In many Saimoes I've seen, the match-up are either set in place sometimes with seedings (e.g.winner of Bracket A face the winner of Bracket B, winner of vs 8th vs 9th face the winner of 1st vs 16th) or randomized each time a round is finished. On one hand with the seedings, it will make sure that the stronger ones will face the weaker ones most of the time, but the next match up are predictable since the winner will definite face the winner from one bracket, not another. On the other hand with random grouping, people will never know who the character will face until the round is over and match up is released, this is exciting but it also creates unbalanced grouping and strong characters can go out on the early rounds.
I wondered if there's a way between the two, and I realized that there's.
First one is to reseed the characters each round, with VF, VD, VF% or whatever equations that can sorts characters into rankings (I think I've seen this partially-implanted in a Saimoe Tournament before, but I can't remember which one).
Although the problem is trying to find the best way of seeding, I can't think of one that is suitable. Something too simple (VF, VD, VA, VF%) can be too easily manipulated and affect by other factors (some match days have more voters than others, intra-series match-up ect.), something too difficult will make the tournament more complicated to understand. There is also a consideration of whether to include stats for just one round or all the results of the year, should the nomination take part, if so then how to deal with the auto-nominated characters.
It' probably better by an example.
Example:
In a ISML 2008~2011 Final League, the match up goes like this:
1 vs 16
2 vs 15
3 vs 14
4 vs 13
5 vs 12
6 vs 11
7 vs 10
8 vs 9
Supposed that 12th seed upsetted 5th seed, no other upsets, which leaves the remaining characters to:
1
2
3
4
6
7
8
12
Under normal match up, 12th seed will face the 4rd seed, which is who the 5th seed would have meet.
1 vs 8
2 vs 7
3 vs 6
4 vs 12
However in this new case, the match up goes:
1 vs 12
2 vs 8
3 vs 7
4 vs 6
Simply put, the previous match-up is disregarded, and the highest seed face the lowest seed, 2nd-highest seed face the 2nd-lowest seed and so on.
Next match up is unknown (sort of), but the match up is not random. It's a little better with randomized grouping but it's still giving more power to luck than stats, a bad seed in the early round then the candidate is doomed to go through 1~4 seed all the way if she is to gain the Tiara.
At this point I'm guessing the elimination period will be something like Best/Guy Contest 2, except with only 64 characters. I think that's simple to understand and I do like it, I wrote this down since I want to see some unpredictability in match-ups without completely relying on random chances.