TRENDS IN OFFENSIVE TEAM ACTIVITY IN BASKETBALL

The purpose of this paper is to analyze the structure of the models of offense and to determine their effi ciency. We were able to determine the teamwork structure by using a system of analysis of the offensive process. The analysis of the team activity as an integrated whole becomes very important. For this research we observed the Tallinn University basketball team, which plays in Division One of the Estonian league. The data was gathered from 600 ball possessions in 8 recorded games of the regular season. The data we collected was analyzed by the means of data mining. This research has helped to work out the basics of the analytical system of the teamwork aspect. The analyzing system of the competitive activity of the game, enables us to fi nd out interesting offensive models from the data. The results show that the team scoring (points/possession) as the main indicator of analyzed offenses was equal to 1.13 on average with a frequency of 48%. The 1—2 s duration period of ball possession in offensive zone proves to be most effective (scoring 68%). The most effective transition period is under 1.82 s (62%). The scoring of set offense is 44% in the mean. On the basis of this information, the coach can fi nd more convenient time lapses in the game’s performance. The analyzing system worked out through that, helps coaches to develop performance and promote learning.


INTRODUCTION
T he main aim of the offensive teamwork is to score points.During the game there are changes in the structure of the game, which are caused by different reasons.For instance, it may depend on the tempo, the defensive formation of the opposite team or on the variety of used offensive models amongst others.It is necessary to analyze this information in order to arrange the training process more effectively.The purpose of this paper is to analyze the structure of the models of offense and to determine their efficiency.To the present day, scientificmethodical analysis of literature shows that the shooting technique, the physical preparation and the physiological aspects of the players have been studied the most (Hudson, 1985;Mcinnes et al., 1995;Miller, Bartlett, 1996;Rojas et al., 2000).Some studies have concentrated their attention on the effi ciency indicators of the game.L. Mendes and M. Janeira (2001) have found that the main impact factor that defines winning and losing teams is defensive rebounding.J. Sampaio and M. Janeira (2003) have found in their research, that in close games (those with fi nal score differences between 1 and 8 points), game type infl uences the result (regular season or play-off) and game location (home or away).Globally, regular season profi le was best discriminated by successful free-throws, whereas play-offs profi le was best discriminated by offensive rebounding.
Team ball possession is considered one of the central indicators.Both offense and defense activity effi ciency is determined by dividing the points earned by ball possession (points/possession) in the game.This kind of a method helps to compare the team's activity effi ciency in different games (Trezeguet, 2003).
It is common knowledge that basketball is a complicated team sport game.Because of that incorporation of the separate technical and tactical elements into the integral whole is one of the most important tasks in the analysis and synthesis of the coaching process.The main idea of our research is based on the principal that the integrative usage of the technical and tactical preparation of the basketball players in it's whole meaning opens new perspectives in the quality development of the game of basketball.

METHOD
The analyzing system of the given research requires the use of video-recorded games and a computer form of notational analysis.The Tallinn University basketball team was under observation during their participation in Division One of the Estonian Championship.Data was gathered from six hundred (n = 600) possessions in eight (n = 8) recorded games.Further processing of the recorded game was done using the DartTrainer Basic version 2.5.3 programme (Dartfish, 2001).
During the observation of the recorded games the following data was fi xed: • the type of offense: fast break, early offense or set offense; • the beginning: after steal, defensive rebound or inbound; • the time in possession during the transition, ball possession in the offensive zone and total time; • the count of elements in the offensive zone during one team's ball possession: dribbles, passes, screens on the ball, screen off s the ball, shots; • the outcome actions.
The time of ball possession starts from the moment when the player catches the ball and ends when the ball either leaves the hands of the shooter or when the opposition takes possession of the ball.Time is stopped in rebound and out of bounds situations.The time of transition and the time of ball possession in the offensive zone were fixed separately.The time duration was measured to an accuracy of 0,02 sec.When fixing the number of elements, depending on the need, one segment of ball possession was observed many times using slow motion or observation frame by frame.
The gathered data was analyzed by the means of WizWhy programme (data mining).Programme summarises the data and presents the main patterns.It meets this target by listing the relations between all the values in each fi eld and the dependent variable.The method employs a unique algorithm that segments numeric fi elds in an optimal way and displays the relation between each interval and the value under analysis.WizWhy lists the rules that relate between the dependent variable and the other fi elds.This analysis of the basic rules and trends results in the summary of the data.The rules are formulated as "if-then" sentences (WizSoft, 2002).National Basketball Association (NBA) coaching staffs started to use PC-based data mining application to discover interesting patterns in basketball data in the 1995-1996 season (Bhandari et al., 1997).
The trend report presents the one-condition relations in the data, and as such it summarises the data.If-then rules represent suffi cient conditions (the "if" condition is a suffi cient condition for the result).Minimum confi dence of the if-then rules is equal to 0.66 and if-then not rules is equal to 0.72.Maximum number of conditions in a rule: 3.

