You are on page 1of 11

1|Page

Math Exploration: Probability that a handball team that became a champion is truly the best
team in the league

Vito Bahtijarević

I am a big sports fan and I play handball. Since I am playing for the Croatian national team and
for PPD Zagreb, I wanted to connect handball with mathematics because it is the topic that
would interest me the most. I wanted to see what are the probabilities of winning a
championship when considering the strength of a certain team because that could change the
way I see the champions in the next years.

Handball is a sport that is played mostly across Europe, but it is starting to be famous in Middle
East, North Africa and some part of South America and Asia. it is played with 6 players and a
goalkeeper. The aim is to throwing a ball into the goal using hands and the team that scores
more goals wins. The most famous handball league in the world is German Bundesliga since the
best clubs in the world are mostly from Germany.

In Bundesliga, a win brings two points, a draw brings one point and a loss brings no points. That
kind of scoring system of handball games is used in almost all of the leagues in the world and
some cup competitions. Bundesliga consists of 18 teams which are analyzed in the season 2015.
/2016. in which every team plays with another two times. It means each team plays 34 games
and the highest number of points possible is 68 points. Number of points is the main criterion of
ranking in the table. In the case of two teams having same number of points such as VFL
Gummersbach and SC Magdeburg in the 2015. / 2016. season (35), the next criterion is the goal
difference (difference between goals scored and conceded. For example, SC Magdeburg had a
bigger goal difference they were in front VFL Gummersbach on the table. At the end of the
season, the table looked like this:
2|Page

P Team G W D L Goals GD Points


1. Rhein-Neckar 32 28 0 4 916:704 +212 56:8
Löwen
2. SG Flensburg- 32 26 3 3 969:785 +184 55:9
Handewitt
3. THW Kiel 32 24 2 6 974:822 +152 50:14
4. MT Melsungen 32 22 3 7 910:825 +85 47:17
5. Füchse Berlin 32 20 3 9 910:825 +85 43:21
6. Frisch Auf 32 19 1 12 888:820 +68 39:35
Göppingen
7. TSV Hannover- 32 14 8 10 891:880 +11 36:28
Burgdorf
8. SC Magdeburg 32 14 7 11 895:880 +15 35:29
9. VfL 32 16 3 13 874:864 +10 35:29
Gummersbach
10. HSG Wetzlar 32 15 4 13 823:822 +1 34:30
11. SC DHfK Leipzig 32 13 4 15 856:904 -48 30:34
12. Bergischer HC 32 9 1 22 815:911 -96 19:45
13. TBV Lemgo 32 8 2 22 847:953 -106 18:46
14. HBW Balingen- 32 6 3 23 850:934 -84 15:49
Weilstetten
15. TVB 1898 32 4 6 22 783:926 -143 14:50
Stuttgart
16. ThSV Eisenach 32 4 2 26 795:1002 -207 10:54
17. TuS N-Lübbecke 32 2 4 26 801:940 -139 8:56
18. HSV Hamburg* 0 0 0 0 0:0 ±0 0:0
1

1
https://de.wikipedia.org/wiki/Handball-Bundesliga_2015/16
3|Page

* HSV Hamburg didn’t play a single game in the league due to bankruptcy

P- position G- games played W- wins Points- points per win: points per loss

D- draws L- losses GD- goal difference

Trophy of the champion symbolizes all of the effort and work which teams and players have
invested during the season. The team that is ranked the best on the table gets the trophy but
the question remains if that team was really the best team that season. It doesn’t matter what
the point difference was between the first and the second team, the first team gets the trophy.
That means that a certain team can become the champion with only a point or not even a point
more than the second team on the table. In that case, people often say that the champion won
the trophy with luck. It can also happen that two teams are the same in all categories and that
they share the first place with only one game to go, but they do not play each other so one
team scores a goal which brings them the trophy of the champion. Can we then say that the
best team won the trophy or is it just one lucky winning of the trophy? In the case of Bundesliga
in 2015. /2016. season, Rhein-Neckar Löwen became a champion with only one point more than
SG Flensburg-Handewitt. Experts could say that Rhein-Neckar Löwen won the trophy because
they were better during the season, but they could also say that the champion was lucky. So, I
am going to check whether Rhein-Neckar Löwen was the best team and if they won the
championship fairly.

