To answer the questions you are posing I think it is best to engage in a short thought experiment.
Let's say there is a support player who is matched with silver players. his vision control and GPM are at diamond level. If all his other qualities are the same as other silver players, this means he is consistently stronger than those other silver players because of his two superior qualities.
Assuming some matchmaking system exists, the players this player is getting on his team ("silver") and on the enemy team are going to have some "power level" that is more or less around the same level of whatever metrics the match maker uses.
So this player is matched with 4 other people who's metrics is similar to his, and against 5 people who's metrics is similar to his. If the match maker has made an error than obviously this player's superior qualities will make his team stronger than the other team and we'll see his team winning.
I mean, not winning in any one specific game, but over time with many teams and many games there are going to be instances where his better vision control and gold efficiency come into effect and will then result in a win for his team. so he's going to get X number of "extra wins", and will basically win more than 50% of his games.
In your scenario we see that the match maker has also made an error, because the player loses 70% of his games. Assuming you are talking a large number of games and not 5-6, This means that his team overall has to be very weak compared to opposing team. Where can this error of measurement come from?
it can be that the vision metrics is simply not calibrated properly: Wards that see nothing or that give irrelevant information are marked as useful, etc. Or that this metric is given too much weight in the overall measurement. Or that there is something really bad in this player's behavior that is measured as a positive value: for instance the extra gold generation comes from taking the ADC's CS. or there is a game-losing behavior that this player does that simply isn't measured like overstaying in lane because you want your BF on the first buy and losing dragon because you've caused a hole in your team's map control when the enemy jungler is looking for a free dragon.
When you measure performance based on the 1001 metrics that you can put into the game, there's always a chance that you are forgetting some obscure but really important factor that causes the player to win or lose.
When you measure based on wins and loses alone, after many games those obscure factors get integrated into your measurement.
