depends on what level d3, d2, or d1

Not 2x.  Depends on level.  At D1 its more than 1.5, less that 2.  Less valuable in the lower divisions.

1 x in d1 for each full letter. imo

my impression has been 1.5x for each full letter in d1, 1.1x in d2 or d3.

thus in d1 a third of a letter grade is 1.144714

In DI my team (D-) went toe to toe with a B+ team, I came close to winning the battle but the difference was approx 3-1.

I think that  D- vs B+ at d1 would be 2.5 levels * 1.5-2 = a factor of 3.5-5 or so.

This is D1.  A+ vs. B+.  Should have included that originally.

Posted by pajamainc on 12/10/2010 10:57:00 AM (view original):
In DI my team (D-) went toe to toe with a B+ team, I came close to winning the battle but the difference was approx 3-1.

1.144714^8=2.948334
(8 steps from D- to B+ (D,D+,C-,C,C+,B-,B,B+))

Posted by jetwildcat on 12/10/2010 10:28:00 AM (view original):
my impression has been 1.5x for each full letter in d1, 1.1x in d2 or d3.

thus in d1 a third of a letter grade is 1.144714

are you convinced the scaling of prestige is multiplicative? i used to just assume that was the case, but about a year ago someone brought up the possibility that it was linear - i.e. if its 1.5 per prestige grade, its 2.0 for 2, not 2.25. i am just asking because at one point i had an impression very similar to yours and that aspect was not part of my picture. now i am not sure which way it is, from a programming standpoint it could easily go either way.

based on data, i really have no idea which way it goes. you would be hard pressed to figure it out in normal battles, which are usually close prestige grades (for example, 2/3rds of a grade with your numbers would be 1.31 or 1.33 depending if it was linear or not), and these differences are dwarfed by the unknown element of prestige (for example, is it a low a- or a high a-).

my reigning theory is that which ever the case is, many of us have it wrong, just as likely myself as anyone else. to me, that helps explain why so many of the best coaches cannot agree on the value of prestige. home to campus visit ratios are also very suspect, but those can be nailed down, so i figure people have to be looking at too complex situations and errors in value of prestige result in errors of value of HV:CV.

nope, not convinced. lol.

it does seem to fit in random situations, such as pajamainc's one above, so i keep using it. i also take comfort in the fact that i doubt many people have more effective approximation systems.

multiplicative is just the way that makes the most sense to me, so i *hope* HD works in a similar fashion

Posted by jetwildcat on 12/10/2010 12:12:00 PM (view original):
nope, not convinced. lol.

it does seem to fit in random situations, such as pajamainc's one above, so i keep using it. i also take comfort in the fact that i doubt many people have more effective approximation systems.

multiplicative is just the way that makes the most sense to me, so i *hope* HD works in a similar fashion

yeah, i am right there with you. also, the people who seem to support the linear conclusion seem to have higher prestige values than me - and its based off some extreme range prestige battles. but one would expect if a person's prestige value estimate was higher than it really was, on an extreme prestige difference battle, they would most likely conclude it was linear. or if their prestige estimate was low, they would conclude multiplicative. so im not sure who is wrong but it seems to make sense how those opinions would diverge.

What's the verdict on D3 prestige value in battles. Every letter is about 30% more?

less

20%? And is the effect multiplicative or additive. Don't battle much here. Would A+ v. C be north of 50% edge?