To explore the problem, I have made and simulated a model of a fictional Croatian college
league. First, I assumed that all of the teams in this league are equally strong. Someone could
assume that all of the teams will have similar number of points at the end of the season and
that the first team won’t substantially bounce from others. But mathematics will show that it
doesn’t have to be like that in real life. In professional handball every fifth game ends up in a
draw. Meaning, the probability for a draw is around 1/5 or 20%. Since the sum of all
probabilities needs to be 1, the probability that a team wins or loses a game is equal to
1 4
1−5=5
4|Page

I assumed that all teams are equally strong; the probability that a certain team wins the game is
the same to that same team losing a game. The chances are than 50:50. Looking at the chances
and the sum of the probability of winning and losing the game, the probability of winning is
1 4 4
equal to the probability of losing and the probability is × 5 = 10, or 40%. On the basis of these
2

information the league could be simulated. I simulated the championship by hand. If there is a
schedule of the games, for every game I just need to draw one ball from the box with 10 balls. In
the box there were 4 red balls (representing the win of the home team), two white balls
(representing the draw) and 4 blue balls (representing the win of the away team). We can see
that the probability (if we draw with eyes closed) that we draw a red ball is 4/10, same is with
the blue ball, and the probability of the white ball being drawn is 2/10. Those probabilities
match the assumed probabilities of certain results. I drew balls from the box with replacement,
meaning that I drew a ball for a certain team and wrote their result on a sheet of paper. I
repeated the process for each of the teams in the league. With that simulation, I got this kind of
table:

P Team G W D L Points
1. School of Medicine 32 18 7 7 43
2. Faculty of Electrical Engineering and Computing 32 17 5 10 39
3. Faculty of Transport and Traffic Sciences 32 15 9 8 39
4. Faculty of Science 32 17 4 11 38
5. Faculty of Agriculture 32 15 7 10 37
6. Faculty of Law 32 15 6 11 36
7. Faculty of Mechanical Engineering and 32 15 5 12 35
Naval Architecture
8. Faculty of Economics and Business 32 13 6 13 32
9. VERN- University of Applied Sciences 32 12 8 12 32
10. Faculty of Humanities and Social Sciences 32 14 3 15 31
11. Faculty of Pharmacy and Biochemistry 32 11 7 14 29
12. Faculty of Political Science 32 12 5 15 29
5|Page

13. Faculty of Chemical Engineering and Technology 32 11 5 16 27


14. Faculty of Architecture 32 11 5 16 27
15. Faculty of Veterinary Medicine 32 10 5 17 25
16. Faculty of Food Technology and Biotechnology 32 9 5 18 23
17. School of Dental Medicine 32 7 8 17 22
P- position G- games W-wins D-draws L- losses

It can be clearly seen, even though the teams had the same quality, the great range of points on
the table (21 point difference between first placed and last placed team), so the champion is
relatively clear (4 points advantage in front of the vice champion). This is not accidental nor is it
because of the “hand” simulation. Meaning, I will show that in most cases it will most certainly
be the case.

What would happen if a visibly stronger team goes in this equalized league? Almost everyone,
including those who were not amazed by the previous results, would say that the stronger team
will win the championship. Will it? Let’s assume that the better team has a 1.375 times bigger
probability to win the game then to lose it and the probability for a draw is still the same 1/5. In
the next table there are the probabilities of all three possible outcomes for both the best team
and the rest of the league:

Outcome Points Best team Rest of the league


Win 2 11/20 (55%) 4/10 (40%)
Draw 1 5/20 (20%) 2/10 (20%)
Loss 0 5/20 (25%) 4/10 (40%)

That means that the mean value for the best team should be:

11 1 5
34 × ( × 2 + × 1 + × 0) = 44.2~44 𝑝𝑜𝑖𝑛𝑡𝑠
20 5 20

(Adding of the extra team in the league increased the number of played games by 2 for each
team, with each team now playing 34 games in the season).
6|Page

Now, I made a similar simulation as I did with the previous league. I had two separate boxes. For
the simulation including the best team, I used 20 balls, out of which 11 were red for the win of
the better team, 4 white balls for a draw and 5 blue balls for the loss of the better team. The
second box was simulated as I did it the first time, including ten balls. The new league now has
18 teams and look at the miracle – the “best” team was not the first.

