Football techniques and tactics statistics: case study of wyscout analytical system

ˑ: 

Dr. Hab., Associate Professor A.A. Polozov1
PhD, Associate Professor A.V. Popovich1
Postgraduate student M.V. Kraev1
1Ural Federal University, Yekaterinburg

Corresponding author: aa.polozov@urfu.ru

Keywords: technical and tactical actions, football, result.

Abstract

Objective of the study was to conduct a correlation analysis to substantiate the independence of evaluation of the game outcome from the increase in the number of registered tactical-technical actions of football players.

Methods and structure of the study. A total of 140 games of the leading championships were analyzed: England, Germany, Spain, Russia, France, and Italy. For each game, the Wyscout report on tactical-technical actions was studied and the total number of tactical-technical actions was calculated. The advantage in tactical-technical actions was assessed by the difference in the total number of tactical-technical actions of both teams. The assessment through the Wyscout Platform involves tactical-technical actions as such and various aggregations that are formed from the correlating characteristics.

Results and conclusions. The analysis of 140 games of the English Premier League, Primera, and Russian Premier League of the 2019/2020 season revealed a negative correlation between the total number of tactical-technical actions and game outcome (r=-0.06). Today, there are at least 10 platforms that count tactical-technical actions in top-level team games. The commercialization of this line of work led to the desire to inflate the amount of tactical-technical actions calculated at the cost of losing the connection between the tactical-technical actions advantage and the game outcome. The authors have previously shown that with the minimum number of factors, the regression equation has the best agreement with the results of the football matches if non-correlating characteristics are taken into account. The negative correlation of the number of tactical-technical actions with the game outcome exhausts the theme of tactical-technical actions, creates a prerequisite for the transition to the game scoring through technical-tactical martial arts, assessed by their cost - impact on the match outcome.

Background. Since 1980 the football analysts have extended the range of the technical and tactical actions subject to analysis [2-5] – from 8 items in the Y.A. Morozov (1980) method [6] to as many as 100 items in the presently popular Instat and Wyscout analytical systems – despite the fact that the higher are the technical and tactical actions numbers, the poorer they are normally correlated with the actual competitive performances.

Objective of the study was to demonstrate, by a correlation analysis, little if any correlation between the formally fixed technical and tactical actions numbers and the actual match results.

Methods and structure of the study. We sampled for analysis 140 matches in championships of England, Germany, Spain, Russia, France and Italy. We analyzed the match technical and tactical actions reports by Wyscout to find the technical and tactical actions totals of the both teams. Then we ranked the matches in a descending order for the match hosts’ advantages in the technical and tactical actions numbers and provisionally split them up into 7 intervals of 20 matches each. The total goals scored in 20 matches of every interval were divided by the goals conceded to find the scoring averages. The scored points were subject to the same analysis: see Tables 1 and 2.

Table 1. Comparative analysis of 140 European championship matches reported by Wyscout

Match

Nation

Score

Difference

Technical and tactical actions

Technical and tactical actions

Technical and tactical actions difference

Points

Bournemouth

Man United

Engl

1

0

1

844

979

-135

3

Arsenal

Woolverhampton

Engl

1

1

0

1045

881

164

1

Aston Villa

Liverpool

Engl

1

2

-1

725

1198

-473

0

Brighton

Norwich City

Engl

2

0

2

1037

858

179

3

Man City

Southampton

Engl

2

1

1

1250

707

543

3

Sheffield

Barnley

Engl

3

0

3

852

958

-106

3

West Ham

Newcastle

Engl

2

3

-1

1058

730

328

0

Watford

Chelsea

Engl

1

2

-1

818

1204

-386

0

Crystal Palace

Lester

Engl

0

2

-2

762

860

-98

0

Everton

Tottenham

Engl

1

1

0

865

910

-45

1

Everton

West Ham

Engl

2

0

2

895

857

38

3

Bournemouth

Norwich City

Engl

0

0

0

954

987

-33

1

Aston Villa

Brighton

Engl

2

1

1

929

902

27

3

Chelsea

Newcastle

Engl

1

0

1

1128

739

389

3

Lester

Barnley

Engl

2

1

1

967

745

222

3

Tottenham

Watford

Engl

1

1

0

1238

823

415

1

Wolverhampton

Southhampton

Engl

1

1

0

950

865

85

1

Crystal Palace

Man City

Engl

0

2

-2

723

1158

-435

0

Man United

Liverpool

Engl

1

1

0

783

1107

-324

1

Sheffield

Arsenal

Engl

1

0

1

750

1105

-355

3

 
Results and discussion. The values were transformed for analysis to put them on a universal basis within the 0-1 range using equation Y = (Xi - Xmin) / (Xmax - Xmin).

