Posted by isack24 on 10/1/2012 3:38:00 PM (view original):
But why would he have dropped?

If 50% of the weight is the most recent season, and his most recent season is undeniably better than the season before, and there couldn't be anything that "fell off" because he has gotten better every season but 1, then it makes no sense that he would have gone down unless the other 50% prestige weighs recent seasons more heavily such that the NC game season was weighed more heavily in the second-most-recent season slot than the third.  But that would seem to imply that the the second-most-recent season slot is a decent % of the "other 50%."

It has to be a weighted average, that's the only thing that makes sense.  So maybe it would be something like below, and the title game appearance moving back a season could have a decently large impact.

current season, 50%
1st prior, 25%
2nd prior, 12.5%
3rd prior, 6.25%
4th prior, 3.125%
etc.

And I agree with Al that the W-L records the last 2 seasons are clearly a huge factor.  A team that went 31-29 in the two most recent seasons probably should be happy they're A- prestige.

Posted by alblack56 on 10/1/2012 4:05:00 PM (view original):

Posted by isack24 on 10/1/2012 3:38:00 PM (view original):
But why would he have dropped?

If 50% of the weight is the most recent season, and his most recent season is undeniably better than the season before, and there couldn't be anything that "fell off" because he has gotten better every season but 1, then it makes no sense that he would have gone down unless the other 50% prestige weighs recent seasons more heavily such that the NC game season was weighed more heavily in the second-most-recent season slot than the third.  But that would seem to imply that the the second-most-recent season slot is a decent % of the "other 50%."

It has to be his record.  14-15, followed by 17-14.  Seble wants the prestigous teams to win a lot of games, even if it trumps post-season results.

I remember a team that was 29-1, SOS 320,  a 1st round loser, and made a BIG jump in prestige. This school had won ONE NT game in it's entire history

Agreed, al, but if the most recent season is weighed so heavily (50%), and then the next four or so are weighed in some way to give the other (50%) how would his prestige have gone down when S1 > S2 in every way, including wins?

EDIT: Unless it's a weighted average, as discussed above.

Posted by isack24 on 10/1/2012 5:09:00 PM (view original):

Posted by alblack56 on 10/1/2012 4:05:00 PM (view original):

Posted by isack24 on 10/1/2012 3:38:00 PM (view original):
But why would he have dropped?

If 50% of the weight is the most recent season, and his most recent season is undeniably better than the season before, and there couldn't be anything that "fell off" because he has gotten better every season but 1, then it makes no sense that he would have gone down unless the other 50% prestige weighs recent seasons more heavily such that the NC game season was weighed more heavily in the second-most-recent season slot than the third.  But that would seem to imply that the the second-most-recent season slot is a decent % of the "other 50%."

It has to be his record.  14-15, followed by 17-14.  Seble wants the prestigous teams to win a lot of games, even if it trumps post-season results.

I remember a team that was 29-1, SOS 320,  a 1st round loser, and made a BIG jump in prestige. This school had won ONE NT game in it's entire history

Agreed, al, but if the most recent season is weighed so heavily (50%), and then the next four or so are weighed in some way to give the other (50%) how would his prestige have gone down when S1 > S2 in every way, including wins?

EDIT: Unless it's a weighted average, as discussed above.

Season 51 hurt him (dropped him from A+ to A) but season 50 was still propping him up.  After season 52, albeit stronger than season 51, his most recent 2 seasons are hurting him.  A- is about the best he should expect given the way the engine works.

However, I agree with Al that this is asinine.  Take a team who's been in the final four 7/8 seasons in a row.  If they play in a tough conference and go 19-and whatever and only make a sweet 16 two seasons in a row, they drop?  Crazy.

Sure, I don't think the A- is the problem.  I think the A+ was high in the first place.  My issue isn't the final grade, but the fickle nature of D2/D3 prestige.

Anyone else see the irony here?

Posted by dahsdebater on 10/1/2012 6:11:00 PM (view original):
Seriously, you guys still aren't getting this?  Trying to argue what season is doing what, talk about a weighted average - NONE OF THOSE IDEAS APPLY IN D2/D3.  It's a floating value.  Right now, he's got some prestige level.  At the end of next season his prestige will be adjusted based on the current prestige level and the prestige contribution of the new season, at roughly a 50/50 level.  IOW, at the end of next season his prestige will basically be the midpoint of his current prestige and the prestige based on next season alone.

In almost all cases it's not a huge difference, but that doesn't make it not worth understanding.  Think of it as similarly to how you might perform a string of mathematical operations.  D1 works like aA+bB+cC+dD, where the capital letters are the prestige values of the previous 4 seasons (now I guess it's more than 4, but the idea is the same) and the lowercase letters are coefficients valuing more recent seasons more highly.  At the end of the next season it will perform the same calculation, but A will become B, B becomes C, etc, and the most recent season appears as A.

