Prediction competition for the Euro 2020 football winners from eRum 2020.

The slides from the talk can be found here.

To participate in the competition you need to upload a $7 \times 24$ prediction matrix. The 24 columns represent the 24 teams and the 7 rows represent the 7 different ranks that can be determined from the tournament: 1st, 2nd, 3rd, 4th, 5th-8th, 9th-16th, 17th-24th.

The prediction matrix should be uploaded here as a PULL REQUEST of an R matrix object that can be read by load() on R v4.0.0 or higher.

The order of the columns (the 24 teams) will be provided here after the final play-off match in November 2020. Then we know the names of all the 24 teams.

The elements of the prediction matrix should be the probabilities that the given country (column) will obtain a specific rank. Consequently, all columns must sum to 1.

The rows of the prediction matrix must sum to c(1, 1, 1, 1, 4, 8, 8), respectively. These numbers represent the number of teams that can be part of each rank.

The deadline for submission i Sunday June 6th, 2021 at midnight CET.

We will use the Tournament Rank Probability Score (TRPS) to determine the best tournament prediction.

The TRPS is defined in this paper and implemented in the socceR package which can be installed from CRAN:

install.packages("socceR")

Smaller numbers represents better predictions.

library("socceR")
m1 <- matrix(c(1, 0, 0, 0, 0, 1, 0, 0, 0, 0, .5, .5, 0, 0, .5, .5), 4)
m1 # Prediction where certain on the top 2 ranks

##      [,1] [,2] [,3] [,4]
## [1,]    1    0  0.0  0.0
## [2,]    0    1  0.0  0.0
## [3,]    0    0  0.5  0.5
## [4,]    0    0  0.5  0.5

trps(m1, c(1, 2, 3, 4))

## [1] 0.04166667

You are free to model the tournament results in any way you want. You can even conjure them up. Run a complicated random forest model. Or make educated armchair guesses. All that counts is the 7 x 24 matrix.

Hopefully we can summarize some of the approaches on eRum 2022. In any case we will have updates and the winner listed here on GitHub.