サッカーの上達にデータを生かす
Summary
TLDRこのスクリプトでは、データがサッカーの向上にどのように役立つかについて語られています。経験や直感に限界があり、客観的なデータの活用が効果的かつ効率的な改善につながると示されています。ゲーム分析、GPSデータ、視覚スキル、目運動解析など、さまざまな研究を通じて、技術の進歩がスポーツにおけるデータ収集を容易にし、それらをトレーニングや戦略の改善に活用する重要性が強調されています。
Takeaways
- 😀 スポーツ界では、経験と直感は重要な要素ですが、経験と直感だけでは限界があり、オブジェクトデータの活用はより効果的かつ効率的な改善につながる。
- 🏆 サッカーの試合分析は、試合でのプレイを改善するために最も重要な活動であり、試合の質的な分析と、GPSや画像処理技術を用いた定量的な分析を行う。
- 📊 GPSトラッカーやビデオカメラを用いて選手やボールの動きデータを自動で取得し、試合の分析に利用することができる。
- 🔍 日本とドイツの攻撃パターンを比較した研究で、攻撃の成功率に差は見られず、攻撃パターンはほぼ同じであることが明らかになった。
- 🎯 ゴールキーパーの守備能力を評価する研究では、ロジスティック回帰分析を用いて、各シュートの難易度と失敗確率を算出した。
- 🏃♂️ GPSデータを用いて選手の持久力や体力を測定し、トレーニング計画を作成することができる。
- 🤔 サッカー選手の視覚スキルを評価する研究では、脳波、筋電位(EMG)、反応時間データを用いて、状況を評価する速さの違いを発見した。
- 👀 目の動きを測定する研究では、上級者と中級者の選手が異なる視点を重視していることが明らかになり、視覚スキルのトレーニングが必要であることが示された。
- 📈 技術の発展により、より多様な情報を収集することが容易になり、スポーツ活動でのデータの活用が重要になる。
- 💭 経験と直感を大切にしつつ、定量的なデータを効果的に適用する理解が必要であり、スポーツの向上にはデータの収集と分析が不可欠である。
Q & A
中谷先生はどのような研究センターに所属していますか?
-中谷先生は筑波大学サッカーコーチング研究センターに所属しています。
スポーツにおける経験と直感にはどのような限界がありますか?
-経験と直感はスポーツの進歩に重要な要因ですが、それだけでは限界があり、客観的なデータをうまく活用することでより効果的かつ効率的な改善が可能になります。
サッカーの試合分析において何を分析する予定ですか?
-試合分析では、どのようなプレイがうまくいくか、どのようなプレイが問題を引き起こすか、特定の状況がなぜ起こるのか、それらを改善するために何ができるかを分析します。
GPSや画像処理技術を利用して得られるデータをどのような分析に使いますか?
-GPSや画像処理技術を利用することで、選手やボールの動きに関する位置関連データや動きに関する情報を自動的に取得し、ゲーム分析に活用します。
日本とドイツの攻撃パターンの違いはどのように調査されましたか?
-攻撃シーンを3つのパターンに分類し、それぞれの攻撃でペナルティエリアへの侵入やシュートの有無、シュートの成功率などを記録して分析しました。
ゴールキーパーの守備能力を評価する指標は何ですか?
-ゴールキーパーの守備能力を評価する指標には、平均ゴール数やシュート成功率などが含まれますが、各シュートの難易度を考慮する必要があります。
ロジスティック回帰分析とはどのような分析手法ですか?
-ロジスティック回帰分析は、データから失敗する可能性や難易度を予測するための統計技術です。各シュートの成功や失敗、それに関連する要因を記録し、これらのデータを分析して回帰方程式を作成します。
GPSデータを使用して選手の持久力や体力を測定する方法はどのようなものですか?
-選手がGPSトラッカーを装着することで、心拍数や総移動距離、高強度の運動時間などを自動的に取得し、トレーニング計画を作成することができます。
視覚スキルの違いはどのように調査されましたか?
-視覚スキルの違いを調査するために、脳波データ、筋電位(EMG)データ、反応時間データを取得し、サッカーのパスシナリオに基づく選択反応タスクを行いました。
目線のデータはどのようにして分析されますか?
-目線のデータは、目線追跡システムを装着して取得し、視線が向いている場所や視線の動きを分析することで、選手が実際に何を見ているかを客観的に把握します。
技術の進歩により収集できる情報が増えたことの利点は何ですか?
