，范加尔漫长而严谨的赛后分析会议大大削弱了球员的信心。后来，范加尔（Van Gaal）改变了他的策略，并通过电子邮件将每个游戏者的错误发送到他们的邮箱中。收到电子邮件后，一些玩家将其删除，而没有查看它们。后来，Van Gaal安装了一个跟踪插件，以检查玩家是否打开了电子邮件。玩家只需用手机打开电子邮件，然后继续自己做其他事情。显然，使用视频分析来解释战术是没有问题的，但它也需要出色的人员管理技能。一些英超联赛球队正在使用一个名为Pushfor的应用程序。这个程序在法律界很受欢迎。它的功能包括发送有关客户（此处指的是玩家）是否已阅读每个页面的信息给发送者（此处指的是教练）的功能。 ）。无论他执教哪个团队，Allardys都鼓励数据分析部门将分析材料发送给球员。这些分析材料可能会引起视频分析人员的兴趣，但是播放器提出的问题有时甚至很难回答。 。
Stevens said: "You don't want the players just sitting there and nodding blankly. You want them to say,'I don't agree,''I didn't understand,' or'What if I did that?' Wait, there are nutritional issues. But once they get serious, the video analysts will have a hard time, because you don't want to be too blunt.
"The software we use now is Hudl. Players can see everything on their mobile phones. Everything I mean includes preparations before the game, summary after the game, their own data, the opponent's data, the opponent's goalkeeper's performance, and the opponent's positioning. The ball and penalty situation, etc. We have a record of what the players are watching and how long they have been watching. But the filling of information does not make any sense. It is true that some players are always positive, but for other players, This is information glut. Some players don't like plenary meetings, so group meetings or single meetings may be more suitable for him. In order for players to make progress, we must be flexible.
Data "tracking" is a very popular concept in recent years, which means to record the trajectory of players and monitor their activities without the ball. Several companies, such as Opta, Statsbomb, and Wyscout, are good at collecting data when players have the ball. We call this "event data"—the so-called "event" refers to events related to the ball. There is a clear difference between the two. the difference.
Players' activities on the field can be displayed using data tracking technology. But this does not show the background environment that caused the player to react.
We use Chelsea midfielder Jorginho as an example. We can count his number of passes and the number of key passes, and we can also monitor the passing distance and the number of sprints of his teammates. However, our current model cannot show whether Jorginho missed a better pass line. A player may continuously send a penetrating through pass while the opponent suppresses his side at halftime, and another player may continuously send a through pass when his side continues to control the ball and dominate the game. In the case where the passing distance is not much different, we can't see the obvious difference between the two from the data, but in fact, it is obvious that the former is more difficult. This is now a more difficult problem.
The current challenge is to combine event data with tracking data so that coaches can evaluate existing players or transfer target players based on this latest model.
Barnob, Opta’s former marketing director, said: “This is the general trend. The commonly used filtering options for tracking data include decision-making power, ball-holding options and the opportunity cost of making a decision. The courage to deal with the ball will also be taken into consideration. It has gradually become an indispensable option. When people combine tracking data with event data, football will show a whole new dimension."
This season, the Premier League officially shared similar information to the teams, but this approach has not yet spread throughout Europe, and the England system is not perfect from the current point of view. Clubs that are at the forefront of the times like Liverpool and Arsenal have developed their own models.
The combination of tracking data and event data can also better reflect the value of a defensive player. In recent years, defensive players will be highly praised for making a lot of tackles, but many people in the industry disagree. In Bierman's book "Football Hacks", he quoted an interview with Harvey Alonso.
跟踪数据和事件数据的组合还可以更好地反映防守球员的价值。近年来，防守球员的铲球技术屡获殊荣，但业内许多人对此表示反对。在Bierman的书“ Football Hacks”中，他引用了对Harvey Alonso的采访。
"If I have to shovel, it means I have made a mistake before," Alonso said. "When I was in Liverpool, I often watched some talk shows. The subjects of interviews were young players from the youth team. The questions included age, idols and their own strengths. When asked about their strengths, some of those children would Speaking of shooting, some of them are tackles. I don’t understand why tackles become a specialty in modern football, but they are also regarded as a teaching content and even integrated into the players’ playing style. How can they do this? Look at the game? Tackles are the last resort. Sometimes you need to tackle, but this should definitely not be encouraged."
