Profile| Accreditation | Circulation Maps | Publications | TV Graphics | Internet | Scouting | Studies | CD-Roms | News |Contact

LINE-UP PERFORMANCE ANALYSIS AND PREDICTION

The English Alphabet is well known to all the civilized world and therefore most of the people of this planet are familiar with it.

It consists of 26 letters, out of which, we can make words, which have various meanings. I’m pretty sure, all of you are familiar with this, and if not, you should consider going back to school!
Let’s pick up 12 random letters out of the alphabet. It can be any twelve, and let’s try to make some 5-letter WORDS.
Here, we already have created six words and possibly there are many more.
However, it is obvious that some combinations of letters make words with meaning and some don’t, like these, which make no sense at all.

Probably or certainly this group is larger, that means there are more combinations of letters, which don’t make sense at all!
The question is: How many combinations of letters is it possible to have?

In other words: If one has 12 letters and wants to create 5-letter words, how many of those can one build?
Let’s look at a simple example.

Out of the 4 letters A,B,C, D we can make 6 combinations of two-letter “words”:

AB, AC, AD, BC, BD, and CD.
Now, if we are also interested in the position of the letters, then, we have not only AB but also BA, that makes 2 permutations, in each, and finally we end up with 12 permutations.
What I mean by this is that a word can be an anagram of another.

In other words, we can make in this sense, not only STEAM but also TEAMS, not only TRAIN but also INTRA and, in both cases, the words have totally different meanings.
Mathematics can help us here to calculate the total number of combinations,
and if we take permutations into account, i.e. if we change the position of the letters, then the total number is a lot larger.
I know these mathematics are “all greek to you” but from our first example of 12 letters with 5-letter words, the total number of resulting combinations is 792,
and since I would like you to believe me, this number comes out of all these simple multiplications.

I know you are sports / basketball people but what does this have to do with you?
Think that the coach has a roster of 20-26 players, like the letters of the alphabet, selects 12 for each game and uses 5 at a time.

Well, according to maths, he has to choose one combination of 5 players out of 792 possible ones, each time he decides to make a substitution.
Like the alphabet, some make sense, some don’t, some are good, some are bad, and this can be the case if we use anagrams.

The equivalent in basketball is moving, for example, the “play maker” to position “5”, or the “center” to position “1”, what we call permutation in mathematics.
The coach knows very well, without even giving it a try that some combinations are totally out of question.

So, before the game he prepares some line-ups, but during the game he tries some line-ups, perhaps by accident (for example, if a player has to come out with 5 fouls or injury).
Let’s see that mechanism.

During the game the coach asks a player to come out and another to come in.

Later on, the same happens. So, gradually, during the game, he makes a lot of substitutions.
Thus, a number of questions arise.
First of all: How many substitutions the coach makes.
Second: How many times does the same line-up appear on court.
Third: How many different line-ups (combinations) appear on court.
And of course, which line-up is best.
We studied line-up performance very carefully and in 1995 during Eurobasket we applied it for the first time.
To go back to the triplet 26-12-5 we can assume that a coach has most of the times a roster of 20-26 players.

For a match he has to select 12 and on court he has to choose 5 at a time.
We will look at the roster of AEK, the champion and leader of this year’s regular season.
AEK will serve as an example for this presentation not only for the impressive season 2001 - 2002 they had, but also because coach Dragan Sakota, belongs to this group of professionals with high intuition for fundamental coaching, like, Ioannidis,Alexandris,Petropoulos, Subotic, Ivkovic, Obradovic, Serf and A1 rookies Yannakis, Pilafidis
Let’s concentrate on the game AEK – Panathinaikos 86-62, a victory of AEK.

Sakota selected these 12 players for the game of Day 7.
These 5 players are the starting line-up. Of course, according to what we proved previously this is one line-up out of the 792 possible ones.

Holden, Zissis, Dikoudis, Betts and Kakiouzis are the chosen, and as we will see later in the game did not have the best performance.
First of all we have a list of all substitutions performed by each team with all necessary information, such as time and number of players.

Next to them, we have all statistics related to each line-up so we can safely say that the first line-up, the starters of AEK, did 12 points, 4 rebounds, etc. in 8 minutes with an evaluation index of 0.19.

On the other hand, the starters of Panathinaikos did 11 points in 6 minutes with an evaluation index of 0.33 and managed to lead the game with one point (10-11).

Instead of going into complicated calculations for the evaluation index, think of it, as if it were something like scoring per minute, but don’t forget the winner has always better tendex than the loser..

This means that if each line-up makes a score of one point difference and the coach makes 20 substitutions the game will end up with 20 points difference.

So, actually, we are trying to calculate the contribution of each line-up to the final score.

Anyway, in this game, Sakota used two line-ups in the first quarter by substituting Mihalis Kakiouzis with Arian Komazec at 01:44, improving his team’s performance impressively.
These graphs display performance of each team during the game.

They show the tendex of each line-up, but, as I said before, think of scoring, or points per minute.
If a line-up scores 2 points per minute, it is better than another, which scores 1.5.
Ideally, every time the coach makes a substitution, he must bring on the court a better line-up, so this diagram does not have ups and downs but only ups.

The situation is similar to a group of 5 workers who build a wall: the wall of difference of two scores. If a substitution is not successful the next workers destroy the wall, which is inherited to them.

And of course, if the substitution is successful then workers carry on building, increasing the difference.

In any case, Sakota used 15 “teams” during the game and he repeated the same line-up only once (end of 3rd quarter – start of 4th), whereas Obradovic used 18 “teams”, three of which were repetitions.
We examined for many years all these in the course of a whole championship of 26 days and we came to various conclusions.

For these conclusions, we used all line-ups, that is, all combinations of players used by the coaches, with their total time, total statistics and performance.
Before the change of the 30” rule a “healthy” team used around 100 line-up combinations during the year, a coach without players, i.e. without choices used around 70, and a coach who had not been able to decide on certain schemes had to try 130 or even more.

Today, these numbers are 110, 150, 180.
Let us look at the top 10 line-ups, combinations, of AEK, sorted by their total time of play.

First of all, Sakota does not depend on one line-up. We can see that the three first line-ups have played more or less equal time during the year.
The best performance belongs to the first selection that means that the nose of the coach is good and if we look at a game-by-game analysis we realize that its performance is very stable, thus the average of 0.580 has been like this in most of the games.
On the contrary, the fourth line-up, which we examined previously, and has an overall index of 0.444, has been unstable, which is the case because of the strength of the opponents.

This means that in some games the opponent is easy to beat, in some the opponent is strong.
A week before the start of the game, the coach is equipped with all this information and of course with all the corresponding data for the opponent.

He can check his good line-ups and select the stable ones in parallel to video analysis, which is offered along with the time codes shown in the previous slide.
Using a stable line-up, the coach can be sure of the expected performance.

At the same time, from the lists he can have an estimate of the opponent’s expected performance and he can act accordingly.

The question is: How long does it take to realize all these after each substitution?
The mechanism is like in elections, where you have the previous results and using a sample of 10% from the current data you estimate the new result.

The only difference, of course here, is that you can alter the result by making a substitution to improve your team’s performance.
Of course, the coach needs a good supporting team, which, during the game, compares the lists, looks at the monitors, in order to follow the current performances.

All these should be performed in parallel to the traditional activities of the assistant coaches.

< Back