Table 2. Key Wyscout indicators ranked by their correlations with the match scores and weight in the regression equation

Indicator

Regression ratio

r

45s-plus ball control

0,900

0,61

Long passes

0,730

0,61

Average passes per match

0,655

0,52

High passes

0,533

0,5

Average shots per match

0,523

0,75

5-15s ball control

0,367

0,28

Ball control in % per match

0,079

0,57

Repossessions and interceptions

0,039

0,59

Ball losses in the last third

0,038

0,65

Average losses per match

0,015

0,34

Attacking/ forward passes

-0,016

0,55

15-45s ball control

-0,099

0,59

Ball control time per match

-0,278

0,46

Passes to the last third

-0,445

0,65

Contacts in the penalty area

-0,569

0,6

Average ball control time

-1,484

0,38

We compared the above data with findings of one of the 1995 studies [1, 6] that used equation ∆ = (N (+) - N (-)) / (N (+) + N (-)). Note that for a regression analysis it is recommended to select the parameters with the highest internal correlations. Having processed the data, we arrived at r = 0.81 correlation of the expected points with the actually scored points: Δ = 0.75X1 + 0.71X2 + 0.62X12.

The team advantages in the shots on goal, field control (penetration depth) and repossessions provide a sound basis for the match result forecasts. When the key factors for analysis are minimized, this equation yields the most reliable outcome. Actually the Wyscout analysis offers only two of three most important indicators and, hence, not effective enough in analyzing the defense quality and shots on goal: see Figures 1-4.

Figure 1. Growth of the analyzed technical and tactical actions numbers since 1980

Figure 2. Correlation of the technical and tactical actions numbers with the match results since 1980

The correlation ratio for the scoring advantage and points per match was estimated at r = 0.871. The correlation ratio for the scoring advantage and technical and tactical actions totals for the 2018-2019 championships of England, Spain and Russia (140 matches) was estimated at r = -0.06. And the correlation ratio for the match points and technical and tactical actions totals for the 2018-2019 championships of England, Spain and Russia (140 matches) was estimated at r = -0.052. An attempt to add new technical and tactical actions indicators yielded in a poorer correlation with the match results in fact.

 

Figure 3. Correlation ratios for the scoring advantages and technical and tactical actions totals for the 2019-20 England and Russia championship matches (n=140)

Figure 4. Correlation ratios for the points scored and technical and tactical actions totals for the 2019-20 England and Russia championship matches (n=140)

It should be emphasized that the technical and tactical actions totals were found virtually non-correlated with the goals scored to goals contended ratios, with the only peak found close to the zero technical and tactical actions total. Paradoxically enough, the hosts’ and guests’ equality in this indicator was found associated with the hosts’ wins; whilst the high advantages in the technical and tactical actions totals were found virtually non-beneficial for the teams: see Table 3.

Table 3. Correlation between the match points and the hosts’ technical and tactical actions advantage for the 2019-20 England and Russia championship matches (n=140)     

Hosts’ technical and tactical actions advantage

Match points

Difference in scores

380,85

0,67

0,92

196

0,88

1,23

110,1

0,76

0,93

34,45

1,22

1,21

-47,25

2,33

2,27

-137,35

0,88

1,08

-330,1

0,71

0,83

 
Conclusion. The present Wyscout and Instat systems analyze individual technical and tactical actions and aggregated technical and tactical actions; with the individual technical and tactical actions including passes, shots on goals, etc.; and the aggregated values including XG, 45s-plus ball control etc. It should be noted, however, that the correlated indicators are normally aggregated. It was back in 1995 that one of us found the regression equation yielding a product fairly correlated with the match result when the technical and tactical actions numbers are minimized; particularly when a special priority is given to the non-correlated indicators like the team advantages in the shots on goals, field control (penetration depth) and repossession percentages. The study found that extensions of the technical and tactical actions numbers never improve the model efficiency and dependability. Therefore, we recommend the technical and tactical actions counts being replaced by the technical and tactical actions cost estimates i.e. their contributions to the match results/ competitive performance.

References

  1. Golomazov S.V., Chirva B.G. Theory and methodology of football. V. 1. Game Technique. Moscow: SportAkademPress publ., 2002. 472 p.
  2. Lyukshinov N.M. The art of training elite footballers. Teaching aid. Moscow: Sovetskiy sport, TVT Divizion publ., 2006, 432 p.
  3. Koloskov V.I. et al. Results of training and performance of Soviet football players at the XXIV Olympic Games. Nauchno-sportivny vestnik. 1989. No. 1-2. pp. 54-61.
  4. Polozov A.A. Rating system in team sports and single combats. Yekaterinburg: USTU-UPI publ., 1995. 110 p.
  5. Polozov A.A. Upcoming match. Computer version. Teoriya i praktika fiz. kultury. 1997. No. 3. pp. 41-44.
  6. Tyulenkov S.Y., Guba V.P., Prokhorov A.V. Theoretical and methodological aspects of football training management. Study guide. Smolensk: TOO IKA publ., 1997. 116 p.