In D3 the calculation looks like (A + B)/2 := B, where A is the prestige of the most recent season and B is the existing prestige value.  You average those numbers and replace B with the new value.  B becomes your prestige for the next season and at the end of the following season the same calculation is performed.  Effectively this makes every season in program history impact the prestige, but each season further you go back the impact is reduced by ~2 (it may be the most most recent season is even more than 50%, but it makes a good estimate).  Thus, the most recent season is 50%, next 25%, then 12.5, 6.25, 3.125, etc.  You never see a season "drop off," but every season but the current season loses about half of it's weight on the prestige rating every year.

I don't really see how what you're saying is different from the weighted average example I posted.  Heck you even used the exact same weights!

It sounds like just semantics to say it's (A+B)/2 versus a full-on weighted average.  B didn't just come from nowhere, it's (likely) a weighted average of all the previous seasons.

@ dahs: 

how do you know this? was it in a dev chat? 

and have we seen this play out: ie, does prestige flucuate like crazy in seasons 1-10 of a world, and then settle in? 

this seems wrongs and a little nuts to me. It seems much more likely that it's a variation of the D2/D1 system, with more value placed on the most recent year... and D2/D1 don't have the same ratios as each other, IIRC. 

Posted by emy1013 on 10/1/2012 6:25:00 PM (view original):
Anyone else see the irony here?

didn't he use the exact same values as Al?

Same as spasticity.

Posted by wronoj on 10/1/2012 6:42:00 PM (view original):
@ dahs: 

how do you know this? was it in a dev chat? 

and have we seen this play out: ie, does prestige flucuate like crazy in seasons 1-10 of a world, and then settle in? 

this seems wrongs and a little nuts to me. It seems much more likely that it's a variation of the D2/D1 system, with more value placed on the most recent year... and D2/D1 don't have the same ratios as each other, IIRC. 

I honestly don't remember how this came to light, but it was more universally known a few years ago.  We have never been told that it was changed, so I would assume it's still true.

dahs .. if your model is correct, then this is ALSO what you are saying

1.  (A + B) / 2 is A/2 + B/2
2.  ((A + B)/2) + C)/2 ... this is the same as A/4 + B/4 + C/2, it continues as below
3.  A/8 + B/8 + C/4 + D/2
4.  A/16 + B/16 + C/8 + D/4 + E/2
5.  A/32 + B/32 + C/16 + D/8 + E/4 + F/2

So, if you had a numbers, and they were

96, 64, 32, 96, 64, 32, 96 ... the running average would be as follows:

(96 + 64)/2 = 80

(80 + 32)/2 = 56

(56 +96)/2 =  76

(76 + 64)/2 = 70

(70 + 32)/2 = 51

(51 + 96)/2 = 73.5


Also ...

96/2 + 64/2 = 80

96/4 + 64/4 + 32/2 = 56

96/8 + 64/8 + 32/4 + 96/2 = 76

96/16 + 64/16 + 32/8 + 96/4 +  64/2 = 70

96/32 + 64/32 + 32/16 + 96/8 + 64/4 + 32/2 = 51

96/64 + 64/64 + 32/32 + 96/16 + 64/8 + 32/4 + 96/2 = 73.5

=======

At some point, the first 2 numbers no longer matter ... lets say that the point is A/512 and B/512 is that point because if both were 100, it would be only 0.390625 (or 99.6 would be the value)

This would mean that this is the "long term equation":

A/256 + B/128 + C/64 + D/32 + E/16 + F/8 + G/4 + H/2 = Prestige ... or:

8 years ago is worth (100/256) or 0.390625%
7 years ago is worth (100/128) or 0.78125%
6 years ago is worth (100/64) or  1.5625%
5 years ago is worth (100/32) or  3.125%
4 years ago is worth (100/16) or  6.25%
3 years ago is worth (100/8) or 12.5%
2 years ago is worth (100/4) or 25%
1 year ago is worth (100/2) or 50%

The point is, 50% is way too much for the most current year.  In real life, if you go to the sweet 16, elite 8, or final four every year and then one year, oops, you get upset by the 15 seed in the first round, your prestige doesn't drop precipitously.  Silly.

How come no one complains when they suck for 5 years and then have one good season and jump 2 letter grades?  I suppose the "earned" that?

Posted by hughesjr on 10/1/2012 11:55:00 PM (view original):
How come no one complains when they suck for 5 years and then have one good season and jump 2 letter grades?  I suppose the "earned" that?

Assumption fail.