-技術の進歩により、スポーツ活動や生理的反応に関する多様な情報を収集することが容易になり、それらの情報を運動場面に活用するためのアイデアや意見が重要になります。
Outlines
😀 データ分析によるサッカー能力向上の重要性
中谷先生は、筑波大学サッカーコーチング研究センターから来て、データがサッカー能力向上にどのように役立つかを語ります。経験や直感は重要ですが、それだけでは限界があり、客観的なデータの活用が効果的かつ効率的な改善につながることが示されています。ゲーム分析の技術や、GPSや画像処理技術を用いたデータ収集方法、さらには日本とドイツの攻撃パターンの比較研究など、5つのサッカー関連研究を紹介しています。
📊 ゲーム分析による攻撃パターンの比較研究
ゲーム分析の詳細を説明し、日本とドイツの攻撃パターンを比較した研究結果を紹介しています。攻撃シーンを3つのパターンに分類し、それぞれの攻撃がペナルティエリアに入るか、シュートを行ったか、そしてその成功か否かを記録しました。結果から、日本とドイツは同じ攻撃パターンを狙い、特にディフェンダーとミッドフィルダーの間のギャップを利用した攻撃が最も得点に関係していることがわかりました。しかし、ドイツの方がより効率的で、より多くのチャンスを創出していることが明らかになりました。
🅰️ ゴールキーパーのディフェンス能力を評価する研究
ゴールキーパーのディフェンス能力を評価する研究について語られており、通常の目標数やセーブ率だけでなく、シュートの難易度も考慮する重要性があります。ロジスティック回帰分析を使って、各シュートの難易度と失敗する確率を明らかにしています。この分析により、ゴールキーパーがどれだけ効果的にシュートを阻止できるかを定量的に評価できるようになりました。
🏃♂️ GPSデータを用いたサッカー選手の身体能力分析
GPSデータを用いた選手の身体能力や持久力の測定方法について説明しています。GPSトラッカーを装着させてデータを自動収集し、トレーニング計画を作成しています。心拍数や総移動距離、高強度での時間や距離など、101種類のデータを取得し、それらを分析して選手のトレーニング負荷を監視しています。
👀 視覚スキルと目の動かし方による選手のスキルレベル比較
視覚スキルや目の動かし方による選手のスキルレベル比較について研究しています。脳波、筋電図、反応時間データを取得し、状況を判断する速度の違いを見つけました。さらに、目を追う技術を使って、上級者と中級者の選手がどのように状況を見取り、判断するかを比較しています。上級者は広い範囲に注意を払い、重要なプレーヤーに焦点を当てているのに対し、中級者は近くのプレーヤーやマークされたチームメイト、開放空間に注目していることがわかりました。
📈 技術開発とスポーツにおけるデータ活用の重要性
技術の進歩により、スポーツ活動や生理学的反応の多様な情報を収集することが容易になりました。しかし、どのようなデータを収集し、どのように分析してスポーツで活用するかについては、スポーツ現場にいる人々の考え方と意見が重要であると結び付けています。経験と直感を大切にしつつ、定量データを効果的に適用する理解が不可欠であり、客観的な自己反映を通じて改善のヒントを見つけることができると語っています。
Mindmap
Keywords
💡データ分析
💡GPSトラッカー
💡視覚スキル
💡ゲーム分析
💡ロジスティック回帰分析
💡経験と直感
💡心拍数
💡目運動分析
💡トレーニング計画
💡競技レベル
Highlights
数据分析在足球运动中的应用,强调了超越经验和直觉的客观数据的重要性。
介绍了定量和定性分析技术,以及如何通过观察和收集数据来分析比赛。
使用GPS追踪器和视频摄像机自动获取球员和球的移动数据。
计算机生成的多角度回放技术在体育广播中的应用及其成本问题。
实验室中手动输入数据进行比赛分析的方法及其准确性问题。
日本与德国足球进攻方式的比较研究,提出了关于进攻效率的假设。
通过比赛分析,发现日本和德国在进攻模式上的相似之处和差异。
使用逻辑回归分析评估守门员的防守能力,考虑射门难度。
通过统计技术揭示守门员面对不同射门情况下的扑救失败概率。
使用GPS数据监测足球运动员的体能和耐力,制定训练计划。
通过GPS追踪器收集的101种数据类型,以及如何根据需要选择数据。
通过眼动追踪系统分析足球运动员的视觉技能和决策过程。
研究发现高水平足球运动员在视觉信息处理和反应时间上的优势。
通过脑波、肌电图和反应时间数据研究足球运动员的认知和肌肉活动。
通过眼动数据分析,揭示不同竞技水平足球运动员的视觉焦点差异。
技术发展使得收集多样化信息变得更加容易,但需要结合实际运动场景的理解。
强调了在运动训练中有效应用定量数据的重要性,同时重视经验和直觉。
建议通过收集和分析数据来客观反思表现,发现提升的线索。
Transcripts
Hello, I am Nakayama, from the University of Tsukuba Soccer Coaching Studies Research Center.
