Norwegian Soccer Football Predictions
Norwegian
Soccer Football
Although I no longer provide Norwegian soccer predictions,
please examine the results for Norwegian football for 1994 and
1995 to see how well the system works for the Premier Division,
First Division (Group 1) and First Division (Group 2).
Your suggestions are always welcomed.
Premier Division-1994 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Rosenborg 101.9
2. Start 100.9
3. Lillestroem 100.8
4. Viking 100.6
5. Kongsvinger 100.1
6. Tromsoe 99.8
7. VIF Fotball 99.7
8. Brann 99.7
9. Ham-Kam 99.5
10. Bodoe/Glimt 99.3
11. Sogndal 99.1 Relegated
12. Stroemsgodset 98.7 Relegated
First Division Group 1-1994 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Strindheim 101.5 Promoted
2. Droebak/Frogn 100.7
3. Lyn 100.5
4. Stabaek 100.5 Promoted
5. Skeid 100.3
6. Moss 100.0
7. Nardo 99.9
8. Tromsdalen 99.7
9. Alta 99.9
10. Mjoelner/Narvik 99.5
11. Stj./Blink 99.4 Relegated
12. Baerum 99.4 Relegated
First Division Group 2-1994 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Hoedd 101.2 Promoted
2. Molde 100.8 Promoted
3. Fyllingen 100.8
4. Eik/Toensberg 100.4
5. Jevnaker 100.1
6. Bryne 100.0
7. Fana 99.9
8. Vard/Haugesund 99.7
9. Aandalsnes 99.6
10. Aasane 99.4
11. Vidar 99.1 Relegated
12. Mjoendalen 99.1 Relegated
The next table shows the average probability of a home win
(H), draw (U) or away win (B) and the actual fraction of
games when an H, U or B happened, for each division. For
example, in the First Division, Group 1, the average
probability of a U was 0.19 while a U actually happened in
0.18 (18%) of the 132 games.
Accuracy of 1994 Predicted Probabilities For Each Division
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Predicted Actual Games
H U B H U B
----------------------------------------------------
Premier 0.54 0.19 0.27 0.53 0.23 0.24 132
First Group 1 0.54 0.19 0.27 0.52 0.18 0.30 132
First Group 2 0.54 0.20 0.26 0.52 0.20 0.28 132
----------------------------------------------------
All Games 0.54 0.20 0.26 0.53 0.20 0.27 396
The next table shows the actual result for each group of
predicted probabilities. All three divisions are combined.
For example, for all games in which the probability of an H
was between 0.40 and 0.59, the average probability of an H
was 0.50. For those games, an H actually happened in 0.50
(50%) of the games.
Accuracy of 1994 Predicted Probabilities for Each Range
(396 Games)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Predicted Actual
Range H U B H U B
----------------------------------------------
0.00 to 0.19 0.16 0.16 0.12 0.33 0.16 0.21
0.20 to 0.39 0.32 0.24 0.30 0.36 0.26 0.27
0.40 to 0.59 0.50 0.00 0.47 0.50 0.00 0.41
0.60 to 0.79 0.68 0.00 0.65 0.63 0.00 0.44
0.80 to 0.99 0.86 0.00 0.00 0.74 0.00 0.00
----------------------------------------------
All Games 0.54 0.20 0.26 0.53 0.20 0.27
Premier Division-1995 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Rosenborg 101.6
2. Bodoe/Glimt 100.8
3. Viking 100.4
4. Molde 100.4
5. Lillestroem 100.4
6. Tromsoe 100.4
7. VIF Fotball 100.1
8. Start 100.0
9. Stabaek 99.8
10. Brann 99.6
11. Kongsvinger 99.3
12. Hoedd 99.3 Relegated
13. Ham-Kam 98.8 Relegated
14. Strindheim 98.8 Relegated
First Division Group 1-1995 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Sogndal 100.9
2. Droebak/Frogn 100.9
3. Moss 100.7 Promoted
4. FK Haugesund 100.3
5. Aalesund 100.3
6. Aasane 100.2
7. Bryne 99.7
8. Fyllingen 99.4
9. Fana 99.4
10. Aandalsnes 99.3 Relegated
11. Sarpsborg 99.0 Relegated
12. Vard/Haugesund 98.8 Relegated
First Division Group 2-1995 Final Computer Ratings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1. Stroemsgodset 101.5 Promoted
2. Skeid 101.4 Promoted
3. Eik/Toensberg 101.3
4. Odd 101.0
5. Tromsdalen 100.4
6. Lyn 100.1
7. Jevnaker 99.4
8. Nardo 99.3
9. Sandefjord 99.0 Relegated
10. Alta 99.9 Relegated
11. Staalkemaeratene 98.9
12. Mjoelner/Narvik 97.9 Relegated
The next table shows the average probability of a home win
(H), draw (U) or away win (B) and the actual fraction of
games when an H, U or B happened, for each division. For
example, in the First Division, Group 2, the average
probability of an H was 0.53 while an H actually happened in
0.53 (53%) of the 132 games.
