12
1 Analysis of Group-Tactical Offensive Behavior in Elite Basketball on the Basis of a Process Orientated Model Hubert Remmert Up to now systematic game observations have insufficiently been used to de- scribe basketball’s tactical structures in detail. Thus, it seems to be urgently necessary to evaluate literatures and coaches’ recommendations by objective game data. Basketball literature has been analyzed to build a process-orientated state-event model that represents players’ offensive-defensive interactions, es- pecially within group-tactical plays in set offense against man-to-man defense. Based on this model, a specific observation system has been introduced to describe game reality by using the method of systematic game observation. Furthermore, the so-called inter-rater reliability of two separate observations has been calculated to guarantee a sufficient quality of the observation system (Cohen’s kappa [κ] for each observational category: coefficients between 0.685 to 1.000). Sixty games of international elite-level basketball were analyzed by using the interactive video computer system VIDEO AS. Results show a sur- prisingly wider variety of offensive-defensive interactions (as sequences of opening action, defensive constellation, and following action within group- tactical plays like screen actions) than described in literature. Suggestions for basketball training can be made considering also rates of success of the different interaction sequences. We conclude that a wider spectrum of group-tactical action patterns, especially within screen actions, has to be developed when practicing with junior basketball players. Key Words: training theory, systematic game observation, elite basketball, offensive group-tactical plays, process-orientated model building Key Points: 1. A process-orientated model was designed to establish a specific method of system- atic game observation to verify and evaluate practical basketball literature’s rec- ommendations with data from elite basketball. 2. Results show a much wider variety of offensive-defensive interactions in modern basketball, especially within screening actions, than described in literature. 3. Suggestions for basketball training and coaching are made due to modern basketball’s game requirements. European Journal of Sport Science, vol. 3, issue 3 ©2003 by Human Kinetics Publishers and the European College of Sport Science Hubert Remmert is with the Department of Applied Training Science in the Faculty of Sport Science at Ruhr-University Bochum, 44780 Bochum, Germany.

Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

  • Upload
    others

  • View
    9

  • Download
    2

Embed Size (px)

Citation preview

Page 1: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 1

1

Analysis of Group-Tactical Offensive Behaviorin Elite Basketball on the Basisof a Process Orientated Model

Hubert Remmert

Up to now systematic game observations have insufficiently been used to de-scribe basketball’s tactical structures in detail. Thus, it seems to be urgentlynecessary to evaluate literatures and coaches’ recommendations by objectivegame data. Basketball literature has been analyzed to build a process-orientatedstate-event model that represents players’ offensive-defensive interactions, es-pecially within group-tactical plays in set offense against man-to-man defense.Based on this model, a specific observation system has been introduced todescribe game reality by using the method of systematic game observation.Furthermore, the so-called inter-rater reliability of two separate observationshas been calculated to guarantee a sufficient quality of the observation system(Cohen’s kappa [κ] for each observational category: coefficients between 0.685to 1.000). Sixty games of international elite-level basketball were analyzed byusing the interactive video computer system VIDEO AS. Results show a sur-prisingly wider variety of offensive-defensive interactions (as sequences ofopening action, defensive constellation, and following action within group-tactical plays like screen actions) than described in literature. Suggestions forbasketball training can be made considering also rates of success of the differentinteraction sequences. We conclude that a wider spectrum of group-tacticalaction patterns, especially within screen actions, has to be developed whenpracticing with junior basketball players.

Key Words: training theory, systematic game observation, elite basketball,offensive group-tactical plays, process-orientated model building

Key Points:

1. A process-orientated model was designed to establish a specific method of system-atic game observation to verify and evaluate practical basketball literature’s rec-ommendations with data from elite basketball.

2. Results show a much wider variety of offensive-defensive interactions in modernbasketball, especially within screening actions, than described in literature.

3. Suggestions for basketball training and coaching are made due to modern basketball’sgame requirements.

European Journal of Sport Science, vol. 3, issue 3©2003 by Human Kinetics Publishers and the European College of Sport Science

Hubert Remmert is with the Department of Applied Training Science in the Facultyof Sport Science at Ruhr-University Bochum, 44780 Bochum, Germany.

