Tournament Elo Rankings

  • It seems the Tournament rankings do not take into account the number of games played. Thus players will not play the full number of games in order to protect their rating. A lucky player could play only a few games, get a high Elo rating, and then stop and sit on the rating.

    I suggest some fixes for tournaments:

    1. The Elo rating be multiplied by the number of games played divided by the total number of maximum games in the tournament. This could be applied during or after the tournament to determine the final standings.
    2. Disqualify players in the final standings if they do not complete the full number of games.

    This penalty probably should also be extended to non-tournament rankings. Inactive players that have not played a minimum number of games in a given month (say the average number of games played per player) could be penalized each month (say 1 percent). This penalty could also be scaled down proportionally for each game played in a given month (up to the monthly minimum number of games).

    So, for example, if the average number of games played in a given month per player is 10, and the current leader has an Elo ranking of 1,467 and only plays 5 games in that month, then they would be penalized (1%)(1,467)(1 - 5/10) = 7 points. If the leader plays 10 or more games in the month then there would be no penalty for that month.

  • administrators

    @paisiello Thank you for your feedback.

    A few additions:

    1. Tournament Elo is calculated in the same way as Automtach Elo is calculated. The only difference is that we pumped up the coefficient to 20 in normal Automatches it is 8. We did this to get more distinguishable results with only a few games played.
    2. the number of games played does only come into play if you set a certain min/max for games played. As long as a contestants have not reached the min amount of games needed they won´t be displayed on the leaderboard at all.

Log in to reply