Sagarin changing his rankings on the fly

[occupant]

So...now Jeff is introducing a new wrinkle into his ranking/weighting.

DIMIN_CURVE...can't say I've ever seen this in his rankings before.

I'd like to know first of all, why he is making wholesale adjustments to his system. Is he trying to compensate for observed inaccuracies or is he attempting to advertise his "system" as a preferable reference tool available for use by the new CFB playoff committee?

Secondly, is this all allowed by the BCS? Do the computer wranglers have carte blanche to modify/skew/recalculate their formulas as they see fit whenever they want?

Thirdly, how can any computer-based ranking be taken seriously when they have NIU currently ranked 3rd in the rankings they submit to the BCS?

 

[yeodonie]

Yeah changing the rules in the middle is horse shit unless you find a bug. It's maddening

 

[Red_Alert]

For all practical purposes NIU beat Iowa @ Iowa. About the equivalent to what Ohio St beat Iowa in Columbus.

NIU also stomped Purdue of which Ohio St should do as well.

If NIU is over-rated, one could say Ohio St is as well.

 

[occupant]

Yeah changing the rules in the middle is horse shit unless you find a bug. It's maddening

I can see how he would want to improve his system...but bugs by their definition are errata. They shouldn't have anything to do with the construct of the formula, only the failure of a function. Every team would be rated by the same template in place, even if it wasn't what you had designed.

This stinks to me.

I look forward to hearing some explanation by the number cruncher himself as to why these changes are being made at this point of the season.

 

[cane_man]

The bcs sticks with original format?

 

[4down20]

I'm guessing DIMIN_CURVE = diminishing curve, something I've considered myself but with the limited amount of games it didn't seem to be a very good idea. Basically, what it means is that the most recent games are weighted higher than the games to start the season. But as I said, when you've only got 12 games in an entire season, 13 with conference championships(and USC this year), it seems to me that the 12 games themselves are pretty much needed completely.

As for changing his system, I find it hard to believe he is changing the BCS formula part on the fly. Adding other rankings I have no problem with it. As for NIU being so high, I don't think he uses predictive measures in his BCS rankings, which means crazy things are going to happen until there is more data. I'd hate to see what my rankings would look like right now if I didn't predict future games.