Quote: Originally posted by dalter on 6/03/2010I think Rails was kidding.
And I agree with you leaving out CT performance -- it's just an arbitrary slice of a couple games.
Nice work!
thanks :)
i really don't think the CT is arbitrary, though. the regular season is only being measured by rpi... if i had the game history i could put together a ranking approximation that is better than the WIS rankings or rpi and then use that. but because I am only using rpi, i really don't think the quality of the regular season rating is that good. it is so easy to manipulate rpi. you can't really manipulate the CT... also, promises go out the window, and regular season experiments (which also hurt rpi) should be over. the CT is the best effort of teams, usually, in preparation for the NT.
it seems nobody agrees with me, so i will leave the CT out, but honestly i think it should be a small part, because it is less volatile than the NT and more meaningful than the regular season (not more important, but less influenced by factors outside the quality of the team, IMO). it was going to be the smallest component in my original design, but not negligible. i still feel a team who is in the top 5 by rpi and wins a tough CT but gets a bad rap early in the NT gets screwed in the rankings, and i think that would alleviate the issue a bit. also teams in tough conferences tend to get hurt slightly in rpi and significantly in seeding, so including the CT would offset those issues. but, i have given up on including the CT. i actually might have lost the rankings code when my computer blew up last week but i don't think so... keeping my fingers crossed.