Well, I'm having trouble sleeping so tried to answer my own question.

Taking Akai's "final" data sets for the original VF5...

*** Version C
kage 35%
lau 31%
eileen 29%
shun 26%
sarah 26%
jacky 25%
jeffry 24%
akira 23%
vanessa 22%
brad 21%
wolf 21%
aoi 21%
lion 21%
lei 21%
blaze 21%
pai 17%
goh 15%

*** Version B
shun 34%
eileen 33%
kage 30%
brad 27%
akira 26%
blaze 25%
pai 24%
lau 23%
jacky 23%
aoi 23%
vanessa 22%
lion 22%
lei 21%
sarah 21%
goh 18%
wolf 17%
jeffry 14%

So the answer is actually NO, VF5R Version B isn't the most compressed VF of all, because the final Version B and Version C pre-Revolution were actually just as balanced.

Wow, didn't realize.

Basically, the tiers are +/- 10% from the average, which is of course by definition 25%.

Well VF5R has two more characters. So a similar spread across 19 characters is more balanced than the same divide across 17 isn't it?

Considering on the tiers, I think two sets are pretty enough actually, and there is no need to sort into three groups.

As long as we pick out some S rank, the rest are belong to one group even they still have some tiny differences.

ice-9, the comparisons are cool and all, but imo the samples will always be biased, when trying to look at them like that for balance.

I think it's more on the players involved, and the current metagame, self imposed codes, and character loyalty, that produce numbers like that.

Of course insert char would be at the bottom, if the amount of players using said char were small in number, the players were weak in comparison, or both combined.

Empnova, the way ice-9 was calculating the percentages is independent of the number of characters available in the game.

When new versions of the game come out, people assume that the changes made will make the game more balanced. But it might not be the case. For the most part, I think it just change what character might do better than other characters (based on comparing the data from the different versions of the game).

Of course, there will always be biases in the data skewing the observations made. This usually become less of an issue with more samples.

As in the past, I most likely will continue updating this for a couple of months out of curiosity.

Wtf Goh is quite popular as a tournament character..

Probably the most spread out in terms of popularity I've seen yet in all of 5's history.

Number of entry Jeffry players = lowest

Number of 2nd place Jeffry players = lowest

Number of 1st place Jeffry players = lowest

Final Verdict = Jeffry sucks in this game as I've been preaching for the longest of times now, this data backs it up.

The data is a very small sample size to support Jeffry is completely toned down. The only thing I think people can actually take from the data is which character is being used the most.