2
$\begingroup$

Let's suppose I have a database of 1000 competitors and 100 events. Each event has anything from 150 to 300 entrants.

The sport (e.g. rally car racing) is a simple against-the-clock event. Participants set off at regular intervals, and the race is decided by finishing times.

I assume there's already established algorithms for ranking competitors under these circumstances. Where should I be looking?

I've read up about Elo and TrueSkill, but it seems both are much more about 1-on-1 competition, or group-on-group, and not many-on-many. While they could maybe be adapted I'm not quite sure how to go about it, and I'm also unsure if they're suitable at all in this case.

I tried adapting Elo to change a competition of many people into hundreds of 1-on-1 matches, but you end up with the problem of a winner "running away" with a huge lead. Maybe with some tweaking of various factors it could work, but I feel I'm probably barking up the wrong tree?

Many thanks

  • 1
    If you ask for the best, you need to specify what you're trying to minimize or maximize. – 2017-01-21
  • 0
    Fair point. I'm not sure I fully know that myself yet. Would you be able to point me towards possibilities other than Elo/TrueSkill, so I can educate myself more on the subject? – 2017-01-23

0 Answers 0