Today, I would like to talk about how data is used to improve soccer.
In the world of sports, not only soccer, the experience and intuition of players
and coaches is an important factor in the advancement of the sport.
However, there are limits to what can be done with experience and intuition
and making good use of objective data can lead to more effective and efficient improvements.
Today I will explain data collection and analysis techniques
while introducing the five pieces of soccer-related research that you can see here.
To start with, let’s look at game analysis.
For soccer players, playing well in matches is the most important thing.
Therefore, analyzing games is the most important activity we can do which influences improvement in soccer.
So, what do we actually analyze from matches?
We carry out analyses to determine which plays go well, which plays lead to problems,
why certain situations come about, and what can be done to improve them.
This is achieved by means of qualitative analysis through observations by people such as coaches,
and quantitative analysis of gathered data. Here, we will look at quantitative analysis.
Quantitative data used in game analysis can be obtained through various means.
GPS and image processing technology have now developed to the point
that the movement data of players and the ball can be obtained automatically.
By fitting players with GPS trackers, and installing video cameras in stadiums,
it has become possible to acquire from each device location related data,
as well as data and information relating to the players’ movements, all within a short period of time.
I’m sure you have all seen the computer generated footage
used in live sports broadcasts which allows replays from all angles.
That said, this method is still special and expensive,
so sports teams and our research center can’t easily use it in our game analyses.
Most of the time,
we record game analysis data by ourselves,
at the sports club or in the laboratory. In our lab,
we decide on the items we want to analyze or know about,
and generally manually input relevant data into the computer while observing game footage.
This takes a lot of time.
Another issue is that there are differences in accuracy depending on the person entering the data.
Here, I’d like to introduce the soccer game analysis research we have carried out at our lab.
We investigated whether there was a difference in the offensive aspect of Japanese soccer
and that of a foreign powerhouse country; in this case, Germany.
It is often said that while Germany approaches the ball proactively,
breaking through the opposing defensive formation,
Japan passes the ball around without approaching the goal and is
unable to break through their opponent’s defense.
However, this assessment is purely subjective,
and had not been investigated using objective data.
Therefore, we raised the hypothesis that Japan has more problems in offense compared to Germany,
and carried out game analysis to test it.
First, we had to decide on which factors to analyze.
Here, we classified offense scenes during game play into three patterns;
attacks using the gap between the defenders and midfielders,
attacks which used space on the sides without making use of gaps between defenders and midfielders,
and attacks which made use of neither the sides nor gaps between defenders and midfielders.
In addition, for each attack, we recorded whether the attackers entered the penalty area or not,
and whether they made it as far as shooting.
If they did, we recorded where they took the shot from,
and whether or not it was successful.
Unfortunately, we could not acquire this data automatically,
so we had to analyze the situation and
enter the data while repeatedly playing back the soccer game footage.
This photo shows the situation at that time.
This is the computer screen that they’re looking at.
While playing back the footage,
they input data by pushing buttons preset for certain data items.
This is the datasheet used in this research.
Over 2,000 data items were collected.
These are the results of the comparison between Japan and Germany.
There was no difference between the number of attacks, goals,
rate of entries into the penalty area, shot rate, or shot success rate,
but we did find that the attack success rate of Japan was low compared to Germany.
For both Japan and Germany,
the attack pattern which used a gap between defenders and
midfielders was most closely linked to scoring.
However, if we look at the results more closely,
we can see that attacks using gaps between defenders and
midfielders were more closely linked to shots and goals for Germany than for Japan.
Based on subjective appraisals,
we hypothesized that there was a difference between the factors of Japanese and German offense.
Through quantitative analysis,
we discovered that both Japan and Germany aim for the same type of attacks,
and that attacks which make use of gaps between defenders and
midfielders are equally important to both,
but that Germany was more efficient and was able to create more opportunities.
From these results, we concluded that in Japanese coaching,
thorough instruction on things like the timing of passes is necessary.
Next, I will show you some game analysis research
which focuses on the goalkeeper’s play,
evaluating the goalkeeper’s defensive abilities.