RESULTS
The results of summary report showed that the team scoring (points/possession) as the main indicator of analyzed offenses was equal to 1.13 on average with 48% frequency.The scoring of analyzed set offenses was equal to 1.06 on average with 44% frequency.
Selection of if-then rules (association rules) for the team scoring (dependent variable) more than the average 1.13 (predicted value) is shown in Table 1.

DISCUSSION
Our analyzing system gives an opportunity to determine the effi ciency of the models of offense.The trend report showed that the fast break was the most effective (66%) game type.The effi ciency of the early offense was equal to the team average.The team used the type of set offense in 77% of all ball possessions with effi ciency equal to 44% (see Figure 1).This results are in accord with earlier research made by Tavares and Gomes (2003) where they found that the main game method used by high level junior teams was the set offense with a frequency equal to 74.6%.Our analysis of the one-condition relations between type of beginning and scoring showed that the efficiency of offense was higher when it starts after steal (59%) or defensive rebound (51%).In situations (302 cases) where the team began their offensive activity with inbound in the defensive e zone the effi ciency (43%) was more-or-less equal to the mean scoring of set offense.Such information gives us the opportunity to rate the team performance separately by offensive types.The analysis of the time in possession showed that the ball possessions (79) with a total time duration of up to 6.04 sec had the 68% effi ciency.The most offenses in this time period belong to fast breaks (Tavares, Gomes, 2003).The scoring drops when time in possession increases (Figure 2).
In fi gure 3 we can see the analyzed team's onecondition relation between scoring and time of ball possession in the offensive zone.The trend report showed that the most effective (68%) duration period of ball possession in the offensive zone was between 1-2 sec.Mainly this period responds to fast break situations.The suffi cient conditions for the result are shown in Table 1 (rules nr 4, 8, 10, 12).It was followed by the 3-10 sec period with 52% effi ciency (317 cases).11-16 sec (165 cases) was the lowest scoring period (36%).The fi nal duration period of ball possession is between 17-36 sec (65 cases).It can happen in situations where players of the opposite team commit a personal foul which is followed by an inbound or the team wins the offensive rebound.The scoring in this period was equal to the team's average.This kind of information tells us that the team has to improve the uninterrupted offensive activity with time duration in the offensive zone longer than 11 sec and increase the use of fast breaks and set offenses with a time duration in the offensive zone between 8-10 sec.
Analysis of the transition period showed that team scoring was higher (62%) in situations with a time duration under 1.82 sec or between 5-8 sec (52%).The fi rst time period belongs to fast break situations.The second is followed by set offense.On the received information we can presume that there are two effective models of offense -fast break with a time duration of transition under 2 seconds which is followed by time in possession  in the offensive zone between 1-2 sec and set offense with time of transition equal to 5 sec and time in possession in the offensive zone between 8-10 sec (see Table 1).
Figure 4 illustrates the one-condition relations between the sum total of all the elements the team uses in the offensive zone during one possession and the result.We can see that, when increasing the number of elements, the effi ciency of the offensive teamwork drops.Scoring was higher when the team used under 4 elements.In game situations this means making a fast break in the offensive zone with the minimal number of elements with high intensity.Two elements are usually equal to pass or dribble and lay-up; three-dribble, pass, lay-up (see Table 1, rules 1, 3, 5, 7, 11).Adding a fourth element in game situations brings the team to early offense, often ending with a jump shot.The next two groups belong to the set offense.In the fi rst, depending on the game situation, the team should perform 5-10 elements.For example 9-10 elements holds the dribbles of 1-2 pla-yers, 2-3 passes, 3-4 screen off's the ball and a shot.If the combination is done skillfully the time duration of ball possession in the offensive zone is between 8-10 sec.The effi ciency of this group was equal to 47% and exceeded the average level of set offense (44%).
In the second group, the team had used 11 or more elements.In the main the time duration of this group is more than 11 sec.The effi ciency of this group (42%) drops under the team average level of set offense.Regulated time (24 sec) that limits offense activity, starts to have it's effect at the ending of the offense.Examples of activities that are not conducive to good results include: using too many elements and performing a shot in unfavorable conditions etc.
The count of screens (both on and off the ball) indicate the combination the team has used.With regards to this point the combination with 2 screen off's the ball scored less (78 cases with 42% effi ciency).The scoring was higher when the team used 1 (32 cases with 56% effi ciency) or 4 (58 cases with 50% effi ciency) screen-off's the ball (see Figure 5; Table 1, rules 14, 22).In situations where the team used screens on the ball the scoring was lower (see Figure 6).This information points out the advantages and disadvantages in the team performance.In this case, the players of the team under analysis have to collaborate using 1, 3-4 screen off's the ball.The team should improve the activity in situations where players are using screens on the ball and combinations with 2 screen off's the ball.One possibility to improve the team activity with screens on the ball is to use lay-ups at the end of the offense (see Table 1, rule 21).
The analyzed team's relationship between scoring and a number of passes during ball possession in the offensive zone can be seen in Figure 7.