Page 2: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

2 / Remmert

Introduction

The systematic game observation is a data recording method that is appropriate torecord the objectively observable actions (respectively events) of the game. Internalprocesses cannot be recorded unless clear, indirect indications are presented by theexpressed behavior of players (9).

Up to now, there is no accepted way to transfer players’ internal tacticaldecisions into countable data. This is one of the main reasons why interactionsbetween offensive and defensive players are poorly regarded by quantifying basket-ball game analyses. Therefore, the presented study uses a process-orientated modelto describe the interactions within group-tactical plays (2-, 3-, and 4-person plays) inset offenses against the man-to-man defense1 as indications of players’ tacticaldecisions. The mainly expected use for basketball experts is to verify the immensenumber of recommendations in basketball literature in how to act against differenttypes of defensive behavior within these plays (e.g., 7, 10, 15).

For successful training and coaching, it is extremely important to take adeeper look at the tactical decisions of basketball players while being involvedespecially in group-tactical plays (e.g., give and go and screen actions):

• The tactical abilities of basketball players possess an outstanding importancefor game performance and contain the possibility of compensating inadequatelytrained abilities and talents, like special techniques or physical speed. Tacticalcompetence allows highly effective acting in decisive situations of the basket-ball game to optimize the main goal of success. Therefore the finishing ac-tions of ball possession leading to shots or losses of the ball, the so-called“offensive finishing actions” (13), are the focus of this investigation.

• Group-tactical plays in basketball are influenced by both preliminary andspontaneous decisions. This marks the central role of group-tactical capabili-ties as a link between team and individual tactics in basketball training pro-cesses.

• An earlier analysis could have already proven the outstanding importance ofthe offensive group-tactical behavior for success in elite basketball, althoughthe individual finishing actions (shots, drives, or posting up moves in one-on-one situations) have been quantitatively dominant, with approximately 75%of all offensive attempts (13).

Attempting to analyze group-tactical plays in a basketball game posed severaldifferent problems. They can be summarized under two main questions:

1. Is it possible to establish a practicable method of game observation fordescribing group-tactical interactions as indicators of players’ tactical deci-sions? The research-methodical aspects of adequate modeling and validationof the designed observation system is closely associated with this question.(It is obvious that a precise model-building relating to the analyst’s goal isnecessary to secure the quality of observational results.)

2. Is it possible to identify typical patterns of offensive group-tactical behavioras “strategic rules” (1, 5, 14) in elite basketball using a broad, cross-sectionalanalysis? The main aspects of the presented investigation relating to this

Page 3: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 3

question are verification of recommendations in how to act within group-tactical plays (literature and basketball experts), relation of the investigateddata to the variable of success (scored points), and derivation of training goalsfrom analyzed data.

Materials and Methods

Model Construction

The present study required the use of a process-orientated state-event model thatwas able to represent the games’ interaction stream as a sequence of states (e.g.,constellations of players) and events (e.g., [inter-]actions of players; 4, 5, 11, 12).

The main interest of the investigation consisted of analyzing detailed struc-tures of group-tactical offense-defense interactions—the exact constellations andactions within the offensive group-tactical plays (e.g., screen actions)—above allthe plays used to finish ball possession with the intention to score (offensive finish-ing actions, see above).

The invented model shows on a rough level (see Figure 1) the group-tacticalplays as so-called offensive interaction units2 (OIUs) between the states of offenseand defense (with change of ball possession [CBP]) or between separate offensiveattempts (without CBP). Furthermore the concrete offensive actions (events), asindicators of decisions made by offensive players while being confronted with the

Figure 1 — Rough structure of the built model.

Page 4: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

4 / Remmert

defenders’ behavior (states), are shown on a more detailed level (see Figure 2; directscreen) as favored by experts and basketball literature.

