3
$\begingroup$

My coworkers and I play board games at lunch. We've taken it so seriously that I've written an application to calculate a "player score" based on how you place in each game played.

Here's how it currently works:

(Positions - Placement) / (Positions - 1) = Score

where Placement is 1st == 1, 2nd == 2, 3rd == 3, etc... and Positions is how many people were playing the game.

4 Man Game Scoring Example

Position  Calculation  Score
1st       (4-1)/(4-1)  1.000
2nd       (4-2)/(4-1)  0.667
3rd       (4-3)/(4-1)  0.333
4th       (4-4)/(4-1)  0.000

Now that that's all figured out, you take your average for all the games played and away you go.

While there may be other issues, the primary concern is that we've had to put a "Minimum Games" requirement when going for the season trophy because if you win a single game, you stand with a solid 1.000 score and if you lose your only game, you sit with 0.000.

We've considered various ways of creating "bonuses" for playing numerous games but nailing that magic number is a hard thing to do. Doing a traditional ELO doesn't appear to be sufficient because it's not 1v1.

If there's anyone with a suggestion as to how to do it better, I'm all ears. And my apologies if this is the wrong forum. The board & card games one didn't feel totally right.

Thanks in advance - Hugh

$\endgroup$
4
  • 1
    $\begingroup$ Elo doesn't require 1v1. Update each player as if he played against the average ratings of the opponents. $\endgroup$
    – user58697
    Commented Aug 5, 2014 at 23:16
  • $\begingroup$ Hmm... that's an interesting idea. I suppose you'd have to treat each game almost as if it were multiple. ie - you played a four player game and took second, you lost one and won two. Thanks! $\endgroup$
    – hughball
    Commented Aug 6, 2014 at 21:34
  • $\begingroup$ In fact, this is how Elo used to be recalculated after a large chess tournament before computers. $\endgroup$
    – user58697
    Commented Aug 6, 2014 at 21:45
  • $\begingroup$ Having a Elo rating set up is the perfect excuse to set up a data crunching server. I shall attempt! Thanks for the suggestion. $\endgroup$
    – hughball
    Commented Aug 7, 2014 at 21:02

1 Answer 1

0
$\begingroup$

One "bonus" type you might want to consider is having the score being based on their highest scores, with decaying after the highest.

So, for example (with a $5\%$ decay rate), let's say you have a player that played $5$ games with scores of $1,1,\frac{2}{3},\frac{1}{3},0$. Then, you could calculate that player's total ranked score as $1+0.95\cdot1+(0.95)^2\cdot\frac{2}{3}+(0.95)^3\cdot\frac{1}{3}+(0.95)^4\cdot0\approx2.84$ points. This scheme rewards playing more games, while still being somewhat fair to players with a better record. You can then adjust the decay rate to make the system suit your needs.

$\endgroup$
2
  • $\begingroup$ That's a unique idea. I'll give it a shot and see what the scores look like and post back. Thanks! $\endgroup$
    – hughball
    Commented Aug 5, 2014 at 22:10
  • $\begingroup$ I queried my data and found that while in the first season, most of the players stayed in the same position in the standing, one player (who played a lot of games) jumped up a position because of the summation of all the scores and the decay keeps the scores from totally ballooning. $\endgroup$
    – hughball
    Commented Aug 6, 2014 at 20:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .