Ranking Idea: Standard Deviations from the Mean

[jipkin]

In the hall of fame, there are number of interesting ranking categories, but none really truly reflect skill. Number of points, number of words, and number of games just reflect how much you have played. Average score would be meaningful except that players can pick and choose which boards they play because of the preview of the number of words etc. I know I only play on the big ones (they're the most fun). High score is obviously pretty meaningless since there's so much luck there.

I think the easiest value to calculate for a player's skill would be to rank that players average number of standard deviations away from the mean. For example, really good players will probably always be three to four standard deviations above the mean for a given board no matter how big the board is, while really bad players will always be three to four below.

You'd have to recalculate everyone's standard deviation everytime a board was played, of course. Not sure how technically demanding this would be... I can't imagine it would be very hard if you had a leaderboard stored for each board.

Thoughts?

[shringeld]

How about assessing a board's degree of difficulty, and working out a form of handicapping that would more or less equilibrate the scores on different boards? So that a very high score on a board with under 200 words would, via handicapping, be equivalent to a very high score on a richer board? Same effect as calculating standard deviations, but less computational effort.

[Chelonia]

You'd have to recalculate everyone's standard deviation everytime a board was played, of course. Not sure how technically demanding this would be... I can't imagine it would be very hard if you had a leaderboard stored for each board.

Thoughts?

It seems to me that something like this is already done, in that it places the score on the curve. I don't know if the cutoffs on that curve (high, very high, etc) are technically standard deviations, but since the program can place it on that curve, it seems to me that it wouldn't be that hard to create a score from that which could be compared to others.

Said she who has no idea how such things are done!