P Team G W D L Points
1. School of Medicine 34 19 8 7 46
2. Faculty of Kinesiology 34 20 6 8 45
3. Faculty of Electrical Engineering and Computing 34 18 5 11 41
4. Faculty of Transport and Traffic Sciences 34 15 9 10 39
5. Faculty of Science 34 17 5 12 38
6. Faculty of Agriculture 34 15 7 12 37
7. Faculty of Law 34 15 6 13 36
8. Faculty of Mechanical Engineering and 34 15 6 13 36
Naval Architecture
9. Faculty of Economics and Business 34 14 6 14 34
10. VERN- University of Applied Sciences 34 12 8 14 32
11. Faculty of Humanities and Social Sciences 34 14 3 17 31
12. Faculty of Pharmacy and Biochemistry 34 12 7 15 31
13. Faculty of Political Science 34 12 6 16 30
14. Faculty of Chemical Engineering and Technology 34 11 6 17 28
15. Faculty of Architecture 34 11 5 18 27
16. Faculty of Veterinary Medicine 34 11 5 18 27
17. Faculty of Food Technology and Biotechnology 34 11 5 18 27
18. School of Dental Medicine 34 8 8 18 24
7|Page

I should mention that it doesn’t mean that a team which has around 44 points with 18 clubs in
the league at the end is necessarily the best or it doesn’t mean that the second team on the
table is the best. Results (with the appropriate use of mathematics) just show that the best
team does not or even with high probability won’t be the first on the table.

In the simulation, I have assumed something about the probabilities of wins, draws and losses. It
is much harder to conclude the probability of the champion to be the best team on the basis of
real point distribution on the table. I also assumed that all teams in the league have equal
strength, which is not the case in any league in the world and especially not in the Bundesliga.
Even the probability of 1/5 for draw hides a small trap. In fact, we can assume that probability
only if we don’t know any additional information. When I would know that a certain team
scored more goals, the probability of a draw would drop to around 5%. It would be a small
probability that the team that scores more goals has a bad game and scores less or equal goals
than the team that scores lesser goals per game. If I also know that a game ended in odd
number of goals, the probability of a draw is equal to 0.

Let’s go back to the question what is the probability that in reality the champion became the
best team. If at the end of one round at tip lie two equal teams, they have an equal probability
that they are the best team. But if a certain team lies on the top of the table by herself the
probability of that team being the best in the league grows in comparison to the point
difference towards the followers. If the goal difference is really big the probability is complete,
meaning 100%.

I will define the strength of a certain team as the number of points during the championship,
with a condition that all of the teams are followed by the same amount of luck. Precisely
because of the element of luck, number of points collected during the championship is not exact
toward the true strength of that team, the results will be scattered within a certain range. If I
label X’ as the number of points achieved therefore X’ is a discrete accidental variable which can
be approximated. The curve that describes the probability of point achievement of same
strength teams is Gauss curve. Bell looking curve with a maximum right above the actual points
8|Page

won; which usually represents the true strength of a team. Let’s look at MT Melsungen that
achieved 47 points.

