Being stuck in the house and without any (or very few) basketball games, we thought it might be fun to share some of our data we use in https://newstats.eu and challenge you to give your predictions on historical data.
We anonymized and uploaded data from the first 5 rounds of 3 different leagues in the corresponding directories.
In each of these folders you will find 2 files:
- train.csv : (example)
It contains the data for the first 5 rounds of the anonymized league. Every pair of rows refers to a single game in that round. As an example:
ROUND,GAME,TEAM,HOME,2P,2PA,3PA,3P,FT,FTA,ORB,DRB,AST,TOV,STL,PTS
1,game_1,team_10,1,19,34,20,7,7,16,9,23,12,23,4,66
1,game_1,team_2,0,21,40,22,10,9,11,8,24,18,16,13,81
In round 1, in game_1, the home team(home=1) was team_10 and away team(home=0) was team_2. The final score was 66-81. The rest of the fields are obvious. The anonymized versions of the teams are only valid within the same league (e.g the team_2 of league1 is not the same team with team_2 of league_2).
- test.csv : (example)
This is the file you need to fill out the last column. This entry:
ROUND,GAME,TEAM,HOME,PTS
6,game_30,team_4,1,
6,game_30,team_3,0,
indicates that in round 6 the home team team_4 plays versus team_3. You will need to fill out the predicted points for each team.
In order for you to participate, just fork the current repo the fill out the test.csv files in the leagues directories. We will rank the submissions on 2 categories:
- Winners prediction (i.e accuracy in predicting the winner in each game)
- Total points in game (i.e accuracy in predicting how many points will be scored in total in each game)
We will publish the winners and you will get a chance to present your approach. You might have questions, comments, so feel free to reach out on our twitter https://twitter.com/NewstatsEu
Deadline: 29 March (tentative)