Video Analysis

The particular parameters of the presented model (for the purpose as described)have been transferred into an observation system, which contains additionally moregeneral categories and categorical classes (see, e.g., “Finishing OIU” in Table 1) toclassify and distinguish the recorded game data:

• General categories: division/sex, team, score team A/B, ball possession afterfinishing action, scored points, finishing event;

• Time categories: period, time code (generated automatically);• Spatial categories: finishing position;• Tactical categories: offensive attempt after, offensive phase, reorganization

between offensive phases, opponent’s defense, number of offensive interac-tion units (OIU), previous OIU, overlapping OIU, constellation before finish-ing OIU, finishing OIU, finishing by assist, opening action, state of defender,following action.

To record the immense data, the interactive computer system VIDEO AS(Video Analyzing System) was used. Every single observation unit (time of theselected offensive attempt3) was analyzed by choosing the correct categories and

Figure 2 — Detailed structure of the built model (direct screen).

Page 5: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 5

categorical classes of the observation system. Altogether, 60 games of nationalGerman and international elite basketball4 have been observed. The complete datawere analyzed by using the statistical package for the social sciences (SPSS).

Validation

The research method used in this study has been validated in the context of verifyingthe scientific quality standards objectivity and reliability. It is discussed to be im-possible to verify the quality standard of validity for the systematic game observa-tion as well, because every interactive sport game presents itself as a unique andnon-reproducible event. Environmental conditions and categorical behavior of play-ers vary from time to time, and the research findings cannot be reproduced asdemanded from reliability-tests within the classical test theory (2, 6, 8). As empiri-cal validating requires some kind of classical reliability test, Lames (9) suggests adifferent way of validating the systematic game observation by determining the so-called inter-rater reliability, which proves the formal exactness of the used observa-tion system. Furthermore, external validity can be proven by consistent judgmentsof basketball experts (2, 9). For this study, external validity can be supposed through-out the foundation of model-building by analyzing basketball literature.

The (empirically examinable) instrumental aspect of reliability gives infor-mation about the reliability of the used observation system, not of the determinedresults! In this study the inter-rater reliability on the basis of a so-called matrix ofcorrespondence (of two independent observations) was calculated for every obser-vational category5 (see Table 2).

Page 6: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

6 / Remmert

The calculated kappa [κ] coefficients should be at least 0.80 to 0.85, as de-manded in statistical literature (8). After the calculation process, insufficient coeffi-cients were discussed with regard to the contents of basketball expertise. The resultwas an optimization of the used observation system that can be accepted as valida-tion of the constructed model. Modifications to observational categories and cat-egorical classes were considered at length for the presented results to ensure they arecomparable to future observations. For example, combining the categorical classes“screen” and “passive screen” increased the calculated coefficients of the categories“previous OIU” and “finishing OIU.” Observational experience showed that it wasnot sufficiently possible to divide players’ group-tactical behavior into active andpassive screening actions as described in basketball literature.

Results

The presented results have been confined to group-tactical behavior in general andscreening actions within 2-, 3-, and 4-person plays in detail.

The group-tactical offensive interaction units (OIUs) are used against theman-to-man defense mainly with the intention of preparing (67.3%) or overlapping(69.2%) individual finishing actions (75.8%) to make offensive team play morecomplex and keep the defending team busy by using fake actions. In addition, it isinteresting to examine the average scored points of every finishing OIU. Only thegroup-tactical OIUs reached a score above the average of 0.80 points per offensiveattempt—1.04 points per complete offense (see Table 4)6.

Page 7: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 7

Beyond these results, it is worthwhile to analyze differentiated offensive-defensive interactions within group-tactical OIUs. The direct screen (screeningactions within 2-person plays—active and passive variants) is the finishing actionmainly used above all group-tactical plays (see Table 3; 10.5% of all OIUs includingindividual actions). Contrary to this, the indirect screen (3-person plays) and mul-tiple screen (staggered and double screen within 4-person plays) are used morefrequently as previous (together 49.7%) and overlapping (together 61.3%) group-tactical OIUs.

Page 8: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

8 / Remmert

What about the concrete offensive-defensive interactions within the directscreen (see previous main questions: strategic rules)? In general, there was observeda wider variety of action patterns than described in basketball literature. Further-more, the non-described (and therefore unconsidered by the original model) interac-tions seemed to be more successful (see Figure 3, e.g., drive against sagging underscreen: 55.0% success ratio in comparison to an average of 39.6% of all directscreens).

