The reason why ranked failed and is a failure is because there’s no performance based matchmaking system, people group to exploit the algorithm, and there’s no MMR decay if you do placements every season.
When performance based matchmaking was attempted back in like 2017-18 they didn’t take into account deaths, so people yoloed out to stat pad sheet dps to get more rank points. Anyone that has played the game long enough knows that sheet number dps really doesn’t matter in a team fight if you can CC and sustain their dps to kill them.
All they had to do to fix it was add KDA or player deaths to the algorithm.
Given that the players are playing regularly (at least a little every season) so should not be becoming less skilled over time. Should there be?
Actually the idea was to try and estimate who did more or less work on a team and skew points awarded. Someone who lost and was pretty much AFK all game would get a significantly larger point deduction than the other people who were trying to play.
There were 2 issues with the system:
People started to try and game the system. Such as by focusing on achieving high amounts of stats in ways that might not contribute to overall winning the game. This is more a phycological issue since the player would still be better off winning (points earned instead of lost), but because they felt that they might lose they instead made them more likely to lose to try and cheat the system.
People were getting unfairly punished in cases where it did not detect performance correctly. For example, a player might have spent a significant amount of time mercing, or wasting the enemy team’s time but according to the performance based match maker it might think they were AFK and underperforming by a lot, even if their plays did result in significant net gains for their team.
The problems with KDA:
As discussed above, people will start to game the system. This includes them avoiding situations where they might die, at the cost of their team losing team fights, and wasting time trying to overly secure kills that another team member was already performing.
The stat is already directly correlated to game outcome. In the 50% of games you win you will have better KDA than the 50% of games you lose, where you will have worse KDA.
As discussed above, it will also result in players being unfairly punished in some circumstances. For example, your lane pushers will likely run up less kills/assists than your assassins or main team. This does not mean they are playing badly, rather that they are contributing in ways that are not correctly detected by metrics like KDA. Another example would be players who are playing melee diver heroes, as they are more likely to die than people sitting safely in the back. This would punish people for taking divers as if the match goes wrong, chances are they will end up with the most deaths due to what they are playing.
A performance‑based rating system is something this game has needed for a long time. It doesn’t matter whether the performance is calculated through a table or through detailed AI analysis of the match — Most importantly, they should finally DO something like this!
They can keep group ratings exactly as they are, but SOLO players deserve a proper SOLO rating!