Indicators of a goalkeeper’s defensive ability include things
like the average number of goals conceded and their successful save percentages.
However, this doesn’t take into account the level of difficulty in stopping each shot.
In other words, stopping an easy shot and stopping a difficult one are given the same merit.
So, we recorded the success or failure of every shot and
major associated factors throughout a game,
and used a statistical technique called logistic regression analysis on this data
to reveal the level of difficulty and the probability of a failed save for each shot.
We collected and analyzed shots
from game footage and recorded whether they succeeded or failed,
as well as 12 factors relating to shooting,
such as the length of time taken to achieve each shot and
whether or not there was a defender on the shooter.
The method of analysis was the same as the game analysis explained previously.
While repeatedly playing back the footage, data for each factor was recorded.
The data for each factor then underwent statistical processing,
and a regression equation was created to estimate the probability of a save being unsuccessful.
With this equation, the probability of a shot save failure could be calculated.
For example, a save with a 0% failure probability is a long shot with a slow shot speed,
coming straight toward the goalkeeper.
A save with a 25% failure probability is one in which the shot is taken
from near the penalty area and under steady pressure from a defender,
so that the shot’s course is restricted and easy to predict,
and the shot comes towards the goalkeeper from slightly off-center.
A save with a 50% failure probability is one in which the shot is taken from near the penalty area,
without pressure from a defender,
in circumstances where it is difficult to predict the shot course,
and the shot comes towards the goalkeeper from slightly off-center.
A penalty kick has a 72% save failure probability.
Finally, a cross that comes in from the side,
leading to an unmarked free shot from near the goal has a save failure probability of 98%,
and is an example of a very difficult shot for a goalkeeper to save.
Let’s try to evaluate the defensive ability of four goalkeepers.
Players A, B, C, and D have each conceded 10 or 20 goals.
Using the regression equation obtained from this research,
we worked out the failure probability of each of these shots,
and from this calculated the predicted number of conceded goals.
Looking at this, we can see that while Player A had a predicted conceded goal number of 10,
they in fact conceded 20 goals.
There is a chance that they allowed easy shots to score.
Player C, like Player A, has conceded 20 goals,
but was faced with a lot of difficult shots, and was predicted 30 conceded goals.
This means that while faced with a predicted 30 conceded goals, he could limit this to 20,
and Player C can be considered as having the highest goal saving ability.
Next, I will present an example of how GPS data is used to measure the physical strength
and endurance of soccer players.
In the world of sports,
it is now becoming the norm to measure the endurance and
physical strength of athletes using GPS data, and create a training plan.
It is incredibly convenient as one can obtain a lot of data
automatically by fitting an athlete with a GPS tracker.
We also use GPS data in the University of Tsukuba football and soccer clubs, so I will introduce an example.
The soccer club uses GPS data from matches and
everyday training to monitor the physical load of its players.
Physical load can be measured by the volume and intensity of an activity.
By regulating the volume and intensity of activities,
we aim to improve performance while limiting the risk of injury to players.
Players wear a tracker on their backs while taking part in games.
The signals are then picked up by receivers on the edge of the sports grounds.
101 types of data can be acquired through GPS.
From among these,
we take volume-related data measurements that include heart rate and total distance traveled.
Similarly, as data related to intensity,
we measure the time spent exercising at over 85% of the maximum heart rate,
and the proportion of the total running distance done at high intensity.
We are able to select the data we require based on our needs.
This data shows a certain player’s movement speed and heart rate during a match.
The blue line is the speed of movement, and the red line is the heart rate.
This is a diagram showing where a single player moved during a match.
As the frequency increases, the color changes from blue to red.
Such data can be obtained from every player who wears a GPS tracker.
This shows the movements of a whole team—in other words, 11 people—during a match.
The movements of all 11 people can be recreated simultaneously.
Using this graphic, it is possible to analyze the tactical maneuvers of the team.
This is a comparison of the data for each team member.
The top shows the maximum speed, and the bottom plots the number of sprints.
We can see that there are individual differences.
These are the records for factors such as the total distance traveled
and distance traveled at high intensity during daily training.
With this, we can monitor the load of that particular day’s training.
We manage the conditioning of our players by using data like those mentioned previously
to draw up a week’s training plan based on the volume and intensity of activity during a match,
and constantly monitor whether day-to-day training is hitting those benchmarks.
Next, I will introduce a study regarding whether a soccer player’s visual skills,
such as their ability to observe and
make judgements based on the game situation, differ depending on their level.
We investigated the characteristics of soccer players who are said to be skilled,
beginning by studying their brain wave, electromyogram (EMG), and reaction time data.