\insert figure 3\Going into detail, the following differences in players’ interactions, as previ-

ously shown by the original model (see Figure 2), are remarkable. Only the quantita-tively relevant cases are listed:

• defender is screened successfully (without switching): dribbler uses the screenfor shooting in 47 cases (following action: shot; success ratio: 40.4%);

• defender sags under screen: dribbler drives to the basket in 20 cases (drive;s.r., 55.0%);

• defender slides through: dribbler drives to the basket in 19 cases (drive; s.r.,63.2%);

• defender steps out above screen: screener rolls away from the basket in 19cases (pop; s.r., 52.6%);

• defenders switch: dribbler drives to the basket in 73 cases (drive; s.r., 28.8%);• dribbler is double teamed by defenders: dribbler drives to the basket in 33

cases (drive; s.r. 36.4%), screener rolls to the basket in 30 cases (roll; s.r.,36.7%), screener rolls away from the basket in 25 cases (pop; s.r., 40.0%).

Figure 3 — Model “direct screen” and rates of success.

Page 9: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 9

Taking a deeper look at indirect and multiple screens, the determined resultsshow different tendencies. While the direct screen has been more successful whenusing different following actions than the original model recommended, the indirectand multiple screens seem to be mainly successful when acting as favored by expertsand literature (see Figures 4 & 5).

Nevertheless the variety of following actions against defensive constellationsis even bigger than shown by the built model.

Discussion

Because the systematic game observation produces results only describing gamereality, it is urgently necessary to add supplementary information from experts tomake suggestions for basketball practice (3, 9). Obviously it is easier to providespecific training goals out of single observations (e.g., scouting an opponent toprepare a team for one special game) than considering data produced by a broadcross-sectional analysis. In addition to that, it doesn’t make sense only to usebasketball’s statistical norm profiles to guide training processes because the gamechanges dynamically (opponents, strategies, rules, etc.) from time to time. Never-theless, it is necessary to analyze present elite basketball, especially within itstactical structures, to give young and talented players a better perspective on train-ing, which is founded on objective data in addition to coaches’ individual recom-mendations.

Figure 4 — Model “indirect screen” and rates of success.

Page 10: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

10 / Remmert

For this, the results concerning the direct screen are remarkable. Many expertstake the view that the direct screen is mainly of methodical interest in modernbasketball because it helps prepare players to use the more complex indirect screenin game situations. In contrast to this opinion, the results presented here underlinethe importance of the direct screen as the most frequently used finishing actionabove all group tactical plays.

Conclusions

General suggestions for basketball training derived from the presented results are:

• It is worth using more overlapping offensive interaction units to reduce thedefenders’ opportunities to help against finishing actions in one-on-one and2-person plays.

• Offensive players should cut and penetrate as often as they can to engage theirdefenders attention (faking and also preparation of finishing actions by fa-tiguing defenders).

• Offensive set plays should consist of a variety of screening actions (2-, 3-, and4-person plays). Therefore, even unorthodox following actions should bepracticed to increase the success ratio, especially within 2-person plays (di-rect screen).

• Center players should work on their distance shooting capabilities and focuson outside shooting (pop) out of the direct screen.

Figure 5 — Model “multiple screen” and rates of success.

Page 11: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

Group-Tactical Offensive Behavior / 11

• Indirect and multiple screens should be used not only for the purpose ofdistance shooting. In many situations, the drive (dribbler) and roll to thebasket (screener) seem to be more successful, especially when the defenseanticipates the following actions, as recommended by experts.

• When defending screens, basketball players should focus on the switch in 2-person plays and the slide through and/or the stepping out (above the screen)in 3- and 4-person plays—by considering the concrete game requirements!

Notes1The man-to-man defense is the basic and by far most utilized defensive strategy in

elite basketball, and the a set offense is more often played than a fast break (about 70% to30%).