TRENDS IN OFFENSIVE TEAM ACTIVITY IN BASKETBALL
There were 409 offenses when players used under three passes.The scoring (53%) of those offenses was above the average (48%).The scoring drops (34%) when players passes the ball between 4 and 6 times (119 cases).
According to the research information we can pressume that effi ciency of the team activity increases by means of fast breaks and skillfully done set offenses.The results of our earlier research (Bazanov et al., 2005) show that both of them have to be done with high teamwork intensity.These results are in accord with research made by E. Tsamourtzis et al. (2005) where they found that the winners indicated more fast breaks, with more successful two point shots and more fast break situations 1 on 0. They concluded that the fast break constitutes an important factor for the attainment of the victory.

CONCLUSIONS
The aim of this study was to analyze the structure of the models of offense and to determine their effi ciency.According to the research a number of conclusions can be made.The team scoring (points / possession) as the indicator of analyzed offenses was equal to 1.13 on average with 48% frequency.The average scoring of set offense was equal to 1.06 with 44% frequency.The offenses with the total time duration of up to 6.04 sec had 68% effi ciency.On the received information we can presume that there are two effective models of offense -fast break with a time duration of transition under 2 seconds which is followed by time in possession in the offensive zone between 1-2 seconds with the minimal number ( 3) of elements in the offensive zone including lay-up and set offense with time of transition equal to 5 sec and time in possession in the offensive zone between 8-10 sec.The main discovered weaknesses of the team under analysis which have to be improved are uninterrupted offensive activity with time duration in the offensive zone longer than 11 sec and situations with screens on the ball.One opportunity to improve both of them is to use lay-ups at the ending of the offense.
On the basis of such information the coach can evaluate the of the team, fi nd more convenient time lapses in the game performance and correct the strategy for future games.The analyzing system worked out through that helps coaches to develop performance and promote learning.

Table 1 . If-then rules for the scoring more than 1.13 points per possession Note
. Minimum confi dence of if-then rules (successful) = 0.66; if-then not rules (unsuccessful) = 0.72.