a proposal to rectify skewed standings
12 Mar '16 13:55I would like to moot an idea for changing the way clan standings are calculated. The sum total of wins per clan is suboptimal for (at least) two reasons: a) it favors large clans which play short timings and therefore rapidly accumulate wins, compared to smaller clans with longer timings, and b) it is open to abuse and manipulation (e.g., collusion).
I propose (for discussion) the following formula instead: the net average rating loss or gain of a clan is to be the measure of the clan's standing.
Example: player X from clan X’ has a 100 point increase in his individual rating, while player Y (same clan) has a 50 point decrease. This makes a 50-point net increase for that clan. Player Z from clan Z’ drops 150 points in his personal rating, while player C from clan Z’ increases 450 points; the net increase for clan Z’ is 300 points. Clan Z’ is therefore ahead of clan X’ on net rating points. Note that it makes no difference whether the individuals of clan X' or Z' are rated higher or lower, only the net change is relevant.
Obvious objection: if clan X’ has 100 players who each increase by 10 points individually, that would be a net increase of 1000 points for the whole clan, whereas if clan Z’ has five players who each increase 100 points, that would be a net increase of only 500 points, which is prejudicial to the smaller clan.
Therefore, not the sum of the individual increases should be reckoned, but the average. This would rectify the discrepancy between large and small clans, irrespective of time limits or number of games played.
This formula would have the advantage of encouraging challenges slightly above one's own clan rating, since wins (assuming one does win) would accentuate one's rating increase, whereas challenging lower-rated clans and beating them would not gain as much.
Caveats:
a) Players should have separate ratings for individual games and clan games, as Suzianne has proposed, to stop skewing clan pairings by throwing non-clan games (sandbagging). Only clan-ratings would be reckoned for clan pairings, as well as for the final tally which clan posts the greatest average net increase in ratings. (Separate ratings have already been implemented for club members, so this would be merely an extension of current RHP policy.)
b) A maximum ratings differential should be observed for clan pairings (the current value of 200 points seems about right).
c) An arbitrary minimum of completed clan challenges would be sensible (just as a minimum number of games is required to calculate an individual's rating).
d) An arbitrary maximum number of clan-to-same-clan challenges would be sensible--a fairly low number, two or three challenges--to eliminate collusion between clans. Exceptions could be made to allow very large clans to play the same other-very-large clan more than two or three times, provided different players were involved in each challenge. Alternatively, clans which like to play each other would be allowed to play any number of challenges, but only up to the agreed-upon maximum would count towards the clan standings.
Request for comments.
I propose (for discussion) the following formula instead: the net average rating loss or gain of a clan is to be the measure of the clan's standing.
Example: player X from clan X’ has a 100 point increase in his individual rating, while player Y (same clan) has a 50 point decrease. This makes a 50-point net increase for that clan. Player Z from clan Z’ drops 150 points in his personal rating, while player C from clan Z’ increases 450 points; the net increase for clan Z’ is 300 points. Clan Z’ is therefore ahead of clan X’ on net rating points. Note that it makes no difference whether the individuals of clan X' or Z' are rated higher or lower, only the net change is relevant.
Obvious objection: if clan X’ has 100 players who each increase by 10 points individually, that would be a net increase of 1000 points for the whole clan, whereas if clan Z’ has five players who each increase 100 points, that would be a net increase of only 500 points, which is prejudicial to the smaller clan.
Therefore, not the sum of the individual increases should be reckoned, but the average. This would rectify the discrepancy between large and small clans, irrespective of time limits or number of games played.
This formula would have the advantage of encouraging challenges slightly above one's own clan rating, since wins (assuming one does win) would accentuate one's rating increase, whereas challenging lower-rated clans and beating them would not gain as much.
Caveats:
a) Players should have separate ratings for individual games and clan games, as Suzianne has proposed, to stop skewing clan pairings by throwing non-clan games (sandbagging). Only clan-ratings would be reckoned for clan pairings, as well as for the final tally which clan posts the greatest average net increase in ratings. (Separate ratings have already been implemented for club members, so this would be merely an extension of current RHP policy.)
b) A maximum ratings differential should be observed for clan pairings (the current value of 200 points seems about right).
c) An arbitrary minimum of completed clan challenges would be sensible (just as a minimum number of games is required to calculate an individual's rating).
d) An arbitrary maximum number of clan-to-same-clan challenges would be sensible--a fairly low number, two or three challenges--to eliminate collusion between clans. Exceptions could be made to allow very large clans to play the same other-very-large clan more than two or three times, provided different players were involved in each challenge. Alternatively, clans which like to play each other would be allowed to play any number of challenges, but only up to the agreed-upon maximum would count towards the clan standings.
Request for comments.
Back to Top