2The OIUs represent those offensive actions (individual or group-tactical plays) thattransfer the course of the game from one state to another. During the basketball game, thismay happen by changing ball possession (scoring or loss of ball: change to the state ofdefense) or not (new offensive attempt, e.g., after an offensive rebounding).

3Offensive attempt: Period of ball possession from gaining the ball to the final shot orloss. This means the possibility exists to start another offensive attempt after a missed shot bysecuring the rebound.

4Sixty games from 1994 to 1999: 6 games of the first German women’s and men’sdivision in each case, 6 games of the second German men’s division, 6 games of women’s andmen’s European league in each case, 5 games of the WNBA, 6 games of the NBA and NCAAin each case, 6 games of women’s international championships for national teams, 7 games ofmen’s international championships for national teams (European championships, Worldchampionships, Olympic games).

5The coefficients C (percentage of correspondence between the two observations) andκ (percentage of correspondence relativized by chance factor) are presented here. κ is alwayssmaller than C because of this relativization by chance (except in the case of perfect corre-spondence).

6Although no statistical correlation could have been proven by using the Chi-squaretest (Pearson’s [U]).

References

1. Barth B. 1999. Strategie und Taktik im Wettkampf. In: Thieβ G, Tschiene P, editors.Handbuch zur Wettkampflehre. Aachen: Meyer & Meyer. p. 361-75.

2. Czwalina C. 1992. Gütekriterien in der beobachtenden Sportspielforschung. In: HagedornG, Heymen N, editors. Methodologie der Sportspielforschung. Ahrensburg: Czwalina.p. 61-72.

3. Hagedorn, G. 2000. Sportspiele. Training und Wettkampf. Reinbek: Rowohlt. 248 p.4. Hein T. 1995. Strategisch-taktische Spielanalyse im Sportspiel Handball. In: Perl J,

editor. Sport und Informatik IV. Köln: Sport und Buch Strauβ. p. 85-96.5. Hein T. 1999. Modellbildung im Sportspiel Handball. In: Wegner M, Wilhelm A, Janssen

JP, editors. Empirische Forschung im Sportspiel. Kiel: Christian-Albrechts-Universität.p. 173-85.

6. Hohmann A. 1994. Grundlagen der Trainingssteuerung im Sportspiel. Hamburg:Czwalina. 332 p.

Page 12: Analysis of Group-Tactical Offensive Behavior in Elite ... · action patterns, especially within screen actions, has to be developed when practicing with junior basketball players

12 / Remmert

7. Krause JV, Meyer D, Meyer J. 1999. Basketball skills and drills. Champaign, IL: HumanKinetics. 206 p.

8. Lames M. 1991. Leistungsdiagnostik durch Computersimulation. Frankfurt am Main,Thun: Harri Deutsch. 257 p.

9. Lames M. 1994. Systematische Spielbeobachtung. Münster: Philippka. 159 p.10. Lieberman-Cline N, Roberts R. 1996. Basketball for women. Champaign, IL: Human

Kinetics. 283 p.11. Perl J. 1997. Möglichkeiten und Probleme der computerunterstützten Interaktionsanalyse

am Beispiel Handball. In: Perl J, editor. Sport und Informatik V. Köln: Sport und BuchStrauβ. p. 74-89.

12. Perl J, Uthmann T. 1997. Kapitel B1: Modellbildung. In: Perl J, Lames M, MiethlingWD, editors. Informatik im Sport. Schorndorf: Hofmann. p. 43-64.

13. Remmert H, Steinhöfer D. 1998. Analyse der individual- und gruppentaktischenAngriffsabschluβhandlungen im Damenbasketball mit Hilfe des interaktivenVideosystems VIDEO AS. Leistungssport 6:47-51.

14. Roth K. 1989. Taktik im Sportspiel. Schorndorf: Hofmann. 277 p.15. Wissel H. 1994. Steps to success. Champaign, IL: Human Kinetics. 215 p.

About the Author

Hubert Remmert has been a lecturer in the Department of Training Science since 1997. Since2000, he has also been responsible for teaching basketball at the same faculty. The author hasbeen an active basketball player and coach for several years in Germany’s top divisions.