Method: I took the ELO of each team's top 6 players. Why 6? Five would probably have done about as well, but I think 6 represents better what a coach will think of as his choices for a given match. Seven might be even better, but by the time we hit even five, let alone six or seven, most teams are down to bad players. So six represents players with a realistic chance of winning if played wisely.
I then took the average ELO for all players and average ELO for each 1-6 spot, and found each player's difference from overall average and average for his spot. I then got three team statistics: average player ELO, average player ELO difference from total average, and the value of the team's ELO spread. Results:
Average ELO, all players: 2133
Average ELO, Ace: 2245
Average ELO, #2: 2170 (-75)
Average ELO, #3: 2143 (-27)
Average ELO, #4: 2108 (-35)
Average ELO, #5: 2082 (-26)
Average ELO, #6: 2054 (-28)
The gap between ace and #2 is, obviously, far and away the biggest on most teams. Some of this is a function of Flash, Jaedong, and Bisu, and some of it is true for everybody.
The resulting ranking:
8. ACE -66 var. 110. One player above average, five below, all players below role average by at least 25 points.
7. STX -20 var. 135. One player above average, five below, one player above role average (#6 hyvaa +13).
6. WJN -8 var. 255. Ace-#3 players above average and role average, #4-6 players below average and role average.
5. SAM +3 var. 123. Two players above average, four below, Ace-#3 below role average by 14- but #4-6 above by 15+.
4. CJ +16 var. 208. Four players above average, two below, all players above role average except #6 sKyHigh -28.
3. K8 +19 var. 196. Four players above average, two below, all players above role average.
2. KT +21 var. 221. Three players above average, three below, all players above role average.
1. SKT +36 var. 282. Three players above average, three below, all players above role average except #4 Hyuk -2.
My belief that STX really is bad has been reinforced, although they're probably not as bad as ACE if I'm being honest here. KT by ELO is much better than I thought, mainly because I always forget how good Mind really is. Woongjin and Samsung have really weird breakdowns: Stars because their 1-3 are so good and 4-6 are so bad (note the huge variance), KHAN because the entire lineup is basically mediocre (note the tiny variance).
General take-aways: ELO variance is basically useless without context: large variance (@200) basically means you have an S-class ace (Bisu, Flash, Jaedong, Hydra, ZerO). Smaller variance (@100) means different things: for ACE it means everybody is bad but not remarkably so. For KHAN, it means everybody is pretty good but not really S-class. And for STX, it indicates a totally mediocre team (I'm being polite here: Bogus is better than I remembered, but the rest of the team is completely eh).
If I get really ambitious tomorrow I'll work out deviations and stuff and see if that gives any more information but right now it's 1 in the morning, lol.