Max Rating Reevaluation?

This type of thing might require too much bandwidth to be practical, but it's just a thought. Instead of max rating for the past 30 days to determine eligiblity for a tournament, what about using an AVERAGE rating for the past 30 days? We have all had the experience of accidentally beating someone much better than our usual best, and sometimes this can lead to a spike in one's rating. Or, sometimes it works out that several people resign or time themselves out by tournament overnight. I think an average would be more representative of a person's level.
Any thoughts?

a median gives a better measurement of the most likely value of a data set where you expect the odd outlier here and there.
Though you could also alpha trim average too.... DSP books where are you?

See also http://www.redhotpawn.com/board/showthread.php?id=8848

Originally posted by nemesio
Instead of max rating for the past 30 days to determine eligiblity for a tournament, what about using an AVERAGE rating for the past 30 days?

I think this would defeat the purpose. Say I'm a rating 2000 player who wants to smurf my way into a lower-rank tournament, where I can likely win. I start tanking games left and right, and my rating immediately plummets. My average rating over the past 30 days starts to fall immediately, too, albeit not as quickly. However, my max rating stays at 2000 for the full 30 days, after which it starts to drop.

In other words, to purposefully lower my rating (for tournament eligibility), I would need to waste an entire month of losing on purpose before seeing any reduction in my eligibility rating. That's quite a bit of work, and hopefully enough to offset any potential tournament-winning pleasure.

EDIT: Perhaps a better example would be the case where I only have to drop a few rating points in order to be eligible for a tournament. I may be able to reduce my monthly average rating in time for the tournament, but I won't be able to reduce my my monthly max rating by losing on purpose.

Originally posted by nemesio
... We have all had the experience of [b]accidentally beating someone much better than our usual best, ....?????[/b]

OOOOO .... I take full credit for all my wins !

Originally posted by godzillion
Perhaps a better example would be the case where I only have to drop a few rating points in order to be eligible for a tournament. I may be able to reduce my monthly average rating in time for the tournament, but I won't be able to reduce my my monthly max rating by losing on purpose.

I suppose this is precisely my point. Let's say I average a 1350 (which, until recently, is about where I am). Let's say in an opened tourney I (accidently) beat some with an 1800 rating, or s/he times out or something. My monthly rating sky rockets. Then I lose a whole bunch of games in this same tourney, my average rating goes back to what it was (or worse), but my peak rating, certainly an inaccurate measure of my skill remains the same for a month. The average monthly rating can only change so much, particularly if you have many games, and would be a more accurate reflection of how a person is playing of late, which is how they should be judged with respect to a tournament.

A person who tanks games probably isn't that good anyway, one has to recognize. And a person who would be able to tank a few because they were close to meeting a rating qualification probably can't win that tournament with ease, anyway (i.e., if I have a 1405 rating and I tank a few to get into at 1400 tournament, it doesn't secure me a win by any means). It doesn't seem to be a big risk factor to me.