I think playing in times where it is morning/midday in europe is the best for pvp win ratio. Americans get tired, youngsters from urop flooding the tiers, adolescent and elder players from russia and europe still at work or in university bed… 
And in a squad, facing strong enemies in the queue for a whole night, can give you a lot of SR, but on the cost of W/L and vice versa.
And then there is personal mood, teams, tier… and currently also a lot of luck with the MM.
Or in which patch epoch one person was how active with which ship.
Point being, even W/L is not something which can objectify skill, nothing is. Skill is after all situational. It’s experience, which should be measured stronger. That is a more empiric number. Total battles, fleet strength, in relation to recent and total activity, etc.
Let epeen numbers be epeen. If I had to approach an MM, categorization of single pilots should be broken down to simple factors (so a couple of categories, like bronze, silver, gold, or an internal 1-10), which are calculated by the stats, instead of creating a MM which actually works on arbitrary number scales. And it should have limits for each factor, so farming a good stat over a point where it does not really say anything significant anymore, is simply epeen growing, but does not skew the classification internally. Otherwise, such players will always shift the boundaries of these numbers statistically, by being exceptionally bad or good on purpose, and the results will become chaotic. That will tell me more about holes in my mechanics and dynamics in the game, than about the player itself.
Also, I would add a self assessment of the pilot, which can influence this, by letting him decide where he is at any time, which could possibly influence his internal classification but only relatively. Finally, this would not be visible to players, so it can be finetuned independently.
Finally I would brainstorm about a (one time) rematch vote, which could give me insights where my MM struck gold, and also look at data from matches, where people left the game or disbanded the group or changed tiers afterwards. If any match gets a rematch with all participiants, I have a solid case to study. Making it one time only, of course, to hinder uncontrollable abuse, so no rematch of a rematch.
But to fix the game itself, I think the MM is not the primary issue, lack of players who stay with the game however is.
) and steadily growing.
