Edited in after question about general approaches.
As
P1naatt1ke1tt0 mentions, that is the end result and theoretical "approach". 6-4 means that equally skilled players at the highest level would out trade each other 60-40 after playing 100 matches. Now, as we all know this may not be as simple as people are after all discovering new tech, or got other random elements pushing the equation and so on. Thus it quickly becomes a social science, where one must mix facts with some degree of subjectivity in order to get anything.
The steps usually follows:
1. A person who mains or play a specific character a lot supply his perceived match-up rating.
If it is based of one feedback only it is perceived as weak, and any other feedback in other match-up discussions would trump it, or slide it towards the middle of none-agreements (depending on how strong that other MU has been discussed).
2. From this a discussion usually follows, with new recommendations for the MU-scoring.
Specific match-ups people disagree with the scoring for are discussed as for investigating tools or options that may have been neglected and so on. Here one usually challenge the specific tools one have towards each other, provide frame data, crushes, certain approaches of interest and in general contribute towards the "facts" for reaching understanding for those viewpoints. Unless there is an agreement the share numbers of sympathizers for a certain setups would determine what score is given, unless the MU-of the counter party has been discussed in great detail and found it agreeable towards the minority vote.
However, this is how it may be viewed in the short term. In the long term one has to consider the following:
3. Relevance lost over time or patches.
Generally, the newer MU or feedbacks trumps older ones for several reasons. First of the longer the current patch has been played, the more tricks has been exposed and been accustomed towards. Generally once one move over to a new patch the game changes, thus the newer is perceived as better and would most likely be taken into consideration until someone voice disagreement. The old data will then only be perceived as relevant until we got some new inputs, where older data is viewed as worse than "1 feedback only"-data.
Also, it is not always easy to gain a discussion with people who was at the forums 1/2 year ago. Unless a discussion follows it would then fall in the category of "1 feedback only"-data, or "weak" (which is better than nothing, by all means).
4. It will probably never be a complete project.
There will always be a certain error because of the degree of subjectivity. Also no one is equally good, and the mind games will always impact the game as well. Thus this would not be a fact based science and the MUs and the following tier list should not be interpreted as universally correct, and only an estimation for what the community as a whole believe to be true, collectively.
5. General go at it on an individual level.
Different feedbacks may have different degree of credibility. This is normally shown in the discussion part, as if one of the parties only supplies numbers but don't involve themselves in the following discussion. Normally the way to think about it is "in all the different scenarios, what options do one have towards the other character?" and in sum, what character would statistically win assuming equal skill and understanding (at the top level)?
An easy way to view this is to think a mirror match. A mirror match would always be 5-5, but even so characters may have certain advantages depending on if they are at frame advantages and so on. If you then take all the scenarios you view as points of interest in a mirror match, and replace it with a different character, then who would you view the odds of being in favor for overall?
The points of interest then may comes down to (example case of thinking)
Neutral game:
- Spacing
- Effect of success
- Probability of success
- CQC (Close Quarter Combat)
- Effect of success
- Probability of success
- Approach techniques
- Effect of success
- Probability of success
Frame advantage:
- Odds of getting it (linked to the neutral game)
- Damage output.
- Mix-up options. (Probability of successful rotation)
Frame disadvantage:
- Defensive options
- Damage output
- Effect
Other:
- Special cases
This is just one way to approach the MU, and may depend on the nature of the character as not everyone fit an A4 standard.
As an example case I can "argue" with myself at Christie vs Hitomi.
Neutral game, 50-50.
- Christie has more of an upper hand up close, but Hitomi make up for it in some better spacing game. All in all they are not great advantages either and even each other out.
At frame advantage, 60-40.
- Christie's options are more at the cost of only slightly less damage. She has a higher chance of gaining deep stun and generally starts of with frame advantage as she is much quicker than Hitomi. However, Hitomi has the potential of never stopping at frame advantage and do decent poke damage with 6t and low kicks. In other words if she first manages to gain the frame advantage she could potentially never let go. This also increase the possibility of gaining a deep stun once she has the frame advantage as it forces to opponent to react.
At frame disadvantage 40-60.
- Hitomi has advanced holds and parry. Her advanced holds in general do more damage if successful. However what really tip this in Hitomi's favor is that her parry gives her that big effect frame advantage. Also Christie's usually go to moves are either high punch or mid punch, making the possibility for success decent or at least limiting on Christie's normal play.
Total: 50-50.
In arguments one may go even further into detail, like stating
Neutral game up close (start of battle)
Christie
High punch 9 frames vs 10 frame high Hitomi
High kick 13 frames vs irrelevant
Mid Punch 11 Frame midpunch vs Hitomi's 13 frame
Mid Kick 12 frame kick (crush high?) vs Hitomi's 13 frame
Both have 14 frame low punch (even trade?)
Special note: Hitomi can parry both high and midpunch.
Judged from this
Christie's possibility to succeed with High p is >75% if they trade blows.
Christie's possibility to succeed with High k is 0% if they trade blows.
Christie's possibility to succeed with Mid punch is >75% if they trade blows.
Christie's possibility to succeed with Mid kick is 100% if they trade blows.
Christie's possibility to succeed with Low punch is >25% if they trade blows.
So, this is the odds "if they trade blows". Judged from this Christie clearly has some good opening options against Hitomi. However, if both characters know the numbers, they'll also see that Christie's options quickly boil down to mid or high punch, or unsafe mid kick. Thus the opening option has a high possibility to be throw punished on block after opener, or can be punch parried.
Thus Hitomi can neutralize 2 options in 1 with punchparry, or block mid k for throw punish or even block for later punchparry upon reaction. It then becomes and option of mind reading.
Of course, the Christie player could also know this, then the option could very well be to throw.
All in all though, Christie players would most likely either attack or throw. Holding is less attractive as they'll have to guess between 4 options, as with striking their odds (if they guess Hitomi will strike) may be 75~100%. Hitomi's case the 3 "real" opening options for Christie may boil down to 2 with punchparry, where the mid k hold may also be an advanced hold, in other words if one guess right (50% chance), the dmg done will be good. However, the option of throwing for Hitomi is low, as Christie would most likely not hold or block.
End note:
When that is said "people" won't or can't think this straight forward on the different MU's as certain MU's the spacing may becomes slightly more irrelevant, the overall trades is very complex and so on. In conclusion it is like P1naatt mentions, that you just have to see overall what the different options between the characters are, and make a sum of all the different scenarios to get an estimate as for in who's favor the balance is tipping.
