Warrior General Games

step 1. each team is given a "raw_rating". this is calculated by adding your win% (win/total_games) and 3*points% (points_for / (points_for+points_against))

Edit(9/26): put a cap on the points_for you can receive from any 1 game at 50+points_against. this dampens some of the nasty effects of 255-0 blowouts.

for example, a 3-5 team who scored 144 and gave up 176 in those 8 game would get a raw of (0.375+0.45*3 or 1.725)
lowest raw: a gutted team who never scored a point would get raw_rating of 0
highest raw: an undefeated team who has never given up a point would get a raw of 4

step 2. for each team, start with 100 rating and go through their schedule. if they won, they get their opponent's raw_rating added in addition to another fraction of that raw_rating based on the score diff. the fraction is based on 50 max, so that if you win by 50 or more, you will get the full raw.
for example, if a team played an opponent with a 2.5 raw and beat them by 25, they would get 2.5 + (25/50)(2.5) added to their rating. worst case is if you win against some gutted team with 0 raw, then you actually don't get any credit for the win no matter how badly you beat them.

when you lose, the opposite calculation is used... you lose more rating for losing against low raw teams + another fraction for the score diff

these changes are calculated for each game on your schedule, but then a 5% backoff factor is put in. this means that your most recent game's changes is fully considered while the game from 1 week before only gets a 95% weight, and the game from 2 weeks before only gets a 90% weight, etc

and that's it, basically these rankings take into account strength of schedule and rewards teams that have had recent success

as always, i'm open to suggestions if you feel certain teams aren't ordered appropriately. i can easily provide the result log of all the calculations if you want to see how things turned out the way they did.

Originally posted by ggezhb
step 1. each team is given a "raw_rating". this is calculated by adding your win% (win/total_games) and 3*points% (points_for / (points_for+points_against))
for example, a 3-5 team who scored 144 and gave up 176 in those 8 game would get a raw of (0.375+0.45*3 or 1.725)
lowest raw: a gutted team who never scored a point would get raw_rating of 0
highest raw: an undefeated team who has never given up a point would get a raw of 4

step 2. for each team, start with 100 rating and go through their schedule. if they won, they get their opponent's raw_rating added in addition to another fraction of that raw_rating based on the score diff. the fraction is based on 50 max, so that if you win by 50 or more, you will get the full raw.
for example, if a team played an opponent with a 2.5 raw and beat them by 25, they would get 2.5 + (25/50)(2.5) added to their rating. worst case is if you win against some gutted team with 0 raw, then you actually don't get any credit for the win no matter how badly you beat them.

when you lose, the opposite calculation is used... you lose more rating for losing against low raw teams + another fraction for the score diff

these changes are calculated for each game on your schedule, but then a 5% backoff factor is put in. this means that your most recent game's changes is fully considered while the game from 1 week before only gets a 95% weight, and the game from 2 weeks before only gets a 90% weight, etc

and that's it, basically these rankings take into account strength of schedule and rewards teams that have had recent success

as always, i'm open to suggestions if you feel certain teams aren't ordered appropriately. i can easily provide the result log of all the calculations if you want to see how things turned out the way they did.

Dork!!! j/k

considering that it's a 16 game schedule, would it be better to use a lower take off percent? maybe 5%, 4%, 4%, 3%, 3%, 3%, 2%, 2%, 2%, 2%, then 1% for further back. I guess i don't like seeing week 1 worth 20% at the end of the season. in the above scenario, it would still count for 65%. just a suggestion since you can't control when you play the best teams and the worst teams.

Originally posted by jeremy042376
considering that it's a 16 game schedule, would it be better to use a lower take off percent? maybe 5%, 4%, 4%, 3%, 3%, 3%, 2%, 2%, 2%, 2%, then 1% for further back. I guess i don't like seeing week 1 worth 20% at the end of the season. in the above scenario, it would still count for 65%. just a suggestion since you can't control when you play the best teams and the worst teams.

i think something like that would be better in a more mature league. the backoff was purposely set that high since many teams are completely transformed over the course of a season, and a higher % would give a more dynamic ranking.

i guess my personal view is that when i look at power ranking at the end of the season, what happened in game 1 doesn't factor in much (it's actually 25%). Anyone else want to weigh in here?

you're absolutely right in that scheduling can totally mess things up.. if you play 5 terrible teams in a row at the end of the season, your ranking is gonna suffer even if you win every game 200-0. none of this would be problematic if there wasn't the ultra bad teams though =(

i like the backoff, and view the power rankings mainly as a
"who's hot" kind of list. it'd be interesting to see what
the progressive backoff would come out to, but i agree that
game1 is meaningless for many if not most teams by the time
game16 is approaching.

Especially this season when so many of us struggled with the new Advanced AI options.

Originally posted by jeremy042376
considering that it's a 16 game schedule, would it be better to use a lower take off percent? maybe 5%, 4%, 4%, 3%, 3%, 3%, 2%, 2%, 2%, 2%, then 1% for further back. I guess i don't like seeing week 1 worth 20% at the end of the season. in the above scenario, it would still count for 65%. just a suggestion since you can't control when you play the best teams and the worst teams.

that's a bad way fo looking at it.

Week 1 is worth 20% of what it was worth in week 1.

It's also been averaged out so that it's only a small fraction of the ranking