i literally have no idea lol. ok let me check for formulas and stuff, but ill explain first to provide context. basically i just took the team dynasty rankings as the starting point. those are fairly focused on NT success and only involve RPI and NT success. i've considered trying to use the NT seeding (from the bracket) and conf/CT titles (not to assign points to those - but in strong confs, to give teams who are really good but lose early in the NT, a little boost). curious if anyone has an opinion on using the conf/CT titles in that way, bit of a fairness issue.
anyway, the individual teams rely less on deep NT runs than i personally would do (i literally value nothing else, regular season is for player development IMO, but this also would make it more luck driven and would kinda screw the folks who care about the regular season). but its still pretty NT success driven. there is also a world strength component that you can think of loosely as population but it actually looks at the last 10 seasons for each program and comes up with a score. this way, a bunch of freeHD users popping into a d3 world who stay for 1 season, count for a little but not a ton more than a 10-sim run. 10 seasons with the same human counts for the most. once the population total or whatever is put together, i take the square root to reduce the effect to what i consider more reasonable levels, and particularly to decrease the gap between very full and very empty worlds - but it really worked, imo, quite well for like, 150 vs 200 person worlds, or 160 vs 180 and stuff. i never intended to compare d3 worlds with 50 people to old school d3 worlds with 200 people. still - none of this is a problem for anything but the all time team/conf rankings. the current world rankings, current cross-world rankings, and current conf dynasty rankings should all be fine.
finally, the individual team scores, which you can see on the dynasty rankings, are then put together in not-quite a sum fashion. it basically softens the impact of the higher scores somehow, or at least that was my intent, so a top-heavy conf doesn't swamp a conf with a lot of depth. i felt like, in individual teams, it makes sense a NT title #1 rpi team gets a lot more points than a s16 team - but in a conference, it feels like this premium on very deep runs is less appropriate than considering depth and such.
alright - enough rambling - here's some numbers:
RPI score - final rpi is mapped to points, in linear fashion. example - the 125 rpi team gets 0.5 points. the 5 rpi team gets 8 points.
rpi 1 = 10
rpi 2 = 9
rpi 5 = 8
rpi 10 = 7
rpi 17 = 6
rpi 26 = 5
rpi 37 = 4
rpi 50 = 3
rpi 65 = 2
rpi 100 = 1
rpi 150 = 0
heres the post season outcomes mapped to points:
nothing = 0
pi1 = 1
pi2 = 1.5
pi3 = 2.25
pi4 = 3
pi5 = 3.75
pi6 = 4.5
nt1 = 3
nt2 = 4.5
nt3 = 6
nt4 = 9
nt5 = 12
nt6 = 17
nt7 = 24 (this means championship)
(note - part of the reason PI title > NT1 here is because in the olden days, the seeding was so bad, it was pretty easy in top conferences to be like the 30-45th best team and have the 30-45th best season but still miss the NT - this is probably a little out of whack today)
so, i was going to next explain world strength, but im thinking for the complexity, its not worth it. i'd love to get some feedback but i doubt anyone cares - if anyone does care, let me know!
conference dynasty calculation - interesting, i think i bailed on a complex scheme. its super simple. its the sum of the 12 teams scores but with each team score raised to the power of 0.8. this basically transforms the numbers to a more even scale so that a few solid teams can add up to one excellent one. some example team scores -> conf score contributions:
40 -> 19.1
20 -> 11
10 -> 6.3
5 -> 3.6
so you can kinda see... on the left, the numbers are double each row higher, on the right, its less than that. not hugely less but about 1.75 higher instead of 2x higher. so if you had like a conf with a 40 team and four 5 point teams and all horrible sims, normally, you'd need a conf with 12 5 point teams to equal it. in this model, you'd need a conf with 9.3 5 point teams to equal it. so its like... a not huge effect, but its meaningful i think, in terms of moving the needle towards depth as compared to the raw dynasty rankings which are quite dependent on deep NT runs. perhaps a factor of .7 would be better?
5/21/2020 6:58 PM (edited)