ELF Glicko2 (ELO) Ratings

On this page you will find the ELO ratings (Glicko2 model) of all ELF teams that have participated in a single season.

The seasons should be viewed in their own individual way.

The calculation is based on the ELO system known as Glicko2, developed by Mark Glickman in 1995.

The initial value for each team at the beginning of a season is 15.

ELF2021 Glicko2 Ranking

The +/- column shows the gain or loss of points compared to the initial value of 15.

Team	Matches	Rating	+/-
Frankurt Galaxy	12	19,4	4,4
Hamburg Sea Devils	12	15,7	0,7
Panthers Wroclaw	11	15,4	0,4
Leipzig Kings	10	15,0	0,0
Cologne Centurions	11	14,6	-0,4
Barcelona Dragons	10	13,3	-1,7
Berlin Thunder	10	13,3	-1,7
Stuttgart Surge	10	12,0	-3,0

ELF2022 Glicko2 Ranking

The +/- column shows the gain or loss of points compared to the initial value of 15.

Team	Matches	Rating	+/-
Hamburg Sea Devils	14	18,2	3,2
Vienna Vikings	14	18,2	3,2
Frankfurt Galaxy	12	16,5	1,5
Barcelona Dragons	13	16,0	1,0
Raiders Tirol	13	16,0	1,0
Rhein Fire	12	15,7	0,7
Berlin Thunder	12	15,7	0,7
Panthers Wrocław	12	14,3	-0,7
Leipzig Kings	12	13,5	-1,5
Cologne Centurions	12	13,0	-2,0
Istanbul Rams	12	11,0	-4,0
Stuttgart Surge	12	11,0	-4,0

ELF2023 Glicko2 Ranking

The +/- column shows the gain or loss of points compared to the initial value of 15.

This calculation also includes the w/o wins so as not to penalize the teams that were awarded the match as a win.

Team	Matches	Rating	+/-
Rhein Fire	14,0	19,5	4,5
Vienna Vikings	13,0	18,8	3,8
Stuttgart Surge	15,0	17,7	2,7
Frankurt Galaxy	14,0	17,6	2,6
Raiders Tirol	12,0	16,5	1,5
Panthers Wroclaw	13,0	16,0	1,0
Berlin Thunder	13,0	16,0	1,0
Munich Ravens	12,0	15,7	0,7
Paris Musketeers	12,0	15,0	0,0
Hamburg Sea Devils	12,0	13,5	-1,5
Cologne Centurions	12,0	13,5	-1,5
Fehervar Enthroners	12,0	13,0	-2,0
Helvetic Guards	12,0	13,0	-2,0
Barcelona Dragons	12,0	12,0	-3,0
Leipzig Kings	12,0	12,0	-3,0
Milano Seamen	12,0	12,0	-3,0
Prague Lions	12,0	11,0	-4,0

ELF2024 Glicko2 Ranking

The +/- column shows the gain or loss of points compared to the initial value of 15.

Team	Matches	Rating	+/-
Rhein Fire	15	18,9	3,9
Vienna Vikings	14	18,8	3,8
Stuttgart Surge	13	18,1	3,1
Paris Musketeers	14	17,6	2,6
Munich Ravens	13	16,7	1,7
Raiders Tirol	12	16,5	1,5
Madrid Bravos	13	16,0	1,0
Cologne Centurions	12	15,0	0,0
Panthers Wrocław	12	15,0	0,0
Berlin Thunder	12	14,3	-0,7
Milano Seamen	12	13,5	-1,5
Frankfurt Galaxy	12	13,5	-1,5
Barcelona Dragons	12	12,0	-3,0
Hamburg Sea Devils	12	12,0	-3,0
Fehérvár Enthroners	12	12,0	-3,0
Helvetic Mercenaries	12	11,0	-4,0
Prague Lions	12	11,0	-4,0

Possible uses

Estimate winning probabilities

These values can be used, for example, to estimate how high the winning probabilities of individual pairings are.

The so-called K-value is used for this. With an Elo difference of 400, the probability of the team with the higher value winning is extremely high.

Example

The 2024 Rhein Fire (.917) with a rating of 18,9 would win 99.2% of the time against the Prague Lions (.083) with a rating of 11 in this model.

If the game is likely to be evenly matched, this value would be lowered.

Example

The 2024 Raiders Tirol (.667) with a rating of 16.5 against the Berlin Thunder (.417) with a rating of 14.3 are closer to each other, so you would lower the K-value here, in this example to 150. This would result in a win probability for the Raiders of 59.5%.

Strength of Schedule

By adding the ratings of the previous season of a division or inter-division opponents, you can create a Strength of Schedule (SOS).

You can find the ELF 2025 Strength of Schedule here

Example

In the following example, we take the Glicko2 values from the (imaginary) match Rhein Fire (18.9) vs. Prague Lions (11.0), which with a K-value of 400 results in the following win probabilities:

Rhein Fire: 97.67

Prague Lions: 2.33 %

Effective Rating Difference (𝐷𝑅)

Description: The effective rating difference (𝐷𝑅) is calculated using the win rates (𝑊) of both teams and a scaling factor (𝐾)

$DR=K\cdot \left(\log _{10}\left({\frac {W_{A}}{1-W_{A}}}\right)-\log _{10}\left({\frac {W_{B}}{1-W_{B}}}\right)\right)$

The idea behind the formula is to translate the strength of the teams (based on their win rate) into a comparative value. The difference between a strong win rate (Rhein Fire: 91.7%) and a weak win rate (Prague Lions: 8.3%) is heavily weighted. The logarithmic ratio ensures that larger differences are exponentially more significant. The 𝐾 value multiplies the result to bring it to a readable scale.

Win Probability Formula (𝐸)

Description: The win probability is calculated using the effective rating difference (𝐷R) and a normalization constant 400

$E={\frac {1}{1+10^{-DR/400}}}$

Why is the 𝐾 value 400 and what does it mean?

Meaning of the 𝐾-value

The 𝐾-value is a scaling factor that determines how strongly the difference in the win rates influences the calculated difference (𝐷𝑅). It is used to adapt the calculations to the conditions of a particular game or league.

Why was 𝐾=400 was chosen?

K=400 is a widely used value in rating systems such as Glicko2 or ELO. It ensures that differences between teams or players are displayed on a scale that can be easily interpreted.

Impact

A high 𝐾 value (e.g. 400) makes differences between teams more significant. Even small differences in win rates can lead to large differences in 𝐷𝑅 can lead to large differences.

A low 𝐾-value (e.g. 100) would give less weight to differences between teams, meaning that the outcome of a game is harder to predict.

Conclusion

The 𝐾 value adjusts the weight of the calculation to the requirements of the league or match. Higher values make differences clearer, lower values equalize the predictions. In this example, 𝐾=400 was chosen to emphasize the large differences in performance between Rhein Fire and Prague Lions.

Example with Values (Rhein Fire vs. Prague Lions)

Effective Rating Difference (𝐷𝑅)

For Rhein Fire (𝑊𝐴 = 0.917), Prague Lions (𝑊𝐵 = 0.083), 𝐾=400

$DR=400\cdot \left(\log _{10}\left({\frac {0.917}{1-0.917}}\right)-\log _{10}\left({\frac {0.083}{1-0.083}}\right)\right)$

Win Probability (𝐸)

Using the calculated 𝐷𝑅=835.6

$E={\frac {1}{1+10^{-DR/400}}}$