Stevens of Crystal Palace said: "The so-called defense is not about what you are going to do, but about what you don't want to do. Is it really good to put the shovel? The traditional teaching philosophy is like this,'Don't put the shovel, Stand in position and prepare to intercept.' It means that if my defensive position is okay, it means that the ball will never pass under the opponent's striker, because the opponent's pass will always be intercepted by me. So defense has always been low-key. Work should never be touted. This is one of the main reasons we introduced tracking data."
Now some companies are developing new products in response to this situation. Sportlogiq is said to be close to success, but there are still many companies trying to fish in troubled waters. The Athletic previously revealed that a club paid a six-figure price to a data company, but after investigation, it was found that the company's data system was full of errors and omissions.
Analytics FC ingeniously provides a new method to measure the value of defensive players, although the current system is not perfect. Steele said: "Our computational model can use tracking data to study players, but it can't cover 90 leagues. From the perspective of scouts, we provide the algorithm to the club to continue to broaden our horizons. The more leagues use our system, the more useful information we can collect.
Our algorithm can measure the quality of defense, which is impossible for other 99.9% models. Here we reverse the previous model. We previously calculated the probability of a player scoring in position X. If a defender blocks a shot, the team's risk of conceding a goal is reduced because of this block. Similarly, if a defensive player makes a steal or interception at a key position, this can be reflected in his value. In order to measure the value of a defensive player, we are no longer obsessed with the number of steals, but instead focus on how much his defensive action reduces the risk of conceding a goal. "
While mathematicians and economists are striving to fight for the right to speak in the football world, more and more teams are spending a lot of money to introduce data analysis systems. However, the new coronavirus epidemic has gradually tightened the financial situation of various teams, and now many teams have given long vacations to signing departments.
SciSports’ Chief Information Officer Bruville said: “We signed contracts with three clubs in the past three weeks, but three other teams refused to pay us. Our service is based on football matches. We help them analyze their opponents, but now Without a game, we naturally have no use. Now even the Champions League is postponed, and the winter for data analysis companies is coming. All data companies will develop new models and then put them on the market during the European Cup. We have also done related things. Although the club’s business is not good, we can still make up for it from the national team. Everything is interrelated."
The road to employment for data scientists remains bright. Although the annual salary of some top teams for data workers is less than £30,000, the income of some senior talents is comparable to that of investment banks and top consulting companies. Steele said: "A lot of teams are now seeking cooperation with data scientists. This is the same as the teams seeking sports science consulting 15 years ago. In the past, many sports science departments graduated from Loughborough University and the University of Bath. Students all have a meager income. They can only go to the gym to do miscellaneous work or work as stadium staff. The coaches are indifferent to these people. They can't get into the football business. People will say, "He didn't make trouble anyway. Just let him do it.'
"Now there are about 10 staff in the sports science office of each team, and there are even personnel who install GPS on players and analyze the collected data. This is really different from the past. I believe that data science practitioners will also have a bright future. Most teams are equipped with a video analyst and a data collector, but many people do not have a data science education background. They are neither mathematicians nor Not a data engineer. The annual salary of high-end talents is quite high, because they could have worked in banks. Look at Will Spearman in Liverpool, a nuclear physicist at Harvard University. He will be paid how much he wants. How many people have his qualifications?"
Will the club go to top universities to recruit data science graduates and give them internship opportunities? Steele said: "I think it is entirely possible. The other departments of a team only need to compete with the parallel departments of their peers, but the data science department is different. The team needs to compete with the government, high-end digital modeling companies and various other departments. Big consulting companies are vying for talent. But one thing can never be wrong: For professionals of the same level, clubs can always attract them at a lower price for only one reason-passion for football."
Finkelstein believes that his prediction 17 years ago has become a reality. At the end of the interview, he concluded: "I recently contacted Ian who works in Liverpool. He believes that the "Fink Think Tank" column is the basis for big data to take root in the football world. I am very grateful, thank you!"