0,06
Mean; 47
0,05
Probability density

0,04

0,03

0,02

0,01

0
0 10 20 30 40 50 60 70 80
Number of points

The height of the points on the curve represent the probability that the real strength of MT
Melsungen is a certain number of points, if we know that at the end of the season MT
Melsungen had µ=47 points, which most certainly represents its real strength. However, that
maximum probability is only (around) 6% because the rest till 100% is distributed to other
possible results. I chose X’ as the number of points and the probability is 6% because I
calculated P(46.5<X<47.5). In the model, I assumed that the probability suddenly drops with the
deviation bigger than 7 points to more or less. Meaning, the probability that a certain number of
points is MT Melsungen’s true strength suddenly drops if that number of points is distanced
from the achieved 47 points by more than 7 points. The curve will look the same for every team
with a slight horizontal translation; every team will have a maximum above the achieved
number of points - µ.
9|Page

Number of points Normal distribution Number of points Normal distribution

26 0,00063312 48 0,056413
27 0,00096201 49 0,054712
28 0,00143223 50 0,051991
29 0,00208921 51 0,048407
30 0,00298598 52 0,044159
31 0,00418147 53 0,039471
32 0,0057373 54 0,034567
33 0,007713 55 0,029661
34 0,01015958 56 0,024938
35 0,01311188 57 0,020543
36 0,01658026 58 0,01658
37 0,02054255 59 0,013112
38 0,02493758 60 0,01016
39 0,02966137 61 0,007713
40 0,03456725 62 0,005737
41 0,03947074 63 0,004181
42 0,04415934 64 0,002986
43 0,04840685 65 0,002089
44 0,05199096 66 0,001432
45 0,05471239 67 0,000962
46 0,05641316 68 0,000633
47 0,05699175

This table shows the probability of MT Melsungen achieving certain number of points

For example, if the observed team had 58 points in Bundesliga, because of the element of luck
everyone knows that may not be the measure of their quality. It is probable that the number of
points achieved by that team is slightly better or worse; and it is less probable that it was much
better or worse result of that team. In this case, anyone can easily think that the real quality of
the team is not represented by the number of points achieved and that it could’ve been slightly
different without changing the quality of the team.
10 | P a g e

If I want to determine the probability that the champion is truly better than the vice champion, I
should compare their Gauss curves. In all possible deviations from the actual achieved points, I
should calculate the probability that the champion achieved “moved” result and that the vice
champion is weaker; and then sum up all of the probabilities. Since, Rhein-Neckar Löwen in
2015. /2016. season achieved 56 points and SG Flensburg-Handewitt achieved 55 points, it
follows that Rhein-Neckar Löwen was actually better with the probability of only 1.6% than SG
Flensburg-Handewitt. That doesn’t say that Rhein-Neckar Löwen (or SG Flensburg-Handewitt)
were actually better than the rest of the league. For that I should make the same kind of
calculation for each team in the league and then compare it. To really prove whether Rhein-
Neckar Löwen was really the deserving champion in the 2015. /2016. season, I would need to
compare probabilities of each team in the league and then compare it with the champion.

In the end, this kind of calculation cannot always be true because of many factors affecting the
results of a certain team. In the beginning, I needed to simplify the calculations and introduce
the impossible situation with every team being the same in strength, so the opinions of many in
this case could be true. I can also discuss whether a probability that a champion was better by
1% or 85% is important information. But these calculations could possibly determine the quality
of the teams in the league and as well predict the champion. For example, THW Kiel in the 2013.
/2014. season had 0% probability that they were the best team in the league after winning the
league at the end because they had the same number of points as the vice champion. But in the
2012. /2013. season THW Kiel had 11% probability that they are really the best team in the
league after they won it. In fact, these information’s could be useful in predicting whether the
champion deserved the title in the certain season. The champion who was in one season better
with the probability of 82% was much better than the rest of the league than a champion in
another season with the probability of 45%.
11 | P a g e

Bibliography
Brueckler, F. M., & Cicvarić, D. (2008). Nogometna matematika i fizika. Zagreb: Školska knjiga.

Haese, Haese, Haese, Mäenpää, & Humphries. (2012). Mathematics for International students;
Mathematics SL (3rd edition). Haese Mathematics.

Wikipedia. (n.d.). Handball Bundseliga 15/16. Retrieved January 25, 2017, from Wikipedia.de:
https://de.wikipedia.org/wiki/Handball-Bundesliga_2015/16

You might also like