The answer is very simple:
it's not easy - in fact it is near impossible - to measure individual skill directly.
The reason being, that KDA and the other metrics don't correlate "well" to individual skill. For instance, above gold rank, CS/min and Rank have no correlation at all.
The reason it is not easy to measure individual skill with metrics, is because the value of a measured event depends on context. For instance KDA. Lets say your team lost a fight resoundly. You got 2 kills and survived. your KDA increases, but the outcome of the engagement is the enemy getting baron. The metrics say you are good. The outcome of the game says you are bad. So the metrics in this place are not telling the truth. Other metrics like vision score are equally as deceptive. Is warding the dot brush against a power farming jungler effective? Maybe warding the wraith camp is better? the score is the same but the results vary wildly.
and generally the measurement depends on already having some good assessment of the individual skill and having an outcome that is only relevant in relation to that assessment.
for instance in baseball, all metrics are relevant within one's league, and are translated into a simple score of "wins over replacement" (WOR). notice, that wins over replacement is a "team result" type statistic. So even in cases where individual metrics are heavily developed and used, the useful result is a team result stat.
In league of legends the large number of opponent compositions allow for measurement of WOR directly.