Are skilled soccer players able to assess situations more quickly?
We developed the hypothesis that
intracerebral information processing differs depending on the competitive level of the player,
and began our experiment.
As a means of evaluating the intracerebral information processing of athletes,
we obtained and analyzed event-related potential linked to thought and cognition through brain wave data,
EMG data to understand muscle movement,
and the reaction time taken to make decisions after an image is shown.
This is a choice reaction task based on soccer pass scenarios.
Please look at the black screen.
The blue players are teammates, while the white are opposing defenders.
There are several patterns of image presented, shown at random.
The participants’ task is to look at the images presented and
decide based on the situation whether to pass to the right, the left,
or the space between the two players.
Once they’ve looked at the situation
and decided whether it’s better to pass to the left or right teammate,
they press a switch at their feet.
If they decide to pass in the gap between the two opponents, they do nothing.
This experiment is repeated 120 times.
This is the brain wave, EMG, and reaction time data.
The horizontal axis is time.
The top graph shows brain waves, and several V-shaped waveforms are visible.
Based on the waveforms, we guess when the brain has started to process visual information or has evaluated a stimulus.
We can confirm the moment a muscle has started to move using the EMG.
The bottom graph captures the moment the foot switch was pressed.
We can see that the timings of the brain reacting,
the muscle beginning to move, and the switch being pressed are slightly different.
This is not directly related to today’s talk,
but there are various interesting things to be discovered by investigating how the brain controls people’s movements,
so I recommend you look into that as well.
The subjects of this experiment were university soccer players of different competitive levels.
We found from the brain waves that
players with higher competitive skills had quicker intracerebral visual information processing
and stimulus evaluation than lower-level players.
Similarly, from the EMG and reaction time data,
we found that players with higher competitive skills had faster physical reaction outputs.
Although there are differences in the competitiveness of soccer players in terms of skill,
it became clear that there are differences in the speed at which they judge the situation,
suggesting the need for cognitive training during practice.
Finally, let’s look at soccer players’ ability to judge a situation by measuring eye movement.
We analyzed optical data from advanced and intermediate soccer players.
We fitted them with eye tracking systems like the one in the picture.
Analyzing eye movement is one way of revealing what a person is looking at.
We can get an objective grasp of what a person is focusing on based on the duration and position of their gaze.
This is a view of the experiment.
A screen is set up on the wall of the laboratory which shows soccer game scenario footage.
We had the participants carry out the task of receiving a ball passed to them
by a ball shooting machine on their left while observing the screen,
and passing to wherever they judged appropriate based on the circumstances.
In order to predict where the gaze might be directed, we categorized target locations in advance.
Here is an example from a different experiment using kendo.
The categories include the head, the torso, the bamboo sword, the wrists, and the lower body.
In this soccer experiment, we categorized spaces and target players in the same way,
then analyzed where the gaze was directed.
Here we have matched the eye movement data with the video frame-by-frame.
This is kendo experiment data, where the top shows an experienced practitioner known as a “master,”
while the bottom is an inexperienced practitioner.
I’m sure you can tell that there is a large difference in the places looked at and the way the eye moves.
The more experienced a person is, the less they look at anything other than the face of the opponent.
Here, we have compared advanced and intermediate soccer players in the same way.
The difference isn’t as clear as with kendo,
but it seems that there is a difference in the places focused on
and the way players move their eyes depending on their competitive level.
Advanced players paid attention to a wider area, being aware of key players such as unmarked teammates,
teammates they are considering passing to next, and opponents.
Conversely, intermediate players paid a lot of attention to nearby players, marked teammates, and open spaces.
Using eye tracking data, we were able to learn how soccer players actually look at things like the ball
and the teammate they are passing to in order to pass correctly.
From this, we are able to recognize skillful techniques and tips for playing well, then leverage this in concrete coaching advice.
In conclusion, with the development of technology, it has become easier to collect more diverse information,
such as the movements made during sporting activities and physiological responses.
However, in order to know what data to collect, and how to analyze it so that it can be leveraged in an athletic setting,
the ideas and opinions of people who are constantly in those athletic settings are necessary.
It is important to have an understanding of how to effectively apply quantitative data,
while also valuing the experience and intuition accumulated through regular sporting activity.
If you want to get better at sports, the first step is to be physically active,
but by occasionally collecting data by using footage and recording measurements,
then using this to objectively reflect on your performance,
you will surely discover some hints to help you improve.
Thank you for listening.
5.0 / 5 (0 votes)