Warding
While generally good (and I appreciate you trying to reach out to the supports), it should be noted that one of the best teams in the EU LCS, SK Gaming, is notorious for their lack of wards. This criteria would also not be applicable on Twisted Treeline.
Staying with team when roaming
Some strategies involve a player being on their own, either counter jungling, split pushing, or defending an objective while the team takes something else. You are directly calling for these team players to be punished.
Picking the right fights
Please provide an algorithm that you can teach to a computer that will identify the "right fights". 3v5's when their 5 all have 2 hp left and your 3 are full health is still the right fight.
KDA
A support that gets you out of a situation alive when you otherwise would have died without getting a kill has nothing in their KDA to show for it (or, if they do, it's usually a death). There are many other times when a player almost gets a kill so many times throughout the game, but can't quite close it. The enemy has to go back, looses pressure else where on the map, but the KDA doesn't change.
KDA is probably the worst criteria to use because it focuses too much on a non-objective. The team with the most kills at the end of the game doesn't win despite their nexus falling. A kill is not worth more than a tower, and in some cases is worth less than the scuttle crab as far as objective control goes. By making KDA part of the criteria for player performance, you also unnecessarily drag out games, with the winning team sitting at the enemy fountain waiting for respawns to pad their obscene KDA.
#TL;DR:
- KDA is the wrong metric to use for just about every reason you can think of
- Other things are too hard to measure using a computer program, and would require an analysis going through every game individually.
- Because of the above points, winning/loosing the game is the only way to judge a player's performance.