Accuracy of 1995 Predicted Probabilities For Each Division
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Predicted Actual Games
H U B H U B
----------------------------------------------------
Premier 0.53 0.15 0.32 0.49 0.20 0.31 182
First Group 1 0.55 0.21 0.24 0.46 0.25 0.29 132
First Group 2 0.53 0.17 0.30 0.53 0.14 0.33 132
----------------------------------------------------
All Games 0.54 0.17 0.29 0.49 0.20 0.31 446
The next table shows the actual result for each group of
predicted probabilities. All three divisions are combined.
For example, for all games in which the probability of an H
was between 0.60 and 0.69, the average probability of an H
was 0.69. For those games, an H actually happened in 0.67
(67%) of the games.
Accuracy of 1995 Predicted Probabilities for Each Range
(446 Games)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Predicted Actual
Range H U B H U B
----------------------------------------------
0.00 to 0.19 0.14 0.15 0.12 0.05 0.18 0.17
0.20 to 0.39 0.31 0.23 0.29 0.29 0.22 0.31
0.40 to 0.59 0.50 0.00 0.48 0.45 0.00 0.46
0.60 to 0.79 0.69 0.00 0.66 0.67 0.00 0.75
0.80 to 0.99 0.87 0.00 0.87 0.78 0.00 0.50
----------------------------------------------
All Games 0.54 0.17 0.29 0.49 0.20 0.31
My name is Ray Stefani. I am a professor of electrical
engineering at the California State University, Long Beach.
It is said that the older you get, the better athlete you
were. I did (actually) compete as an undergraduate at Notre
Dame in swimming (making eligibility for the 1960 Olympic
Swimming Trials at 1500m) and as a graduate student for the
University of Arizona in bowling. For about a year I trained
for the modern pentathlon, choosing to return to a Ph.D.
program rather than the Olympic Trials in 1968. I became
interested in sports predictions as a result of my Ph.D.
topic in statistical estimation.
From 1971 until 1980 I provided weekly predictions for
American football to a number of newspapers. During 1981-82
I spent a sabbatical in Switzerland where (in addition to
technical work in my field) I studied European
betting systems (tipping). From 1980 to 1994 I did
predictions for Australian Rules Football. You may find the
section "How the System Works" to be interesting. Recently,
I have been interested in rating systems for the major
international sports and in the physiological explanation of
the differential performance of male and female Olympic
champions.
I have traveled to Norway many times and traveled the
circuit from Oslo to Stavanger to Bergen to Oslo often and I
once took a boat trip from Bergen to Kirkenes and back. I
can read Norwegian fairly well but my spoken knowledge
leaves a lot to be desired.
I thought I might use the WWW to provide predictions for
Norwegian Football. I hope you find them to be interesting.
|
